More on Servos?

[jacampb2]

I didn't want to hijack the other thread, but all this talk about Servos has got me considering converting my stepper driven b'port S2 CNC boss to servos. I've started doing a lot of plastic machining/engraving on the mill, and the factory steppers are starting to show their limits. I can easily rapid at 100IPM or better, but I had to tune the acceleration so low that I can't even approach these speeds while machining. For a small engraving w/ lots of small moves I normally top out at ~30IPM. I can come closer to 50-60IPM during straight moves like cutting out a part from a sheet, but I would like to get into higher feeds for the plastic stuff.

I'm trying to educate myself on servos, but there is a huge hodgepodge of information out there, do any of you have any pointers on where to start learning? I found this website DIY DC Servos (http://www.truetex.com/servomod.htm) and it seems interesting, but will this perform similarly to a off the shelf DC servo?

I was considering just purchasing new steppers, as the consensus in one of my past threads was that there will be an improvement on speed and power over the 30+ year old factory steppers on my mill. I hate to throw more money after steppers though and still not be happy.

I know that AC servos seem to be the way to go, but in order to afford it I would have to buy surplus, and in order to do that I need to know how to size and match the drives. DC servos seem to be a lot cheaper to pull off, but is there going to be any kind of performance increase by using them?

I would like to be able to get my feed rates up into the 100IPM range, is this unrealistic for a knee mill w/ a ~42x18 inch table size? And one last question, to keep costs down, can you intermingle steppers and servos? My Z axis is fine with stepper drive, as is my rotary table, the only axis' im interested in speeding up are the X and Y.

Any info is very much appreciated.

Thanks,
Jason

[Hood]

Performance of DC servos will be fairly similar to AC servos so no problems there getting the accel you want. DIY servos versus real ones I would think the likely difference would show at slow speed where the DIY oenes may be prone to cogging. However I really dont know much about the subject so take that with a pinch of salt :)
One thing I would say is get servos that are big enough to drive 1:1 as you will get better acceleration performance, but thats just my opinion, I am sure there will be many who disagree.
As for mixing steppers and servos, if you are meaning with Mach then there is no problem as Mach doesnt know or care what the motor type is.
Hood

[macona]

My mill at home has brushed DC servo motors and I can get rapids up to 400ipm. I think they are 2:1 but they are so massive that there is no acceleration issue here. My little lathe is 1:1.

As for drives I would go with the Whale3 from http://cncdrives.com

You should be able to pick up some old motors on ebay.

There are also a few sellers selling AC brushless servo packages with three motors and drives for a really good price. Usually Yaskawa or Mitsubishi motors. Given the choice I would go with those over DC brushed. They are an integrated set and can tune themselves. I never have had to tune the mitsubishis on my lathe. They tune themselves on the fly.

[kf2qd]

for Brushless servos you might take a look at Yaskawa (or if you have a good Omron distributor you might look at Omron, they rebrand Yaskawa), Mitsubishi or Automation Direct.

FOr brushless you will need to buy the Drive (or amp), cables and motor. Most brushless drives can be setup for Step/Direction, Clockwise/Counter-Clockwise or analog control. You can also adjust the number of steps per revolution.

Seeing as how most brushless motors have a max speed of 3,000 RPM you will probably want some form of gear reduction. Timing belts (inexpensive)are a common method or a planetary gearbox (expensive) are the most common ways of getting a usable shaft speed.

Brushless motors are MUCH smoother than steppers and you have the ability to set the number of stepps per revolution of the motor so you can get very good resolution, even if you have to change the step/rev because of gearing. Typical is either 1000 or 2000 from the factory, but it is easy to change those number to something that will work better in you application.

If you are currently using Steppers you will need to use some gear reduction if you use brushless because there is quite a max rpm difference between them.

I have worked with both the Yaskawa/Omron drives and the Mitsubishi drives. They each had their little quirks and niether was easier/harder to use than the other. You can run drives up to 750W on single phase 220V, bigger drives will need 3 phase. You would need to order an interface cable as they tend to use a connector that is a pain to solder on. They need power and a drive enable signal and they are good to go.

[Hood]

You can run drives up to 750W on single phase 220V, bigger drives will need 3 phase. You would need to order an interface cable as they tend to use a connector that is a pain to solder on. They need power and a drive enable signal and they are good to go.

I have 3KW drives that run on single phase also I dont think its that hard for someone with basic soldering skills to wire up cables themselves, my drives have 44pin high density D connectors and its not a problem and a heck of a lot cheaper than paying for pre wired cables..

Hood

[macona]

I would say that 80-90% of the brushless drives on ebay are analog input velocity/torque models. On mitsubishi model look for the model ending with an "A" or "AN" and Yaskawa the model ends with a "P". These drives are positioning drives and they will accept step and direction inputs.

The Japanese drives do not really allow using other brand of motors with their drives, especially with newer sets as the encoders are connected to the drive by a serial communication method. This relays motor information to the drive and also can provide absolute position indication.

On the other hand many of the American and European drives will support almost any motor. The drives are programmable in software to the specific characteristics of the motor. I am currently using an Allen Bradley 1398 series drive to control a Parker Compumotor motor. All that was needed was a little adjustment of the encoder to get the timing where it drive was looking for it and it runs fine. I have drives by Aerotech, Glentek, Electrocraft, and SimpleServo (AC Lenze).

Motors are a different story. I have a whole bunch of different ones sitting around for the event I have a use for them. Some have only hall and tachometer feedback. These will work fin with any analog input drives as they dont need an encoder. The control that runs the drive will look at feedback from an external source. This could be an external encoder, resolver, abscoder, linear scale, etc.

Motors with resolvers usually require a specific drive for resolver based motors. A resolver is a rotary transformer that looks for a shift in the output to determine position. Resolvers are absolute positioning devices. They are very rugged and rather expensive. Since they are analog the minimum useful resolution is very good, down to several arc-seconds.

Small motors 400w and under can have top speeds from 3000 to 4500 RPM with intermittant speeds up to 6000 RPM. The larger motors tend be found mostly in the 2000 RPM varieties. Mitsubishi offers larger motors in three speeds, 1000, 2000, and 3000 RPM.

So far almost all the drives I have messed with that say three phase in run fine off of single phase. They have no phase loss circuity so they dont know the difference. You may have to watch you acceleration times or set the maximum torque setting in the drive to about 70% though. Under heavy acceleration the bus voltage will drop and can trigger an alarm on the drive. You cal also add more bus capacitance to the drive as well. The drives usually have separate bus power supplies from the control section. The control section also usually controls a contractor that applied power to the drive electronics after it passes self test and the drive is enabled. Dont leave this out when you build your final design.

Bayside makes some nice gear heads. Watch out though. Servos are not standard NEMA frames. They are almost always a metric frame that has a larger shaft, centering boss, and the mounting bolts are on a larger bolt circle. I have modified gearboxes that were intended to fit NEMA 23 sized steppers to attach to the equivalent size servos.

If you can find them use harmonic gear boxes, they have zero backlash.

[Sparky_NY]

I recently finished retrofitting a bridgeport V2E3 which was a servo machine with boss 8 control originally. I used the cncdrives dugong servo drives, I have been extremely happy with them. Their whale servo drive is on the small side for a machine the size of a bridgeport.

The factory motors are powertron servos, 6000rpm max although with the original 2:1 belt drives and original rapid speed of 250ipm the motors only reach 2500rpm in use.

As for mixing servos and steppers on the same machine, that is no problem if you are using Mach 3 for control. I have dugong servos for XYZ, a spare drive for future use to power the knee and a stepper running my rotary table 4th axis.

Having had a stepper bridgeport before, I can tell you the difference in noise and smoothness is unbelievable.

On your particular present application. If you do a lot of small moves, you probably want to tune for best accelleration at the sacrifice of rapid speed. That would probably make quite a difference in real world machining in your case.

[jacampb2]

Lots of good info so far. I am trying to keep up.

The machine already has 2:1 reduction from the steppers to the screws. I am not opposed to going 1:1, but that seems like it will take a monster motor. We are talking about a fairly heavy table here.

Where do you start on sizing motors? I see that a lot of surplus stuff is rated in Kw or HP, but rarely torque like you typically find with steppers. How do I figure out a baseline of what I will need?

How much is this likely to cost me? I have lots of amateur electronics experience, I am not afraid of soldering, wiring or reading schematics. I am a pretty good scrounger, and if I can educate myself enough about the subject, I don't think it will be necessary for me to buy anything in "kit" form. Can anyone give me an guess of the cash outlay if I find the majority of what I need on the used/surplus market?

Does anyone have any more thoughts on the DIY DC brushed servos linked above? This kind of thing is right up my alley of how I like to tackle problems...

Thanks,
Jason

[jacampb2]

Okay, so I did some more research and I am still confused. Here are some more questions:

Is peak/stall torque for a servo similar to holding torque for a stepper? Meaning, if my steppers are 1000 Oz/in holding torque, am I looking for a servo w/ similar "stall torque"? I know that steppers have a "torque curve" and drop torque quickly as speed increases. I also understand that Servos have a more constant torque curve, but it seems quite low. I am looking at these servos from Keling drives (I have had good luck with their products for the first retrofit of the mill) Clicky, first servo in the list (http://www.kelinginc.net/ServoMotors.html), they are rated at 1125 Oz/in peak, but their constant torque is 226 Oz/in. I believe this makes them ~700W motors, is this going to be enough w/ my 2:1 reduction? Will it come even close to cutting it if I go 1:1?

I did some of the formulas on the geko faq, and it seems that I should really be looking for a 1.5HP servo, however, anything in that range, even used on ebay is going for $300+, is this really what I will need?

Thanks,
Jason

[Sparky_NY]

There is a poster at both the mach yahoo group and cnczone, I believe it is Himykabbile (hope I spelled that right). He used kelling 850oz peak servos for his bridgeport clone with gecko drivers. I used a belt drive reduction. He got 400ipm rapids but runs it lower than that because he says its scary.
(go to cnczone, the bridgeport section, there is a thread he started on a new Z drive, that is the guy)

Edit: LOL, I went to the zone and seen you already posted your questions and got replies from the guy I mentioned above.

[jacampb2]

Yes, but thank you. I typically try to wring all the information I can out of the available sources :D

Later,
Jason

[jacampb2]

Well, I have three 3Kw AC servos coming thanks to a guy on cnczone who pointed them out to me. Once they come, I am going to have my work cut out for me finding a compatible drive. Once again, if anyone can point me in the right direction for research, it is appreciated!

Thanks,
Jason

[jacampb2]

I found some big analog servo amps that I purchased on a whim... A-M-C S100A40, I haven't received the servos yet, but these drives should be able to handle any requirement they might have. They are good for up to 400Vdc buss and 100Adc peak current. I only found 2, but I don't think I'll need anything this big for the Z axis, so I'm not really concerned about finding a 3rd drive at the moment, I can always leave the Z stepper driven for now.

I know the drives are not positioning drives, the are +/-10V analog velocity or torque control. I intend to build the YAPSC:10 servo amp controller. It is an open source project detailed on CNCzone. It appears to be a lot like the old Pixie controller, but the software interface is excellent and it appears the designer has most of the bugs worked out. I realize that this might not be the best route to take for a newbie to servos and control, but the price is right! I only gave $50 each for the amps, so I am not out much if it works.

Let me know your thoughts on my plan. If it is full of holes, I am not out much on the drives, I can still part them out, as the IGBT modules in them are worth far more than I paid for them :)

Later,
Jason

[jacampb2]

Awe hell, it turns out these amps are sinusoidal command only... I either misread the data sheet, or read the one for the 100A40 (the S is the key). Is there any control solution that will take step/direction? Any affordable solution at all?

Thanks,
Jason

[macona]

The delta-tau control is the only control I know of that will drive them.

3kw??? What are you thinking? :confused: That would be good for a spindle. Massively overkill for axis control short of a big HMC or boring mill. 3 KW is over 5HP! And that continuos, peak they will run 3 times that. That will twist a ball screw like taffy. No forgiveness.

The milltronics VMC we had at work only had 900w motors on it and it was a 40 taper machine.

[jacampb2]

The delta-tau control is the only control I know of that will drive them.

3kw??? What are you thinking? :confused: That would be good for a spindle. Massively overkill for axis control short of a big HMC or boring mill. 3 KW is over 5HP! And that continuos, peak they will run 3 times that. That will twist a ball screw like taffy. No forgiveness.

The milltronics VMC we had at work only had 900w motors on it and it was a 40 taper machine.

Well, the price was right, there is no reason I can't current limit them to keep the torque in a manageable range, is there?

Can you tell me more about the delta-tau control? Is it a stand alone controller? It seems that Galil (?) makes a controller that will run them, and it is Ethernet or PCI based, however, it looks like it doesn't run with any of the hobby HMI softwares....

Like I say, I probably screwed up on this one, but I will happily use the parts out of the drives for some other project, so it isn't all bad.

Thanks,
Jason

[macona]

Yeah, you can turn them down but now you have a huge intertial mismatch which will make tuning difficult. That and a 3kw motor usually weighs in at 30 to 40 lbs. Thats a lot hanging off a moving axis like the X.

Galil is the company you are thinking of. And there is a plugin for Mach to run one of the cards. Never used one. The cards tend to be pretty spendy. More here. http://www.machsupport.com/forum/index.php/board,81.0.html

There is a guy on ebay that sells used mitsubishi and yaskawa servos in packages of 3 like this:

400W

http://cgi.ebay.com/YASKAWA-SIGMA-II-400W-SERVO-PACK-MOTOR-3AXES-CNC-3_W0QQitemZ250578900029QQcmdZViewItemQQptZLH_Defau ltDomain_0?hash=item3a57aa943d#ht_6906wt_1339

750W

http://cgi.ebay.com/YASKAWA-SIGMA-II-750W-SERVO-PACK-MOTOR-3AXES-CNC-2_W0QQitemZ250566840366QQcmdZViewItemQQptZLH_Defau ltDomain_0?hash=item3a56f2902e#ht_6802wt_1339

I would go with something like this before messing around the way you are talking about. The drives in the auction above take step/dir input and are as plug and play as you are going to get. I have had a hard time finding drives the size you are looking for. I think i ended up finding an aerotech drive for a 3.5kw motor I have. Never have gotten around to trying it as I need to change the encoder on the servo motor.

[jacampb2]

The servos are 26Lbs each, and I still don't have a lot of info on them. They showed up here yesterday, and have no more name plate data than what I had from the auction. They are really much smaller than I expected. They may be 2-3 inches longer than the factory steppers on the mill, but no where near the diameter. Torque rating is 9.54Nm, or 1350 Oz/in, I was mostly guessing, but I figured if the holding torque on my steppers was 800 Oz/in and then 2:1 reduction to the ball screw, that this servo in direct drive configuration would be a reasonable match. Like I mentioned before, I don't fully understand how the differences between rating steppers and servos match up, but I know I can fairly easily overload the steppers, and so I wanted to be certain I overdid it-- evidently I did :)

Maybe the 3Kw rating is peak? I sent Nachi an email requesting info, but is there a way to determine the voltage and current w/o a data sheet? It is fairly apparent which leads are motor winding leads, they are like 12Awg. I was thinking that maybe it could be done by measuring the field coils resistances?

The motors also appear to have brakes, they cannot be turned by hand, so either there is a motor brake or they are fubar! Hopefully I will get some pictures and a bit more information today. I want to remove the encoders and see what they are, I think they may actually be resolvers and will have to be swapped.

I'll keep you all posted. Macona, thanks for pointing out the mach support page on Galil. It looks like it actually supports a lot of their interface cards. Maybe those drives aren't a lost cause after all!

Later,
Jason

[Hood]

Dont remove the encoders (if they are) unless you have a means of aligning them again, some drives have a means in their software for offsetting but I still prefer to have it physically correct in the first instance.
As for 3KW servos, cant remember what machine you have but I wouldnt be too stressed about putting them on a mid sized mill, I have 3KW ones on my Beaver mill, the ballscrews are 32mm dia.

Hood

[jacampb2]

The machine in question is a Bridgeport Series 2 CNC. Ball Screws are 1.25-5.

Bellow are some pictures of the servo and the encoder/halls. I can't find anything on the encoder yet, the only numbers on it are MFE0011BOMA * 04080068N. It is definitely an optical encoder in there, but I have no idea if it is incremental or absolute. I have the wiring colors if they happen to be standardized and anyone can help me...

I have figured out the actual servo wiring, the RED, WHITE and BLACK leads are the three phases, the two small yellow are 24Vdc brake, and the small blue wires are thermal switch. The greed is earth. I suspect the encoder is going to be more of a bitch.

Question on determining motor voltage for rated RPMs. I have heard this method will work with a DC brushed motor, but will it work for a AC servo? The method is to spin the servo at 3K rpms and measure the voltage produced. Since they rotor has permanent magnets, I suspect it will work for a AC motor too, but thought I would ask before I tried it.

Here are the pictures, if anyone can come up with more data, or point me in the right direction I would be forever indebted!

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_003.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_004.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_005.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_006.jpg

Thanks,
Jason

[Hood]

With a series 2 I would have no hesitation using 3KW servos, not needed probably, 2KW would be plenty but you are not going to twist these ballscrews IMO.

Afraid I cant help regards the encoders, from doing a google on the motor number it seems to be a panasonic/matsu****o motor but no info to be had so far.

Hood

[jacampb2]

Thanks hood, I came up with panisonic being a possible OEM too, but dead ended there...

I managed to get out to the shop today and do a bit more testing. I spun one of the servos on the lathe and measured AC output phase to phase. The lathe has a 7.5HP spindle motor and tops out at 1500 RPMs, and it couldn't get the servo up to that speed-- blows my mind because it is a pretty good sized lathe...

I ended up testing them at 1050 RPMs, which is 35% of their rated speed. I got just over 35Vac RMS between phases. Multiplied by 2.857, that comes out to 100Vac at 3000 RPMs. Am I on the right track? That means for the servos to hit there full rated 3Kw output I need 100V, 30A, or is there and efficiency factor I need to figure in?

Thanks,
Jason

[macona]

Nachi does make motors. Funny thing about Japanese companies is they dont like to give out info, at least not on the english sites.

Those are definitely 3kw motors. Heavy power leads and size indicate that. My mill is a Series II copy and it came with 540 oz/in motors on it. And it will break ball screw mounts in a heartbeat.

The mitsubishi 3,5kw I have wants 145v on its windings. Some motors will be all the way up to 200 to 300v depending on manufacturer. If you get allen bradley drives you can set the voltage. Low voltage will limit the top speed of the motor. You will probably need a drive somewhere in the 240v, 40amp per phase range.

[Hood]

Jerry, whats the gearing on your mill? Just thinking the mounts must be pretty weak if 540 ozin can break them.

Jason
Most of the AC servos I messed with come in three voltage flavours, 110v, 230v and 460v, kind of sounds like yours may be 110v.
Problem you are going to have with any drive is you will need to tell it a lot of info about the motor so unless you can get that I think you may have problems. Maybe an email to Nachi would get you some answers?

Hood

[jacampb2]

I have emailed them, and got a response this evening, they have forwarded my email to the robotics division, and so far their attitude seems helpful, so I will keep my fingers crossed.

Thanks,
Jason

[Hood]

Japanese companies are usually very helpful, hopefully Nachi will be the same :)

Hood

[macona]

I cant remember if it is 1:1 or 2:1. I think 2:1. The ball screw bearing mounts are cast aluminum. These are pretty decent sized glentek PM servos. About 1.5kw. It has way more power than needed.

Like I mentioned before. The Milltronics VMC we had used direct driven 900w yaskawas on all axis. This includes the 40 taper milling head with a 10HP motor on it. Way more massive than any part on a series II. And it moved it at 800ipm.

[jacampb2]

I found a Galil card that does sinusoidal command, it is supported by the Mach/Galil plug in. I should know later this morning if I am going to be able to purchase it. It looks promising though, maybe I'll be able to use these big drives after all.

I'm sure I am opening a HUGE can of worms, especially considering I have no idea what I am doing, but hopefully I/We :D will figure it out.

Later,
Jason

[Hood]

macona
If you can break the ballscrew mounts with under 8Nm they are in my eyes not very substanial ;)

My Beaver mill came with 7 or 8Nm DC Servos as standard, its 1:1. I think Series 2 Bridgeports will be around 3.5 to 4Nm but they are geared probably 2:1 so that would be around the same as the Beaver.
As for your Miltronics, would that happen to be linear rails? and gearing?

Hood

[Hood]

Jason
Do you have the software for the drives or do you just have buttons on them for adjusting the parameters? or???

Hood

[jacampb2]

Jason
Do you have the software for the drives or do you just have buttons on them for adjusting the parameters? or???

Hood

As far as I can tell, these amps are as simply analog as possible. They have no tuning or feedback loops to the drive themselves. All of the control has to be handled by a controller. These are probably as close to a voltage/current amplifier as you can get. You command them with 2 sine waves 120° apart, from what I understand, the amp calculates the 3rd phase and amplifies them and outputs amplified waveforms to the servos. The controller has to close the loop.

I missed out on the Galil card this AM, but I am currently trying to get a guy to let go of a 7 axis card for less than a princes ransom. Hopefully I will be able to get something. I am sure it is just going to be a matter of time, there is a ton of this stuff on ebay. Just have to sift through and find what you need.

Thanks,
Jason

[Hood]

I would imagine you need to set such things as v/1000revs and max voltage and current etc to suit your motor. I think some of the old AC drives may have done that via dip switches, maybe yours are like that?
I had/have some old analogue Osaia drives here and they have what they call a personality module, that is programmed with the motors specs but you still have a serial connection so that you can do the velocity or torque tuning. The bad thing about the personality module is it cant be programmed by the drive and I think it was about £60 to get one done.

Hood

[macona]

I had a problem with the drive and the motor ran away on the Y axis. Lots of momentum plus the motor broke the casting. Migged it back together and so far so good. I think it may have been cracked to begin with.

The milltronics does have linear ways. No gearing, just a direct coupling to the servo. And on the Z axis (Entire head moves up and down) there is no counter balance. So the motor supports the entire weight of the head once the brake is released. And it moves pretty darned fast.

[Hood]

I had a problem with the drive and the motor ran away on the Y axis. Lots of momentum plus the motor broke the casting. Migged it back together and so far so good. I think it may have been cracked to begin with.

Yes think it must have been, either that or a poor casting.



The milltronics does have linear ways. No gearing, just a direct coupling to the servo. And on the Z axis (Entire head moves up and down) there is no counter balance. So the motor supports the entire weight of the head once the brake is released. And it moves pretty darned fast.

Well I am surprised to hear that if the head is direct drive, I could see the X and Y being OK direct with a linear rail machine but I know a 3/4 KW AC Servo was no use for the quill on my Beaver NC5, I tried it and although it worked I could not get the acceleration anywhere near what I was looking for.

Hood

[jacampb2]

Well, I got the Galil card! It is a DMC-1770 Brand new in the package. I gave $400 for it, new retail is $3k, and it is still supported, and works w/ the mach plug in. The Card is 7 Axis, it turns out I am fortunate I did not get the first one I tried for, it was a 4 Axis card, and as it turns out, to run the sinusoidal command drives, the controller needs to allocate two DAC's per axis, so the 4 axis card would have only been 2 axis, and I would have been sorely disappointed...

Anyhow, I only have the two sinusoidal drives, so 4 of seven axis will be allocated, I think I will shop for a +/-10V command drive for the Z, that takes one axis, VFD control for spindle will be an additional axis, and I can outfit my rotab as either stepper or buy another servo for it and utilize the last axis. The biggest downfall of the card is that it is ISA bus, I dug through my heap of computers for one with an ISA slot and it was no go. I ended up buying an industrial MOBO on ebay, 1Ghz P4(IIRC, it might be P3, I don't do intel) and a ISA chassis for it w/ 10 ISA slots...

I think the big thing is going to be keeping an ISA slotted machine around for backup. I'll probably buy an old desktop board, AMD 1.6Ghz or so was the last generation AMD boards that had ISA. I just buy a few now (they're dirt cheap on ebay) and keep them around for backup.

Thanks,
Jason

[jacampb2]

Back to the servos... I still have to decode these encoders. I have found the VCC and ground, I hooked it up to a 5Vdc supply and things get weird. I followed 6 leads (+2 for VCC, grd) back to a differential line driver, I figured that I had lucked out and found the differential encoder outputs. Measured with a volt meter, some of these outputs swing with shaft rotation, at least one stays at 0Vdc, and 1 stays at 2.5Vdc. I also have three more leads from the encoder that do not go to the line driver. I assumed these were the hall outputs, but they don't seem to change at all.

So, I took one of the servos down to my basement electronics lab and started probing the encoder w/ the scope. The lead w/ 2.5 VDC on it all the time is actually a square wave output. I didn't have time to do much with it yet, I got called to babysit my kids for a few hours... The leads that do swing, do so very rapidly w/ servo motion, even turning the thing as slow as I can by hand, I get probably 50 pulses on the scope. My scopes probes suck so bad that I am getting a bunch of noise and other artifacts that don't help. I am beginning to think this encoder is some form of serialized data output. If that's the case, I am going to have to replace the encoders, right? If so, do I have to have halls too, or can I get away with the cheap CUI capacitive encoders from digikey?

Thanks for all the help so far!
Jason

[macona]

It wouldnt surprise me if the encoders are serial. A lot of new ones are. That pulsing line is probably the clock line.

Replacing encoders is possible. You will need to find an encoder with hall commutation. There are several companies that sell them Renco being one of them. Then you need to find out the pole count of the motor. You can figure out this with a scope. Hook leads to two of the three motor wires and turn the motor one turn. If you see 4 positive peaks it is a 4 pole-pair or 8 pole motor. Different manufacturers use different terminology when specing their encoders.

[jacampb2]

Turns out there are 2 lines with a high freq triangle wave. They lead to pin 13 and 14 (AM26C31) on the line driver, which are the differential outputs from one channel. The output is a 450Khz 2.5V triangle wave as close as I can tell. The other are square wave, and frequency is dependent on how fast I turn the rotor. I still can't get anything out of the three leads that don't go to the line driver, perhaps they need pulled up, maybe open collector outputs?

I'll try to determine pole count in a little bit. I sure wish Nachi would reply with a data sheet on this thing...

Thanks,
Jason

[jacampb2]

4 positive peaks per rotation. And, let me tell you, that is far harder to count than it sounds. I ended up having to video the scope, and play it back frame by frame to get an accurate number...

So, 4 poles, now, if I am brave I will pull off one of the encoders and measure shaft size and see about ordering something.

As for Halls being necessary, the manual for the Galil card say that they are not necessary, even for AC servos, that the controller can do a phase search at start up. Do any of you know anything about this? It seems a lot simpler if I can do w/o the halls and just use an incremental encoder. Do you still need to "align" the encoders if there are no halls?

Thanks,
Jason

[Hood]

My drives can self sense, the motors dont move far, probably only 1/2 rotation, whilst the drive senses but means you have to make sure you arent close to any obstructions when you power up the drives.
I priced a Renco encoder for a motor that had a resolver and it was around $300 - $400 if I remember and then it was going to be about the same to send it to the UK LOL.

Hood

[macona]

4 positive peaks per rotation. And, let me tell you, that is far harder to count than it sounds. I ended up having to video the scope, and play it back frame by frame to get an accurate number...

So, 4 poles, now, if I am brave I will pull off one of the encoders and measure shaft size and see about ordering something.

As for Halls being necessary, the manual for the Galil card say that they are not necessary, even for AC servos, that the controller can do a phase search at start up. Do any of you know anything about this? It seems a lot simpler if I can do w/o the halls and just use an incremental encoder. Do you still need to "align" the encoders if there are no halls?

Thanks,
Jason


Yeah, i cheated on that one. I used a storage scope.

The galil does a search? Interesting. I guess I can see that. It just applies two sines and watches which way it moves.

I found some renco encoders on ebay that would/will work for mine for about $80.

[jacampb2]

The galil does a search? Interesting. I guess I can see that. It just applies two sines and watches which way it moves.

What it says is that the motor has to be allowed to go through a magnetic cycle in each direction, and it looks for the 0° phase. Better yet, here is what the manual has to say about it. It sounds like this is the "dangerous" way of doing it, I have heard it termed wake and shake on CNC zone...


If Hall Sensors are Not Available: To initialize the commutation without Hall effect sensor use the command, BZ. This
function drives the motor to a position where the commutation phase is zero, and sets the phase to zero. The BZ command argument is a real number which represents the voltage to be applied to the amplifier during the initialization. When the voltage is specified by a positive number, the initialization process ends up in the motor off (MO) state. A negative
number causes the process to end in the Servo Here (SH) state. Warning: This command must move the motor to find the zero commutation phase.
This movement is instantaneous and will cause the system to jerk. Larger applied voltages will cause more severe motor jerk. The applied voltage will typically be sufficient for proper operation of the BZ command. For systems with significant friction, this voltage may need to be increased and for systems with very small motors, this value should be decreased. For example, BZ -2 will drive the X axis to zero, using a 2V signal. The controller will then leave the motor enabled. For systems that have external forces working against the motor, such as gravity, the BZ argument must provide a torque 10x the external force. If the torque is not sufficient, the commutation zero may not be accurate.

The only thing I am not 100% sure about, is if this is saved in the axis tuning, or if the command has to be issued every time you start the machine. This info is in the "getting started" section of the manual, and all the previous steps are about setting up the motor properties, so I assume this is a saved setup. I will do more research and reading over the weekend. I would certainly be a lot happier buying the $30 CUI encoders than new rencos or similar. In fairness, I looked up what it would take from Renco, and it was $80 per axis w/ commutation tracks, but that is still more than I would like to spend right now. I've so far dropped about $1k on something that was just a little pea brained thought a month ago-- LOL, I told you guys I was in a spending mood!

Thanks,
Jason

[Hood]

It will be every time you start the machine as the Galil will have no way of knowing if the motor has moved while its been switched off. I dont use self sensing on my drives but I did try it out and as said it does rotate the motor one way then the other. My drives have the capability of using a standby voltage and although I have never tested I presume, that if the standby is present, even if the main drive power is switched off the self sensing wont be needed on next start.

Hood

[jacampb2]

Right, I understand that it will have to do it every time the controller powers up, what I was saying I was unsure of, was if I would have to manually issue the command to do it, or if it is saved as part of the initialization parameters.

Thanks,
Jason

[macona]

There may be one other snag. I think the mach plugin will have to send this command out. I have never even looked at the plug in. It might be possible to run a script at power on to send the command. The next problem is what happens if you do and estop and kill power to the motors. Bringing it back up could result in very bad things happening when it goes to find home commutation so you can jog away. If your cutter is in the work, well, SNAP!

I have heard of this commutation as wake and shake. I think one of my aerotech drives has it. There was one drive I was using that rotated the motor one turn to find encoder Z and base the commutation off of that. Commutation would then be calculated from the incremental encoder (The encoder count was set in software) I think they called it electronic commutation.

[jacampb2]

The next problem is what happens if you do and estop and kill power to the motors. Bringing it back up could result in very bad things happening when it goes to find home commutation so you can jog away. If your cutter is in the work, well, SNAP!

My gut says that recover from e-stop wont be an issue. This is action that is 100% dependent on the controller card, as long as the card retains power, I would think that it will keep track of the encoder position. I think the scenario where this could happen would be a complete power failure and PC reboot while the mills running. In which case, it will probably snap the cutter when it suddenly looses power while still in the toolpath.

Like I said before, I am going to have to read more about this, but it seems to be part of the initial motor setup/axis configuration. My hope is that the controller stores all this information and does not need to be manually initialized. The manual also details the use of this command with the use of hall feedback to more precisely align commutation, that makes it difficult to believe it has to be run every time. Halls are configured with a similar command, you shouldn't have to initialize them every time either. I am sure this will be a long road, but I will keep you all posted as I learn.

The plus side to all of this is it sounds like the Galil/mach combo is one sweet deal. All accounts are it is integrated well and it is butter smooth compared to PP operation.

Later,
Jason

[macona]

If I recall right the Aerotech drive I have does the same thing. The drive can figure out how far out it is and you can adjust the encoder to get it into alignment or adjust the offset in software. I think the is similar for the AB drive I am using in my lathe as well.

Using it to align a hall vs running without halls altogether is a totally different story. I would just spend the $80 and get the encoder. I prefer machine tools that dont jump around when you turn them on myself!

[jacampb2]

Nachi got back with me today, I got info on everything but details on the encoder. Supposedly the data sheet for that is only available in japanese. I requested it anyhow, maybe the graphs will show something.

Long story short, the encoder that is there is serialized, it looks like the two high freq pulsing leads are data, they are labled Rx and Rx, the other four are labled as complimentary A and B outputs. The extra leads are also used in communication, charging the battery, and mem clr. So, long story short, it still may be quadrature output on the A&B channels and I think maybe it stores its current position and reports it on restart. It also may be bianary output on the A&B channels, in which case it is probably useless to me...

I'll be back with more info later. Thanks for all the help.

Jason

[macona]

FWIW, I can read some Japanese

Its basically a combination absolute/incremental encoder.

[jacampb2]

FWIW, I can read some Japanese

Its basically a combination absolute/incremental encoder.


Macona, Nachi got back to me with the Japanese data sheets. It looks like they are not for the same encoder (different model number on the data sheets) but I think they show the absolute channel of my encoders from looking through the graphs. I asked Nachi if it would be alright to request assistance in some translation and they said it was fine but requested that I not publicly post the documents.

Would you be willing to take a look at them and tell me what you can? If so, PM me with an email address and I will forward them to you.

Thanks,
Jason

[jacampb2]

Well, pieces are showing up in the mail every few days. I got my Galil Card, now I am just waiting on the PC components to show up so I can start building the new controller PC.

I got some more correspondence from Nachi Robotics. They sent me a Japanese data sheet for their newer encoders, but are still trying to track down the info for the ones I have. It sounds like what I have here is a incremental/absolute encoder. 2500ppr on the differential quadrature channels, and absolute position data sent serially on the 3rd channel. The data sheet says RS-485 serial communication. Right now, I am thinking of using the "wake and shake" functionality of the Galil card to find commutation. Obviously that is not the best way, but the commutation data from the encoder is included on the serial output. It sounds like I can pay Galil $800 to write firmware that uses the serial channel, but if I were going to do that, I could just buy new encoders w/ hall tracks.

I have a few new questions. I have been going over the motor data sheet and it specifies the voltage constant as 17.9 V*10^-3/min^-1. What does the min^-1 (-1 is superscript) mean? I assume it has to be some rotational unit, but what is it? The other question is this: The data sheet specifies the number of poles as 8, can I assume that this means 8 individual poles and 4 pole pairs. When I did the scope check for pole count earlier, I came up with 4 pole pairs. I feel confident that this is correct, but I thought I would ask if it is common to see individual pole count in the data sheet.

Thanks for all the help!
Jason

[macona]

Different companies will list their poles differently. What you got from the scope matches the data sheet.

In the software for the galil there may be a place to enter this info. It is often needed for tuning.

[Hood]

The voltage constant will be 17.9v per 1000rpm

Hood

[jacampb2]

I finally got the new PC done for the mill. I ended up buying an Itox industrial MOBO, ATX form factor, P4 533Mhz FSB, 2.53Ghz P4, 2 gigs of ram. The board has 3 ISA slots, 3 PCI, and AGP. I'm going to keep the 1Ghz PIII SBC and ISA backplane for backup.

The mobo has a CF slot on board, and originally I bought a 8gb CF card to put the OS on to make it peppy, but I could not get the damn thing to boot from the CF. So, it is ordinary old hard drive and the CF is going back, it was to much money for the little gain anyhow.

After many false starts w/ the CF card, I finally got XP pro on the box, up and running and the burned copy of the driver CD that the seller sent me was garbage, nothing but the owners manual would open. Luckily I am set up back to back with my office machine and I DL'ed all my drivers to flash and got it running. Installed the Galil terminal, installed the card and fought my way through setting up an ISA card (been a while...) and the terminal connects and communicates with the card just fine.

There are a lot of hidden costs in this Galil setup. I think I mentioned, I gave $400 for the card, that is just the start. Their breakout modules cost close to $400 each new, and even used their are really proud of them. They give pinouts, and I decided to tackle breaking out the connections myself, but even that is pricey. I ordered $150 in interconnects from digikey the other night. The high density 100 pin connectors each go for about $15. On top of that I needed a variety of IDC type connectors to breakout the encoder headers and the 5th-7th axis. I am probably going to be in for about $750 by the time I finish up the interface.

I think it is going to be worth it in the end, but I still have a steep road ahead.

Thanks for all the input so far!
Jason

[jacampb2]

Well, this is getting expensive... I finally got my third drive. I've been keeping an eye on ebay for the last few weeks and a decent deal finally showed up. This one is another A-M-C drive, but is +/- 10V analog command. A bit smaller than the others, it is rated at 400Vdc, 40A max, 20A continuous.

This drive does it's own commutation, so I ended up ordering the previously mentioned Renco encoder, and since they are reasonably priced I ordered 3 to I won't have to do wake and shake on the other drives. Total for three Renco encoders, 2048PPR w/ index and hall tracks was $215 shipped. Not to bad IMO, but I will have to align them with the field. It looks like it is not going to be very difficult thanks to excellent instructions on the Renco web site.

I did the CAD drawings for the BOB's in eagle, and the mill is cutting the last one right now. They are nothing fancy, they just break out two 50 pin IDC headers and one 26 pin IDC header to screw terminals. The Galil controller uses a 100 pin high density MDR connector for axis 1-4 and IDC headers for everything else. The 100 pin MDR connector is a B*tch! I ruined one already trying to make the cable, so I ordered two more and and going to try my hand at making a fixture to align and hold everything. 3M gets over $3K for their "hand" version of the crimper, so I figure I can go through a few $15 connectors as a learning experience and still be ahead of the game.

Hopefully I'll have my BOB's and cables made by the end of the day and will take some pictures.

Thanks,
Jason

[macona]

Well, this is getting expensive...
Thanks,
Jason

Told ya to get that set of 3 yaskawas or mitsubishis! ;)

-Jerry

[Hood]

What are the model numbers of the encoders you are getting, just checked with the quote I got and I must have annoyed them or something as they quoted me $320 plus $150 shipping to the UK for one.

Hood

[jacampb2]

RCML15, they are a normal stocked model. Maybe you selected one of their smaller models which appear to olny be custom built. The RCML15 is a 1.5" and all the encoder counts up to 2048 can be ordered right on their web site. The RCML15 goes up to 5000 ppr, but anything above 2048 is a custom order. Also, it does say on their site that there is a 15% premium for international orders.

Later,
Jason

[jacampb2]

Told ya to get that set of 3 yaskawas or mitsubishis! ;)

-Jerry

Yeah... yeah... but I should have one sweet motion controller when I am done.

You probably don't remember, but back when I first got this mill you tried to talk me into going servos at that time and I couldn't fathom wanting the machine faster than what it was going to be capable of with steppers. I should have listened then, because the retrofit cost me about $1500 then, and now most of the stuff won't even be used this time-- not that it went to waste, I'll find some use for all the stepper crap eventually.

Later,
Jason

[Hood]

I need the bigger encoders as the shaft size is 10mm and I think the biggest the 1.5 inch ones go to is 8mm
Could I suppose have made up an extension shaft so I could mount the encoder further out but in the end I picked up a couple of motors on US Ebay and got them sent to a friend who stripped off the encoders and sent them to me. Total cost of that was under $50 :)

Hood

[Hood]

Probably too late as you seem to have got the breakout etc well on the road but I found this place when I was searching, bit cheaper than Galil but its just a straight connector board.
http://cyberresearchuniversity.net/store/cables-wiring/terminal-panels-breakout-wiring-termination-blocks/scsi_breakout_panels/TP_DIN100S_1670.2.htm

I have never done anything with my Galil, I picked it up at the right price so I can wait for the right machine to come along.

Hood

[jacampb2]

Yeah, I saw those, but still $110 each and I need two for my card. Plus you still need the auxillary breakout for inputs and halls, I think I have less than $30 in the pair of mine.

[jacampb2]

Well, it took a bit longer than expected, but I finished up today. The new fixture helped a ton for the MDR connector. I actually just copied the lower platen from 3M's pictures of the crimper, machined something similar out of scrap 1" ABS and used 3M double sided super tape to keep the wires aligned as I put them in place. Crimping it turned out to be an interesting exercise. I kind of mangled the shell, but it still works. If I end up doing any more of these, I am going to build a little press for it as well. Can't crimp them in a standard IDC press because the pair of ribbons come out the center of the connector.

Anyhow, it worked. Both BOB's are done, bellow is a picture of the cable and the BOB. I know the layout of the BOB isn't stellar, but it allowed me to keep it single sided. If it is to crowded in the inside parallel rows of screw terminals, I'll just separate the boards into individual 50-50-26 pin boards.

So, here are the pics:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/low_mdr001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/low_mdr002.jpg

Later,
Jason

[Hood]

Looks like they should work well :)
Hood

[Jim Blackwood]

I don't intend to hijack your thread but this seems relevant to what I'm trying to do so I'd like to keep up with what you're doing. I have some good sized servos on my x/y axis. Parker Compumotor 3.36 KW, peak stall 45 Amps so I'm guessing they'll take a pretty solid driver. I aim to send a note about those Dugong drivers to ask if they'd recommend them. I don't think I'll need full performance out of the servos or anywhere near it as that has the potential to give me nearly 1000 ipm feed. Hopefully their drivers will handle my other servos as well.

I figure to work back towards the controller from the servos, is this a good approach? I've barely even scratched the surface regarding the rest of the system, although I do have one or two computers that I think might do a decent job. I may be able to order servo drivers right away. Are you happy with the ones you bought?

Jim

[jacampb2]

If I recall from my research the dugong drives max out at 80Vdc and 20A. What is the voltage for your big servos? Most of the higher power servos are going to be higher voltage. Running a 200V servo on 80V max is going to significantly decrease your speed. The granite devices drives may be a better option, they are priced a bit higher, but can do 160Vdc and 20A. Either of those drives will take step and direction, so you can use the standard Mach3 or EMC2 parallel port interface.

The reason I went w/ Galil was because I bought before I thoroughly researched on my drives. My two sinusoidal command drives will only work with a few industrial type motion controllers. They are huge drives though, and more than capable of running my servos to their full potential, so I think it will have been worth it in the end. A lot of the Galil cards will also work with Mach3 via plug in, and should be far superior to the PP interface. The Galil card can handle 3Mhz encoder frequency, 12Mhz quadrature data. This is far beyond what you can get out of Mach w/ PP interface. That allows you to use a much higher count encoder and get increased resolution and accuracy.

You can get the Galil cards fro cheap too, I just picked up a second one for another project. 3 axis card for $100. The Galil cards will allow you to use inexpensive analog servo amps too. You can find 200Vdc 20A brushless analog command amps on ebay all day long for under $100 each. I just picked up a few A-M-C 400V 20A +/-10V analog command amps for $65 each.

But, what Macona has to say earlier on is a very good route to take as well, buy a package with drives and motors for under $1500 and you are doing damned well. I'm happy with where I am going with this, but even though I started w/ the servos first, I may have been money ahead to have still just shelved those and bought a package. I am to stubborn for that, so here we are :D As it is, I paid more for 3 new encoders than I gave for the 3 servos to start this off.

Later,
Jason

[macona]

The Dugongs are 160v at 35 amps.

Unless those 5HP motors are on a HMC these drives will easily handle it. They are brushed servos, right?

[Jim Blackwood]

The servos are brushless and the reply I got was that the Dugong driver will not run them so I'll be looking for another solution. I'm still learning. About those Galli cards, I'm guessing those go between the controller and a servo amp. So is a Servo Driver a combination of the two or do I still need interface cards? I can see this is going to cost a little more than I had hoped. So $100+ for Galli cards and $65+ for amps, or buy servo drivers from Granite or somebody instead? Will a PC work as the controller or do I need another interface of some kind there also?

It's hard for me to imagine getting 5hp out of the Parkers but I guess it's really the torque that's important, although my table really isn't that hard to move. I guess I'll have plenty of rigidity in the motion assemblies.

Here are the specs:
- X and Y axis: Parker Compumotor brushless motor

Model #: Z/ZX 610 Ser #: 92C206D-81

3.36 KW max power

Stall Cont. (RMS Amps) 15.0

Stall Peak 45.0

Max Speed 7000 rpm

KB (V/Krpm) 27.8

Inertia (g/cm*) 2512

TC (Nm) 6.51

TP 18.6

Volts (Rms L/L) 230

Res L/L @ 25*C 0.38



- Quill and rotary table: Pacific Scientific Brushless Servo

Model #: R34SENT – TS – NS – NV – 02 Ser #: 637783

TCS 25 Lb-in.

ICS 4 Amps

Ke 80 V/Krpm

RL 6.9 Ohms



- Knee

Motor Power Company (Italy) Motion Line

Type: Penta 5M 30 B14/M63 CN Ser #: 0288140

P/N: 1007505500008

Watt: 360

Volt: 60

NM: 1.15

FF: 1

RPM: 3000

Amp: 7.7

Duty: S1

Prot IP: 54

Ins Cl: F

This motor drives through a gear reduction angle drive. All others are direct drive.

I don't know much about the last one (Italian). But anyway all the motors except the two Pacific Scientific ones are mounted and coupled so I'd like to just go ahead and use them if I can. I realize they are probably overkill.... I do that with my car too. But not having the finely tuned sense of what's needed like the old timers here I reasoned it was better to go too big than too small. What probably threw me off is that these motors apparently deliver incredible power for their size.

Jim

[jacampb2]

First off, let me say I am still a noob at this, I am learning fairly rapidly, but don't let my comments be the end all in your decisions. If I make a mistake in an explanation, I am sure other will correct me, and I appologize ahead of time.

Servo drives and Servo amps are basically the same thing. The dugong and granite devices amps are positioning amps, that is fairly rare in servo amp offerings. Most amps rely on the controller to do the positioning. Positioning amps can take step and direction, like Mach uses, and use their built in control to reach and hold that position. Both the dugong and the granite amps were designed with the hobbiest in mind, not so much the industrial customer. I believe the Allen Bradley 1398 series drives are also capable of taking step/direction, I am sure there are others that people can clue you in on as well.

The way the Galil controller works, is it takes the place of the PP interface in Mach. It is a PC card that handles all IO w/ the machine. What you need to be aware of, is that Mach can only use one Galil card at a time, so buy a card with enough axis to support your machine, including spindle control if you are going to control spindle speed with mach. W/ the galil card, the PP interface is disabled. The galil cards are available w/ up to 8 axis per card. I bought ISA bus cards because they are cheaper, however, if you can't build a PC, then it would be better for buy their ethernet based or PCI based controller. ISA is pretty much a dead end computer architecture. You can still buy current industrial boards w/ ISA bus, and build a decent machine with ISA slots, but you are not going to find any off the shelf PC's w/ ISA anymore. Check out the Mach3 forum and the Galil section before buying anything, there is a limited number of controllers that will work.

As for cheaper analog amps, they work just as well as expensive ones. As the cost goes up, it is typically because of more features and control on board in the amp. It takes the load off the controller, but it isn't always needed. You can find all sorts of plane jane analog amps out there, and if you go w/ a decent controller like the Galil, then it does the majority of the processing and positioning.

Good Luck,
Jason

[macona]

Many of the newer brushless drives support positioning. Allen Bradley makes a real nice drive. I am using one of the 1398 series drives in my lathe and Hood uses some of the 2098. There are newer ones available too.

[Jim Blackwood]

I've been trying to absorb this and it seems to be making a little more sense. Sounds like a laptop PC is not the way to go if the Galil cards plug into a motherboard slot. Luckily for me we will be replacing my son's PC (Sony Vaio) with a laptop so maybe I can use that. It was a very good machine in it's day but that was awhile back. I don't know if it uses ISA or PCI architecture but I'm guessing PCI, I think it came out towards the end of the switch-over. I should probably dig up the specs on it. I have 4 axis on the machine and a 5th for the rotary table. Can a servo drive run the main 2hp 3 phase spindle motor? That would be pretty cool. I have a VFD on it right now but it's undersized. Works for most things though. Then there are the carousel, tool changer, and drawbar to sort out and I don't know exactly what will be needed for each of those but some sort of output at least and the carousel and the tool changer each have motors on them. Not servo or steppers, although I suppose they could be fitted if needed Right now they use microswitches for positioning, although I could fit proximity sensors since I have some surplus ones.

So maybe I should just go for an 8 axis board? And my luck they are probably scarce enough that I'll have to buy new I'm guessing. Looking at their prices I can't see any way clear to do that.

But with the Galil I can use the cheaper servo amps? And if one proves to be not powerful enough I guess I can just be swapped out for one that is?

With this setup can I record manual moves so that I can use them to create a run program later? That would be a really big bonus and something I really want to be able to do.

Then what about power supplies? Do these servo amps still require big honkin DC power supplies to run them, or can they now take their power from the AC line like a VFD?

Sounds like I should plan on another month or so just to figure out what to buy. I'm reading and studying but it goes slow. I'm not a real quick study on this stuff.

Jim

[macona]

A laptop is never the way to go on a CNC machine. There are energy savings processes that occur and interrupt the pulse train.

You need a VFD for the spindle.

The toolchanger system can be controlled by a PLC or maybe a ModIO. I used a modIO on my tool changer for my little lathe.

There is a way to record manual moves, but these are all through the MDI, keyboard g-code input. If you are expecting to turn and wheels and it records, well, this isnt a Haas.

Some servo drives have integral power supplies, some dont.

From your questions I will say this. You may be getting in way over your head on this one. I would start small. This is a giant project that will probably take months.

[jacampb2]

You shouldn't need an 8 axis card. Not unless you go w/ sinusoidal control amps, these take 2 axis per amp to run. All other command types only use one axis. So, buy +/-10V analog command amps and you only need one axis per amp.

The galil card can control the VFD via the 10V analog command input available on nearly every VFD. It is possible to convert the spindle motor to servo motor and use a servo amp as well. I can't see much real benefit though unless you need spindle positioning, or very precise spindle speed control.

Macona has you covered on the tool changer, PLC, or you may be able to write a VB script for mach to use the IOs you have available w/ the galil card. I don't really know much about tool changing, but the galil cards have many more IOs than a parallel interface does. 16/16 uncommitted inputs/outputs respectively. That is not counting the IO already there for encoders, 5 differential inputs for each axis, and inputs for forward/reverse limits and a home on each axis. Galil's 17x8 and 18x8 cards also include an additional uncommitted 64 I/Os. Galil also makes a PLC that might work for you, but you will need to do more research. I believe with their Ethernet based controllers, that a few have gotten Mach running with the PLC and Controller on a single machine. Do some reading and asking on the Mach forum.

Don't buy new unless you really want to or cant wait. I see the 8 axis cards quite often, and sometimes new on Ebay, most people want in the neighborhood of $1500 for a new one on ebay, but this is still significantly less than new from Galil. I got my seven axis card for $400. Watch for them to come up for auction instead of BIN.

If you are prepared to take your time and learn what you need to know, then I say go for it. I was pretty much in the same boat as you a couple months ago. I didn't know a brushed from a brushless servo... If you need the machine to run tomorrow, then I agree w/ macona, you're in over your head. In my case, I have the mill running pretty much perfect on steppers right now, so I can afford to take my time and get all my ducks in a row...

Later,
Jason

[Jim Blackwood]

Appreciate the advice guys. Yep, no doubt about it, I'm in over my head. But somehow that doesn't seem like a valid excuse for throwing in the towel just yet. I may still be able to sort this thing out. Time is pretty much on my side, it's taken me about 5 years to get the mill cleaned up, back together and usable manually and I must say, though it is lacking in ways it is a big improvement over the little bench top mill I've been using. Guess I got in a hurry because I have a few bucks I can spend on it right now, and it sounds like my best bet might be to set up ebay searches for 7 and 8 axis galil cards and see what comes up, then after that worry about what PC and what amps. How's that sound? I can build a PC (done it before) to match whatever card I find. Maybe that will give me a better chance to get familiar with how everything works together.

But let me ask this. I know some mills have resolvers on them for manual input and I just assumed the signal went to the controller and then on through to the servos. The resolvers I've seen (several years back) were expensive, on the order of several hundred dollars each, but for the way I work it would be worth it to have that. Especially if the controller will keep track of the resolver moves. Can that be done? (I saw something on the Galil site that led me to think that their newer cards have some sort of capability along those lines)

Jim

[Jim Blackwood]

I have another question, what is the significance of the different series numbers in the Galil line such as 1080, 1280, 1780, 1880 etc? I understand the next to last number is the number of axis but what does the first number pair mean and which are the most desirable boards?

Jim

[Hood]

With this setup can I record manual moves so that I can use them to create a run program later? That would be a really big bonus and something I really want to be able to do.
Jim

You cant actually turn manual handwheels and record the code but there are a couple of similar ways in Mach. First is you can MDI a line of code at a time and have it write to a file so that you can just run the file after that.
The other way is the JCode plugin, it records your jog moves and writes the code. I have never tried it so not sure if it can record MPG moves (Electronic hand wheel) or if its just keyboard jogs.

I have a DMC 2140 Galil that I bought and have never used and probably wont, not sure if I will be selling it but I would probably want around about £200 plus shipping for it, so would likely make it expensive for you USA guys. Its a 4 axis standalone controller, Ethernet or Serial comms I think.

Hood

[jacampb2]

You can use a manual encoder/handwheel in mach for manual axis control, you can also use a PC joystick, or Xbox game pad. Lots of options in the manual control department. I can almost guarantee you that once you find a good software (cad/cam) package you like, you won't be worried about manual input anymore. I too was wondering the same things about manual input and recording when I first set up my mill. I never got around to doing it, and I dont miss it either.

I would think that if the servos are not trying to hold position, but the encoders are still powered, that Mach will still keep track of position, even if you use manual hand wheels. I know that Plasmacam brand plasma tables work like this. You can manually drag the gantry to wherever you want and it keeps track of position.

With a 5 axis machine, plus spindle control, you only really need 6 axis. If I were you, I would search for that as well, fewer axis will be cheaper, as soon as you go over 4 axis, you get all the additional inputs for the 5-6ths axis, even if you don't have them.

The PC is really no problem. You can still buy industrial motherboards that take current intel processors, and have ISA if that is what you end up with. Itox is the brand I ended up going with, it's running a 2.55ghz pentium, 2 gigs of ram, and XP pro, plenty fast for the mill controller since it will be doing nothing else and the motion control card takes pretty much all the load of processing.

The difference in the first two numbers is the controller model series. 10xx is an old ISA bus card and is not even supported by Mach. 12xx is PC/104 bus, I haven't seen anything about this on the Mach forum, but it seems to be the PC/104 version of the 17xx controllers. PC/104 is the same buss as ISA, however, it is for SBC (slot based computers) and the cards are stackable. I'd ask on the Mach support forum before I went with one, because they are not listed as a supported controller. 17xx series is the ISA buss cards. 18x0 series is PCI bus, identical to the 17xx series except for bus interface. Both the 17xx and 18x0 are their "Optima" line, which was their flagship line a few years ago. Now it is the "Accelera" and they are denoted as 18x6. The 17x8 and 18x8 are the Optima cards with the extended IO option. 22xx is Optima Ethernet based, and 20xx is Optima USB based (I can't recall if the USB one is supported, I think that they got it running, but I dont recall off hand). 21x3 is the Econo Ethernet based controller, and 40x0 is the Accelera Ethernet controller (both current models) I am not sure if these are supported by mach or not, but I think I read they were.

The Accelera controllers seem to be the most feature packed, but I haven't seen many come up on the bay. The Accelera cards support contour mode where all the rest only work in linear interpolation mode. I'm not sure how much of an impact this will have on your decisions.

Good Luck,
Jason

[Hood]

I would think that if the servos are not trying to hold position, but the encoders are still powered, that Mach will still keep track of position, even if you use manual hand wheels.

No Mach is open loop, if using digital servo drives or external motion controllers then the loop is closed to them but Mach doesnt know anything of it. You could track the encoders in Mach and probably write some VB to update the machine DROs and if you were clever enough you could likely write a plugin to do that but is there really a need?


21x3 is the Econo Ethernet based controller, and 40x0 is the Accelera Ethernet controller (both current models) I am not sure if these are supported by mach or not, but I think I read they were.

Never really looked at the Galil much but the one I have is an Optima series, WTF that means I have no idea :D Heres a link to it http://www.galilmc.com/images/products/optima/dmc2100.html.old It is supported by Mach as thats the reason I bought it when I saw it on eBay UK

Hood

[fasto]

Then there are the carousel, tool changer, and drawbar to sort out and I don't know exactly what will be needed for each of those but some sort of output at least and the carousel and the tool changer each have motors on them. Not servo or steppers, although I suppose they could be fitted if needed Right now they use microswitches for positioning, although I could fit proximity sensors since I have some surplus ones.

Hi JIm, the umbrella toolchanger on my Hurco uses a lot more I/O than you'd think, in fact most of the electrical cabinet is taken up by relays for the toolchanger!

From memory, it's got Carousel Left, Carousel Right, Carousel Up, Carousel Down, and Spindle Unclamp for pneumatic valves (ie, outputs).
It's also got Carousel Rotate and Carousel Direction outputs that are very cleverly set up to use a 3-phase motor as a reversible single phase motor.
So, that's 5 outputs for pneumatics and 2 for electricals.

For inputs, it's got a mess of microswitches. For the carousel, Carousel Left, Right, Up, Down. For the drawbar, Clamped and Unclamped.
It's got 2 proximity sensors: Tool #0 (to home the carousel) and Geneva Wheel Parked to know when to turn off the carousel rotation motor.
It's got three pushbuttons, Spindle Unclamp, Carousel Rotate CW and Carousel Rotate CCW used for manually changing tools and manually rotating the carousel. It's got an input from the spindle drive for "orient complete" as the orientation is done in the spindle drive. It's got a input from the spindle drive for "spindle stopped" because it won't let you manually unclamp the spindle if it's running - the Spindle Unclamp relay is locked out if the spindle is not stopped, in fact.
So, that's 13 inputs.

It's got the toolchange logic both in the relays and in the control, which you'd need to put together somehow in your PLC. Tracking tools is done in the control, so you know how far to index the carousel to get to the next tool. Somehow you'll need to implement that.

These simple things are somewhat more complicated than you'd think!

[Jim Blackwood]

I know the toolchanger and carousel won't be the simplest thing, so I'm leaving that for last. But, if I need servo drive for it I need to plan ahead. I don't think it will be necessary though. The carousel uses a 4 bit microswitch encoder for position and a mechanical latch. It may be rotated pneumatically, I'll have to look. The tool changer rotates via a motor of unknown type and also has pneumatics. The drawbar tensioner is something I'll have to make, and spindle indexing is likely to be an interesting challenge. Maybe I can use the VFD's jog function and a prox.

But for now that's getting ahead of myself. I presently have 4 axis on the machine with a 5th servo to add to the rotary table. I can't think of what I might need a 6th axis for, and perhaps I should just look for a 5 axis card. If sometime in the future I end up with something exotic like a 2 axis rotary table I'll just have to upgrade. Anyway my gut feeling on this is that I won't have too many choices and will pretty much have to make do with what I can find so it may be a minor point anyway.

Sounds like I need to steer away from the lower number Galil cards or at least the 10 series if I can. I looked at the computer I'll have available for this. It uses PCI buss and is a 1Ghz machine with 386M of RAM. I thought it was a little better machine than that but I guess it'll do OK at least to start with. Y'all have convinced me that Galil is the best way to go so I'll concentrate on getting one of their controllers and then worry about the servo drives. I got a note back from Granite that said they thought their drives would handle my servos pretty well, with some limitation of the Parker servos so I'll take another look at their site. I don't know what kind of money I'd run into there but I guess I might as well find out. I'm guessing these are some of the more capable (and therefore more expensive) servo drivers.

I know all of you feel that saving manual inputs probably isn't practical and maybe you're right, but in my shop I don't run production. Occasionally I'll make duplicate parts but more often they change from one part to the next. Sometimes only one small element changes, sometimes the whole part. On rare occasions I may do a half dozen or so identical parts. In most cases, the changes are made "on the fly", and if I could compare it to a standard manufacturing environment, it would probably be one where you have a very excellent, highly skilled, greatly talented and vastly experienced "old man" master machinist who has just about seen and done it all, and he's working for a very poor engineer who doesn't have any skills, doesn't understand his tools, and rarely gets anything accomplished. Obviously here the old man is going to have the job done before the kid finishes his coffee in the morning, even if he has to use a full manual machine to do it. And I'm pretty sure you've all seen this scenario played out or one similar to it. Well in terms of engineering you can compare me to that kid. So why do I want a numeric machine? Because it can do the moves more smoothly than I can. It can do a better job of making the cuts, drilling the hole patterns, and just about everything else a machine can do. It can save canned cycles and use them again later with different dimensions and different tools. In the end, I'm way more comfortable with handwheels than I am with buttons. I know it's backwards, I guess I am too. It took me FOREVER to take up typing and I still do it with 2 fingers. ( my speed has gotten better though)

Jim

[jacampb2]

A 1Ghz machine is bellow the minimum requirements for Mach3, From what I understand though, this is mainly due to the Mach pulsing engine which is not used with a motion control card. Long story short, I read on the Mach forum that one gentleman has a galil card running in an 800Mhz machine w/ mach and no issues.

FWIW, I paid I think $40 for the Itox board I bought, $.50 + shipping for the processor, and I think $29 for one stick of ram from newegg, I had the other stick in my parts pile. I also had a 80gig drive and CD roms and what not around, pretty cheap for a fast(er) machine and still retaining ISA buss. Under $100 anyhow for the whole PC.

I hear you on manual control. My BP has no handwheels at all. I thought it was going to suck, but it doesn't. I can make a CAD change and send it through the CAM software faster than I can machine it by hand. For rinky dink crap that I dont want to draw, well, I still have a manual machine for that! That tends to be the little one off "by eye" widgets anyhow, so I take the time to draw most of my stuff in CAD. It gets easier w/ time, and you seem sharp enough. I stand by my opinion that you are not going to miss it once you start using a good CAD/CAM package...

Later,
Jason

[Jim Blackwood]

Thanks Jason, that helps. Hopefully I've just simplified things some, or at least added some definition. I just accepted the counteroffer on a 2160 6 axis Galil controller. I probably should have checked with you guys first, hope I didn't make a $500+ mistake.

Turns out both the motors on the carousel and the tool changer are 115v single phase so that shouldn't be a problem. So I'm left with one unused axis which could be handy eventually. What is the best way to check out that controller when it comes in to make sure it isn't screwed up? Also is there a source for the cannon type connectors on the Parker Servos?

Jim

[jacampb2]

Turns out both the motors on the carousel and the tool changer are 115v single phase so that shouldn't be a problem. So I'm left with one unused axis which could be handy eventually. What is the best way to check out that controller when it comes in to make sure it isn't screwed up? Also is there a source for the cannon type connectors on the Parker Servos?

Sounds like you did alright. I only have a couple of minutes, but to check it out, very basically, DL the user manual for the controller from Galil, you have to register, but it is free. Follow the setup chapter and install smart term on your PC and then the card. Smart Term is free, but has no advance tuning features, the mach team recommends smart term and not any of their other software packs. The setup chapter in the manual will describe basic tests to ensure the controller is properly communicating. I'll try to get some more info for you tonight.

Later,
Jason

[jacampb2]

Jim, can you take a picture of your connector on the servos? By "cannon" do you mean a threaded round connector with multiple pins in it? If so, it is likely a AMP brand connector. If you can't find the mate in their online catalog, I may be able to help you. Both Digikey and Mouser stock most of the AMP lineup.

Later,
Jason

[Jim Blackwood]

Jason, I've got a photo but apparently don't have permission to upload images.

Jim

[jacampb2]

Jason, I've got a photo but apparently don't have permission to upload images.

Jim

Yeah, you can't upload photos directly to this BBS, you have to use an external host. There are lots of free ones, take a look at the sticky thread at the top of the general forum on how to post pics.

[jacampb2]

I made a little more progress today. I got my differential encoder break out board designed and cut one on the mill and soldered it up. The Renco encoders I bought are single ended, so this little board (1.5"x1.6") has a AM26C31 Quad differential line driver and takes the encoders wire connections via screw terminal blocks, passes through the halls and converts the single ended encoder data to differential. All of this is brought out to a DB15 connector. A lot of the stuff I'll be using had DB15 for interconnects, so to consolidate the number for connectors I have to go with, I figured I'd use it here too. This board will fit in the back of the motor/encoder housing and the DB connector will face down, under the cover, it should be fairly safe.

Here are the pics, I still have to test it to make sure the line driver works as intended.

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/encoder_differential_bob001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/encoder_differential_bob002.jpg

Later,
Jason

[Jim Blackwood]

That's a right interesting little board, I've been looking at that and hoping I can get by with not doing it.

Well, it looks like I've got plenty of time to sort this out, as I've run out of money again and still need the servo drivers. But the good news is that the controller came in and it looks good at least. I'll try to get some pictures uploaded to a site in the next day or so in order to post them here. Looks like I'll have to get several connectors and figure out how to wire them, starting with power. I think I can upgrade the processor in that Vaio computer to maybe a fast enough level, and then I guess I can buy the servo drivers one at a time as I go along.

Well, keep plugging away Jason. It's about the only thing that really works for me.

Jim

[Jim Blackwood]

I got Facebook working so here's a photo of the connectors on the Parker motor:

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/?action=view&current=MVC-547F.jpg

And here's one of the Galil controller:

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/?action=view&current=MVC-550F.jpg

I'd have posted the photos themselves instead of the links but I haven't figured out how to do that yet.

Jim

[macona]

Those are Bendix brand connectors. I think ITT Cannon makes the smaller one too.

Be ready for sticker shock.

[macona]

The connectors on the galil are often referred to as mini-centronics.

The white power connector is a AMP brand connector. You will want to buy the crimping tool for the pins. You can get cheaper ones from molex for about $15. Well worth the money saves sooooo much headache.

I have to say I am curious how the Galil will work out. I want to re-retrofit my mill and get some brushless servos on it. I have had a couple issues with the drives that are on it now. I have a couple 1kw Mitsubishi's for the X and Y, a 1.5kw for the knee, and either a 2, 3.5 or 5kw for the spindle. They take step/dir or analog in. It would be nice to eliminate the parallel port.

[jacampb2]

The galil D shaped connectors are Tyco/Amp brand Champ series 2 connectors. 3M's "MDR" line are also a match. I found some others in my searches, but the 3M's were the easiest to find. They are high density d-shape centronics type if you are looking in catalogs. Both Mouser and Digikey has a decent line up of the MDR stuff, but be forwarned, they are a bitch to crimp. You may be further ahead to buy the pre made cables from Galil or 3M.

Macona, I am taking the stab on the conversion in two weeks or so. I'll keep you all posted.

[jacampb2]

BTW Jim, if your current servo connectors are as high priced as Macona says, you might also consider looking through the AMP online part finder and seeing if you can find a harsh duty/weather tight type connector pair from them for a more reasonable price. I know AMP makes circular DIN style and several other similar types of connectors to what you have, make sure you pay attention to current rating for the motor power connectors, you don't want to get a hold of something good for only a few amps and end up pissed! I'm not sure if you are going to get out of it any cheaper, but it is worth a shot.

Later,
Jason

[macona]

Now I am thinking of saying to hell with Mach and replace the whole control with a Milltronics Centurion 7. I can get the whole thing for about $2k. Then I can add a spindle encoder and have rigid tap. Probably wont do it till this summer so I can move the machine outside to work on it.

[Jim Blackwood]

I've been reading the manual, some old programming is coming back to me. First things first, I need to get power to the box (controller). Is it possible to just buy a power cord? The manual has no reference for which pins of the connector do what and since there are 7 of them there's no way to guess. I've been reading about interconnect modules, seems like another expensive part that does what? Connect the controller to the amp and the encoders basically? I don't suppose there'd be a connector with ribbon cables you could split out for that? I think I'll leave shopping for cannon plugs until I get the power cord sorted out.

Macona, that sounds ambitious to me, and I'd like to hear how you get on with it.

Jim

[jacampb2]

Jim, the ICMs are basically what the boards I made are. They are expensive from Galil, but they aren't necessary. They are just break out boards from the HD centronics connectors. You can get the pinouts from the datasheets on Galil's website, if they aren't in your manual as well, mine are in the manual for my card.

I would think Galil will give you the pinout for the power supply, but I am sure you can also purchase a supply. Galil seems to be more than helpful when it comes to integrating their stuff, so I don't think the PS will be secret, just ask them.

I might be able to make you some breakout boards (ICM) if you want, but I will have to look over your controller pinouts to be sure what I already designed will work. I don't really want to make any more MDR cables than I have to, but I can probably help you out if need be. The factory cables run over $100 a piece, and this is from anyone that makes them, not just Galil!

You could conceivably just cut a single MDR cable in two and terminate it at barrier strips for your connections, not as nice and simple as a BOB, but it will do the same job.

Later,
Jason

[macona]

Actually installing a milltronics control would be sooo much simpler than doing what you guys ares doing. If I get the Centurion 7 control it is two parts. First part is the control itself. Its a panel maybe 20" x 20" with the LCD, MPG, buttons, and keyboard. On the back is the computer. PC104 based computer boots off a ramdisk (Supports a HD too, as well as Ethernet). This whole thing is maybe 6" deep.

The second part is the interface. It is a unit maybe 12" x 6" by 1-1/2" thick. It is connected via fiber optics to the control panel. it handles all the IO. Analog signals for the axis's and spindle. Digital inputs for limits, mpg's, spindle and axis encoders, etc. I think it also allows you to connect linear scales as well. Then there are outputs as well for everything else like coolant, tool changers, etc. I cant remember for sure what the connectors for the digital IO are.

If I had known I could get a control like this for 2k I would have never touch Mach. The conversation is great on these controls. You can do just about anything on the control short of profiling.

[Jim Blackwood]

Jason, I may have to take you up on those break-out boards.

I contacted Galil and as it turns out my controller was shipped without an internal power supply so beware any Galil controllers with a molex(?) connector for power! Presently I'm trying to talk them into selling me a replacement power supply, and I sure hope it doesn't all go downhill from here. I can't really afford to take a loss on this thing.

Macona, I'd like to learn more about that Milltronics control. Are you using a new or a refurbished unit, etc?

Jim

[macona]

Just find the pinout for the power. Probably +5v, -12, 12v, etc. Nothing big. The connector is made by AMP. The pinouts for it should be in the manual with the rest.

The control is rebuilt. Its a model behind current.

[jacampb2]

The pinout for your power connector is in the manual. Page 221, zoom in on the board picture. It is from top to bottom, +12Vdc, +5Vdc, +5Vdc, ground, ground, -12Vdc, the connector part number is from molex (again, you should be able to get it at digikey or mouser) 26-03-4061, and the pins for the connector are 08-50-0189.

You can easily get all the power supply voltages you need from a hacked ATX computer power supply. It needs 1.1A on the 5v line and 40mA on both the -12 and +12 lines. That is available even from the smallest computer power supplies. Google ATX lab power supply, and you will get a lot of instructions on how to pull it off. Or search for my old thread about my bridgeport in the general section of this board, I explain how to do it in there, but I didn't use the -12V lead in mine.

Later,
Jason

[jacampb2]

So, I just had a look at your picture again, and I think you got the older 21x0 serries, I was thinking you had the 21x3... Anyhow, the power requirements are the same, but you are going to have to open the case and see how that big power plug is wired to the board. The pinout I gave you, was for the header on the board, and it appears that it is still the same for the older version you have. My guess is that your external plug just wires to the internal header. Look there, and you should be able to figure out which pin is which on the external plug, but I don't know what the part number is for that nine pin plug...

[macona]

I have had some atx supplies missing the -12. Most modern boards really dont need it any more. For a supply I recommend a 1U ATX supply. Pretty small.

[ianjkirby]

Hi guys,
I have seen servo controllers which are fast, and reliable. You will have to make up your own mind as to their prices. Have a look at http://www.cncteknix.com
While I know these people, I have no financial interest in their business, but can assure you that they are genuine.
Regards, Ian.

[Jim Blackwood]

You know, I never even considered using a power supply out of a computer, it just never occurred to me. And not long ago I threw away a brand new ATX power supply when I was tidying up. That's timing for you.

Well, Galil has been as helpful as you could hope for, and are willing to sell just the power supply and connector to reconfigure the unit. (Not exactly cheap though at $150 + shipping, but about half what it would have cost to let them install it.) The seller refunded the money to pay for the power supply rather than have the unit returned and pay shipping both ways, so I'm sort of inclined to just go that route. That way it's all self contained, it just plugs in the outlet strip, and I'd have what I bargained for to begin with. One of the attractions of the unit I bought was the self contained power supply. In retrospect I guess I should have been suspicious of the power connector but being it's the first time I bought one of these I just missed it and it isn't a purchase option shown by Galil so I really had no reason to question it. Just one of those unfortunate coincidences really. On the other hand, maybe I could bring a power plug out from the Vaio computer I will be using with the controller, it probably has the correct voltages and probably has a spare plug on the power supply. The down side of that is that it's my son's computer and I won't get it for about another month so I won't be able to test the controller, which I think might be important and which it'd be nice to be able to do here in the office using my laptop.

Jason, I think I'd like to have you make me up those bits if you don't mind too much. Could you send me a PM and give me your terms for the break-out boards? From what I can see it looks like you did a pretty good job on yours and it sure would be helpful.

I've been using the mill quite a lot lately manually and turning cranks can sure get old. It's giving me motivation to get on with this. I fitted the new pulleys for the Y axis so now the 2 main axis are complete mechanically and those are the two Parker Compumotors. Maybe it's time to get back to looking at servo amps.

Jim

[jacampb2]

Jim, save the $150 and do the ATX hack. They are very reliable power supplies, OC and O/U voltage protection. Even the Chinese ones typically only have issues when the cheap fans die. I bought a box lot of like 10 ATX supplies when I did my original B'port conversion, I use them for all sorts of crap.

I would be willing to bet that the board diagrams in your manual match the board inside your case. Just open it up and see how the plug is wired.

I wouldn't run the controller off the PC's supply that you will be using, if you do that, you loose your opto-isolation on all the I/Os.

As for the BOB's, your axillary I/O connector is different, it looks like it is a 80 pin MDR, my BOB would not be directly compatible, but when I get some free time, I will try to redesign it for you. As of today though, my mill is down for the Servo retrofit, it could be a while before I am back up, but I will PM when it is.

Later,
Jason

[jacampb2]

Progress for today: I am installing my new encoders and going through the alignment process. I got one differential driver board installed in the encoder cover, it came out pretty well. I am going to get all my encoders aligned, and then move on to finishing the cover mods to hold the driver boards. Pics will follow later this afternoon!

Later,
Jason

[jacampb2]

Encoders are installed and halls are aligned, no pics, I was exhausted by the time I finished yesterday. Suffice it to say, there is nothing to be scared of, it is a very simple process w/ a dual channel scope. I am working on finishing up my encoder covers right now, then I am going to mill the Nema 42-->servo adapters I designed this morning over breakfast :) Hopefully I will get some pics today.

Later,
Jason

[jacampb2]

Well, the encoder covers are all done, and that is as good as it gets for today. Paint is taking far to long to dry, and I'm going to call it a night. Here are a few pictures of the finished product.

This particular cover cracked in two when I started to modify it. I had to tig it back together, you can see in the bottom of the inside where I milled the weld bead out of the inside of the cover. It is solid and turned out pretty good. The other two covers came off w/o a hitch.

And since I forgot to mention it, I did test the differential line driver yesterday and it works as advertised!

Pics:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_encoder001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_encoder002.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_encoder003.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_encoder004.jpg

Later,
Jason

[Jim Blackwood]

That looks very good Jason. Maybe I should pull the end cover off one of my Parker motors to see how it is equipped. I'll try to do that today and maybe post a photo.

I took your and Macona's advice and ordered one of the small atx power supplies. No cover but it'll be inside a cabinet and was just 10 bucks with free shipping. By gradually chipping away at this I hope to one day have a completed machine. Which will be nice because right now I have no means of measuring travel. In spite of that I manage to do useful work with it, and it will add immeasurably to my shop's capabilities once it is fully functional.

Jim

[jacampb2]

Thanks!

I got a bit more done today. I repainted the encoder covers yesterday, evidently, the rattle can I had was past it's shelf life, it didn't dry completely in 3 days time, so I stripped it and started over.

So, anyways, I needed to come up with some sort of liquid tight enclosure for the main motor power wiring, since these servos came off a robot, they terminated the harness in a 18" pig tail w/ quick connects on the end. At the motor itself the wiring went through a grommet and a little aluminum cover w/ gasket to the motor, probably splash safe, but not sealed enough for the mill.

I looked through the water tight nema enclosure offerings at McMasterCarr, but they are pricey-- starting around $60 for a small one. I decided to go to my local big box home depot and see what I could rig up. I ended up with 3 die cast aluminum single gang boxes rated for wet location.

You can see all the gory details of what I did to them bellow, but the short version is, I milled the threaded conduit boss flush on the inside back of the box, and I milled the back outside of the box flat. I milled .125" off the top of little aluminum covers that originally sealed the motor. I basically turned them into little riser blocks. I made gaskets to seal between the aluminum box and the riser and assembled it all back together and painted them to match the motors. I used a water tight compression/grommet connector for wire entry. I am probably going to have to find the same thing in right angle style, because it looks like they may interfere w/ the encoder connector. The other hole in the box just got a threaded plug that was included in the package. The boxes were only $4.50 each, so it seems it will be a good inexpensive solution. The only thing I am not happy with is the covers. They are blank single gang SS covers w/ a foam gasket. I will probably machine some nice covers w/ real gaskets in the near future.

Anyhow, that is it for today, tomorrow I will hopefully get to the motor mounting brackets...

Here are the pics:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_modifications001.jpg


http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_modifications002.jpg


http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_modifications003.jpg


http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servos_modifications004.jpg

Later,
Jason

[Jim Blackwood]

That's pretty creative. I think your servos look larger than mine. Anyway I got a couple shots of the encoder.

Jim

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=MVC-578F.jpg

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=encoder.jpg

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=MVC-578F.jpg

[jacampb2]

That's pretty creative. I think your servos look larger than mine. Anyway I got a couple shots of the encoder.

Jim

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=MVC-578F.jpg

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=encoder.jpg

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=MVC-578F.jpg


Those pictures look like a bearing, not an encoder??? Did you take the right cover off the motor? There should be several wires attached to the encoder...

My servos probably are bigger, they are 3Kw motors, ~4hp. The series 2 mill is quite a lot larger than an ordinary bridgeport though. Macona mentioned earlier in this thread that I bought serious overkill, but I like to go big :)

[MaxHeadRoom]

That's pretty creative. I think your servos look larger than mine. Anyway I got a couple shots of the encoder.

Jim



They look to me like they could be resolvers?
If 6 wires, this would be an indication.
M.

[MaxHeadRoom]

The Accelera cards support contour mode where all the rest only work in linear interpolation mode. I'm not sure how much of an impact this will have on your decisions.
Jason

Actually the Galil cards back to the legacy DMC-1000 have had Contouring, linear Int and Circular Interpolation all along.
M.

[macona]

That's pretty creative. I think your servos look larger than mine. Anyway I got a couple shots of the encoder.

Jim



Previous art! My mill has weather proof j-boxes on it for connections too.

[macona]

That's pretty creative. I think your servos look larger than mine. Anyway I got a couple shots of the encoder.

Jim

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=MVC-578F.jpg

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=encoder.jpg

http://s940.photobucket.com/albums/ad247/Jim_Blackwood/Moog/?action=view&current=MVC-578F.jpg

Yep, those are resolvers. There are a few drives out there that do support resolvers and feed back simulated encoder signals back to the controller. I have some superior electric ones for a couple motors of mine. Pacific Scientific also makes a drive that does it.

Another option which i used on my mill was to mount an external shaft encoder to the cover plate on the back of the motor and use a coupling to connect to the shaft sticking out from the tachometer.

-Jerry

[jacampb2]

Actually the Galil cards back to the legacy DMC-1000 have had Contouring, linear Int and Circular Interpolation all along.
M.

Yes, I know, but according to the Mach team that developed the plugin, you can not use contour mode with the anything other than the accelera controllers. I am not going from personal experience, just what the developers have to say in the plugin's instructions and the support forum. I have no idea why this is the case, I will try mine in contour mode and see, but I suspect they did their homework.

[jacampb2]

Previous art! My mill has weather proof j-boxes on it for connections too.

Doh! Now you are starting to sound like Evan! If I had known you had done it too, I would have given credit where credit was due. Sorry!

It just goes to show that great minds do truly think alike :D

Later,
Jason

[MaxHeadRoom]

Yes, I know, but according to the Mach team that developed the plugin, you can not use contour mode with the anything other than the accelera controllers.

I was just questioning the statement in general, I don't know what the limitations of Mach are, I use the Galil with other software so it may not apply.
M.

[jacampb2]

I don't think it is a limitation of Mach3, but rather something in the plug in. Maybe they had difficulty implamenting it w/ the earlier controller versions. Supposedly the plug in, w/ an accelera controller, and Mach3 will do contour mode.

I'll let you know more when I get up an running. Out of curriosity, what software do you use the Galil with?

Later,
Jason

[MaxHeadRoom]

I don't think it is a limitation of Mach3, but rather something in the plug in. Maybe they had difficulty implamenting it w/ the earlier controller versions. Supposedly the plug in, w/ an accelera controller, and Mach3 will do contour mode.

I'll let you know more when I get up an running. Out of curriosity, what software do you use the Galil with?
Jason

I looked back into the contour mode an I believe it is a timing issue, communication with the controller and the card is critical with the contour mode due to the timing involved.
The software is custom for the older legacy cards, DMC-1000/1500 series.
M.

[Jim Blackwood]

OK, resolvers vs encoders, what's that mean in a practical sense? More money to make it work? Functionally, how are they different from each other?

I seem to recall reading that my controller would do contouring but I guess that's something I can get into later. I'm still waiting on my power supply, and I need to get on digikey and look up some connectors and get them ordered. Otherwise nothing new here.

Jim

[MaxHeadRoom]

There are quite a few differences, resolvers decode a sin/cos signal at every point on 360deg electrical rotation and as such require analogue decoding.
They are much more immune to vibration and high temperatures than digital encoder.
Resolvers have three windings, one, a primary, is used to induce a field in the rotor, the other two are used to detect the phase angle between the signal induced in them by the rotor.
Resolvers have largely been replaced by quadrature encoders particularly now that commutation tracks have been included on the disk and have replaced the need for hall effect devices.
The digital quadrature encoder has two wave forms 90deg apart, hence quadrature, one is used for count the other direction.
Modern controls often use the quadrature pulse edges to multiply either by 2 or by 4 to increase the resolution above the basic pulses/rev.
Today you will probably find more drives that use quadrature encoders over resolvers.
M.

[macona]

You will find resolvers still in harsh environment equipment. Encoders rely on a glass disk that has an etched or thin film deposited on the surface.

There is an IC out there by Analog Devices that will drive a resolver, process the signal from it and output it as a quadrature encoder signal. I was going to do this on my CNC lathe for the spindle resolver but I just decided to yank the resolver and install an encoder in its place.

[jacampb2]

Well, I got my motor adapters made. 1" aluminum plate, I still have to thread the two bolt circles, but I already checked fit on the servos and they are perfect. I'll take some more pics of them tomorrow w/ them bolted up to the servos.

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapter001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapter002.jpg

Later,
Jason

[jacampb2]

Well, I got the adapters all finished up today, all the holes are tapped and they mount up perfect.

Next on the agenda was lengthening the metric shaft and adapting it to something imperial. The stepper motor shafts are .625", and originally I was going to turn an adapter to go from the servo's 22mm shaft to .625-- well, w/o spacing the servos out even further, like another .75-1", the bore for the servo shaft would break through the smaller shaft. I could have pulled it off by shortening the servo shaft a bit, but I wasn't comfortable w/ that thin of a shoulder between the shaft sizes.

My Z axis timing pulley is in really rough shape, and the only reason I didn't buy a new one last time around is that these 11 tooth H series pulleys don't seem to be available anywhere. I am thinking I am going to order 22 tooth pulleys and bushings for all axis. This will mean the drive will be 1:1 now, The servos should have plenty of power to make up for the lack of reduction, and my overall speed will be even higher. Maybe to high. Give me your thoughts, I don't want to end up w/ an odd ball rev/inch ratio, that is why I am thinking matching the driven pulleys. I could get 14 tooth pulleys, but how will this effect my over all machine?

Anyhow, the shaft adapter is 1.375 OD on the servo end, 22mm bore, 1.1875 OD on the shaft end. I'm open to any suggestion on pulley configuration, I was planning on using pulleys from Mcmaster, between the bushes and the pulleys I am looking at about $60 per axis. Then I will also need new belts if I go as big as a 22 tooth pulley.

Here are the pictures:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters002.jpg

silly picture limit....

[jacampb2]

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters003.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters004.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters005.jpg

Later,
Jason

[macona]

Watch your loading on those bearings. That quite a shaft extension. I would almost put a bearing in adapter plate. a low profile bearing would work fine.

Dont worry about odd ratios. Mach can handle it no problem. I have some real odd ball ones on the Z of my mill and the little lathe.

[Jim Blackwood]

What was the reason for the long shaft extension again? I agree with Jerry, that seems like a lot of shaft extension if it's side loaded with a pulley, but of course the servos do have bearings on each end spaced pretty far apart so you might get by with it.

I went with straight 1:1 pulley ratios also on the x and y axis but just for simplicity, as it meant one less variable to deal with later. I don't think from what I've heard that it is really an issue as it can be taken account of in more than one way later. Looks like it's coming along well Jason.

My power supply came in today so naturally the first thing I did was to pull the board and cut off all the connectors so I could sort out the wires. I have scads of black and red (5v), some yellow (12v), one blue (-12v), one white (-5v), three oranges (unknown), a green, grey, brown, and purple. I know there has to be a circuit completed to turn it on so any advice here would be a big help. Thx,

Jim

[Hood]

Green to Black on an ATX PSU (if thats what you have? ) will switch it on.
Hood

[jacampb2]

Jim, don't turn the supply on w/o a load on the 5v line. Check out this link and search atx supply on the same website for more "how to's" Cliky (http://www.instructables.com/id/Converting-a-computer-ATX-power-supply-to-a-really/). Most guide out there say to use a power resistor from 5V to ground. On all of my supplies, I just use an led and current limiting resistor from 5V to ground. It provides a small load and gives you power on indication. I have had some supplies that would run w/o a load on the 5v line, but I have had a few go smokey when I tried w/o it too. I think it is better safe than sorry for $.50 in parts.

Macona, not a bad idea on a bearing in the adapter. I hadn't thought about that. I am not sure if it will be possible to get it concentric this late in the game-- maybe I could use an over sized bore and retain the outer bearing race w/ a trio of machine screws w/ washers-- I have seen this method in commercial equipment.

I don't think the spud shaft is as big as you guys seem to think-- it add's 1.25" to a shaft that was just over 1" long to begin with. Do you really think that is excessive? The stepper motor shafts are at least this long, only .625" diameter and likely much smaller bearings in the motors...

Anyhow, the shaft has to be this length to allow the driver pulley to line up w/ the driven pulleys. I am using all the factory mounts and this is where it has to be to get alignment.

Somehow I had it in my head that the rev's per inch should work out to some sort of whole number. Maybe it has more of an impact on steppers-- w/ the old setup, one step was .0001". Anyhow, I ordered parts today to replace the Z pulley. I'm going w/ a 14T H series pulley there. I ordered new bushings for the X and Y axis pulleys. I'll be able to retain the same pulleys that are there and increase the bore to 1". Those will be some tricky stubs, but I think they will be tough enough.

Thanks for all the input and advise!

Later,
Jason

[Jim Blackwood]

I picked up a pair of 10 ohm 10 watt power resistors at radio shack for about 2 bucks, removed most of the unneeded wires from the board and soldered the resistor in place so I'm about ready to test. Have to bring in a cord from the lab and maybe put LED's on the outputs and I should be ready for the smoke test. I think I'll run by RS with the DMC-2160. They carry some connectors that look a lot like the one installed for a power connection and I already have the tools for those. Get that all squared away and I'll be ready to fire up the controller.

Jim

[MaxHeadRoom]

Incidentally there is a PDF on the Galil site regarding an interface for resolvers.
http://www.galilmc.com/support/appnotes/miscellaneous/note5415.pdf
Max.

[macona]

If you do look into those converters you will be wanting to be sitting when you get the price.

I got a couple of these ICs. I had planned on building my own converter.

http://www.analog.com/static/imported-files/data_sheets/AD2S1200.pdf

[partssisparts]

heres a Resolver to Quadrature Converter

http://pico-systems.com/resolver.html
mike

[macona]

I looked into theirs when I was doing my conversion. Decided it was just cheaper to replace the resolvers with encoders.

They do have a serial to quadrature converter board that looks interesting. I have a 3.5kw Mitsubishi servo motor that I have yet to find a drive for and never have gotten around to changing the encoder. Waiting to hear back from these guys if the board will work for mitsubishi.

[Jim Blackwood]

$150 each it said. How much are the encoders and where would be a good place to go for them?

Jim

[macona]

About $80 each. To me its kind of a wash. $210 more for the cards but I dont have to tear things apart.

[Jim Blackwood]

I see, makes sense. Guess I should tear one down and get the shaft size and start shopping. Might work around those expensive connectors in the process.

I got the controller powered up but haven't established communication with it yet. I have Galil Tools (lite) installed on the PC (the Vaio, not the laptop) and am connected via serial cable and have the connections window up with entries for Com3 19200, Com1 115200, Com3 115200 and Com1 19200 and get a timeout error if I highlight them and click "connect" and that's as far as I've gotten with it so far. Guess I need to go back and look at the manual again, terminal communications were never an easy thing for me.

Jim

[Jim Blackwood]

When connecting a PC using a serial cable, do you have to use the main serial port or can you use the aux serial port? If you have to use the main port I may not have the correct cable.

Jim

[MaxHeadRoom]

The PC's I have that have 2 serial ports, are identical RS232 standard, it is just the COM address that is different.
When working with RS232 its handy to have a signal break-out box, you get a better feel of what is happening.
Correction: see post #148.
Max.

[macona]

You also might need a null modem cable or adapter for the cable you have.. It just swaps the receive and transmit pins. I need one on my Aerotec drive but not on the Allen Bradleys.

Timeout means it is sending a signal and not getting a response.

[jacampb2]

I thought that that was a ethernet based drive. I know the ethernet drives have RS232 as well, but do you have to use it for setup? From what I remember you will not be able to use it via RS232 w/ mach.

[jacampb2]

I made a bit more progress over the past few days. Not feeling so well, we had a cold snap here and I of course started to come down w/ a spring cold...

I got the shaft adapters all finished, keyways cut, and the bolt pattern adapters are all bolted up and good to go. Got the new 14T timing pulley installed on one of the servos, the others I made the shaft adapters step down to 1" shaft so I could buy replacement taper lock bushings for the pulleys that are there. All that went well, then today I pulled the cover off the Y axis motor and found that it's timing pulley is fixed bore. I think I am just going to order another 14T pulley and use those for X and Y and the Z will get one of the left over 11T pulleys. I really can't believe this thing had three different types of timing pulleys-- the Z is a cast aluminum job that is garbage and nearly destroyed, the X has a taper lock bush and is steel or cast iron, the y is fixed bore and steel or cast iron. Must be someone had their fingers in the pie before I got it...

Here are the pictures:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters006.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_nema_adapters007.jpg

Later,
Jason

[MaxHeadRoom]

When connecting a PC using a serial cable, do you have to use the main serial port or can you use the aux serial port? If you have to use the main port I may not have the correct cable.

Jim

I mistook your meaning, the main port should be used on the Galil unit and I think you will find that the 9 pin cable from Galil to PC has to be pin for pin.
All the other stand alone units I have used uses this configuration.
There is a diagram in the manual that shows the pin outs.
The aux port reverses the input/output direction so the cables are different.
I prefer the DOS based DMCTERM for testing, if you have a PC with Win98 you can run it.
Max.

[Jim Blackwood]

I just achieved communication with the DMC 2160!!:) :) :)

Using Galil tools lite I found the right combination when I plugged it in using an ethernet crossover cable, and then found the controller under the "No IP Address" tab. Clicked the assign button and there it was. It looks like I have some sort of program it will execute, and a fair amount of reading to do to see what it all means. It's a wonderful thing when technology does what it should.

Jim

[jacampb2]

Finally, I am off work for a while and am back at this. I am starting on the DC PS, that's the last thing on the list before setting up everything on the bench and beginning the rough tuning...

What I did today was complete my rack for the DC filter caps and made buss bars, ect... This would have been done a week ago, but last week the motor on my lathe died a sudden death. I had to make the 3/8" coupling nuts because I needed some custom sizes to make this work. I was drilling the hex stock and the lathe went boom. A few hundred bucks later, and it is back up and running well enough for me to finish the nuts and get this thing together.

Material for the rack is .125" and .25" acrylic. I chose acrylic because, #1 I had enough scraps to do this piece and not have to cut into any virgin stock, #2 because acrylic has a reasonable dielectric strength and will be a good insulator for the ~360Vdc buss, and #3, since it is clear, I can easily do a visual inspection to be sure the caps aren't beginning to bulge or leak.

The caps are 400Vdc, 450Vdc surge, 4300 uF each. The buss bars are cut from .75x.1875 copper stock. This should be more than enough capacitance to filter supply ripple.

I think I am going to run the DC supply from single phase 220 instead of the 3 phase 220, just so I don't have to start the rotary converter when I am using the 110V high speed spindle to do engraving. It means running another 220 line in the shop, but it should save money in the long run not having to have the converter running whenever the mill is.

Here are the pictures:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/DC_supply_cap_rack001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/DC_supply_cap_rack002.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/DC_supply_cap_rack003.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/DC_supply_cap_rack004.jpg

Later,
Jason

[macona]

Well, it is certainly shiny! Massively overkill, but shiny! You could have used circuit board thickness copper strap the same width and still had anough current carrying capacity. Those straps are what I would see in a BIG mig welder!

[Jim Shaper]

Those straps are what I would see in a BIG mig welder!

Just making sure the drive isn't the only thing that blows when his wiring shorts. :D :p

Does look real nice though.

[jacampb2]

Well, it is certainly shiny! Massively overkill, but shiny! You could have used circuit board thickness copper strap the same width and still had anough current carrying capacity. Those straps are what I would see in a BIG mig welder!

LOL, the copper flat stock was left over from that big H-bridge I built to run my off road recovery winch a few years back. I figured that there was nothing else I was likely to need it for, and this used up all but about 9 inches of it.

I hopefully will be back at this around noon today, still trying to scrounge the stuff up to build a PS on the bench...

Later,
Jason

[jacampb2]

**Cross posted w/ my PS thread in the general section**

Here is a schematic of the updated PS. I *think* this should work, give me your thoughts. I still don't have any idea how to size the pre-charge caps. I am not finding much info online...

Here is the schematic:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/servo_PS_new.png

Thanks,
Jason

[macona]

Dosnt make a whole lot of sense to me. Way too complicated. Precharge caps?

Main Switch to bridge rectifier. From bridge rectifier to cap bank there is a N.O. single pole contactor on one of the poles, dosnt matter positive or negative. In Parallel with the contactor is a power resistor. A on-delay timer closes the relay after a couple seconds or so to close the contactor. Thats it. You can install a bleeder resistor across the cap bank to bleed it. Also a LED with an appropriate resistor for the voltage is a nice addition, it is a visual indication of bus voltage. Most power supplies have them. At 400v buss you will need a 22k ohm resistor at 10w. This will even act as your bleed.

You really dont need all the relays you could leave the main one controlled by the enable outputs on the drives or estop.

[jacampb2]

Well here is how it should work, the 110V control voltage is fed through the NC aux contacts on the main contactor and causes the pre-charge contactor to pull in supplying 220V through the in-rush limiting capacitors. The relay coil across the DC buss has a series resistance of 220K which should cause it to close at ~220Vdc. This relay feeds the main contactor control which then closes, the aux. contact go open and the pre-charge contactor drops out. If the voltage drops bellow ~180 Vdc on the buss, the cycle repeats...

Do you think it is not going to work, or that it is unnecessary. The thing is, I have all the components I need to do it this way, and I like the idea that it will always engage the pre-charge automatically. If I went resistive on the DC side I need a contactor that can handle that DC voltage/current and also a timing device of some sort, which from my search it is pretty expensive.

I have absolutely everything except maybe some wiring terminations in order to do it the way I drew it up. Also, this is the way that Jerry recommended, and he has helped me with a lot in the past and has never steered me wrong.

Thanks,
Jason

[macona]

It might work but the thing is contactors dont pull in at a specific voltage. It may pull in significantly lower. Also the contactor will hold closed probably down to 1/4 of its rated voltage.

It just I have seen a lot of precharge circuits and none as complicated as yours.

Here are some cheap 220v on-delay timers on ebay:

http://cgi.ebay.com/OMRON-TIMER-DELAY-RELAY-SENSOR-H3BA-5A-200-220-240V-/110541549818?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item19bcca30fa#ht_5310wt_913

http://cgi.ebay.com/Pair-2-OMRON-Industrial-On-Delay-Timers-H3BA-/160443054977?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item255b26bb81#ht_500wt_1154

[jacampb2]

nowhere are the contactors pulled in w/ less than their control voltage of 110Vac, the one that connects the mains does so when a small relays coil comes up to about ~70% of the buss voltage, the relay closes and connects 110Vac to the mains contactor...

[jacampb2]

Well, I made some progress today. I got a long ways toward building a supply on my bench for testing the supply design, and if that goes well, then testing the servo amps and beginning tuning. All I have left is wiring up the 110V control side of the system, probably 30 more minutes of wiring. My wife asked me to come in for the afternoon, so I am on hold 'till tomorrow...

Here is a picture of the mess/PS on my bench, also a pic of the BOB w/ it's labels on now, and my attempt at DIY shielding on the IDC cables. Seems like it should work-- better than nothing anyhow.

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/bench_ps001.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/bench_ps002.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/bench_ps003.jpg

Later,
Jason

[jacampb2]

**Cross post again, sorry**

I gave the supply a test run on the bench today. The good news is, the AC caps work great to limit inrush, the buss takes approximately 10 seconds to come up in voltage. The bad news is, the relay that is supposed to sense the buss voltage did not pull in when it should have. In fact it never pulled in at all, and the buss made it up to 300Vdc through the caps.

So, it is back to the drawing board to find something reliable to engage the mains contactor, I am first going to try dropping the coil resistance of that relay and experimenting a bit, but if that doesn't provide reliable results then I'll have to start working on some other method.

Thought you all might want to know I lived through the initial testing...

Thanks!
Jason

[Jim Blackwood]

Looks like good progress Jason.

I'm curious why all the concern about inrush current. Is that really a big problem? I mean, I can see there is nothing to regulate the supply or limit the voltage but if that's the problem wouldn't a zener diode do the trick?

I've not gotten any farther on mine and may not for another month or so since I'm paying off some things including my controller. But it's just a temporary hold. I've noticed I have a very slight bit of backlash in one of my ballscrews and I'm wondering if a linear glass scale might be a good feedback device. Is it easy to add those to the system?

Jim

[MaxHeadRoom]

Generally using a scale to eliminate backlash requires dual feedback, there is an excellent video on the Galil site that explains the dual PID loop configuration for this, using just the scale can cause hunting of the motor if backlash exists, not just for Galil systems.
http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods
Max.

[macona]

Inrush can trip breakers, blow fuses and is hell on contacts. Very few large power supplies do not have current inrush limiting in one form or another.

My machine had Heidenhain scales on all three axis when I got it. It was using them for the only feedback. It was rather sluggish because of it since the gain had to be really low not to oscillate too much.

I tried it again after I retrofitted it and it did not work well. Ended up installing encoders on the motors.

There are several controllers out there that will support dual loop feedback. But even most commercial machines have only the encoders on the motors. Backlash is calculated and entered in the control and it takes it from there. Backlash comp in mach works just fine with servos when you can turn it up to 100%.

[MaxHeadRoom]

Personally I very rarely see any kind of inrush circuit on all the machines I have come in contact with over the years.
Of all the power supplies I have built over the years I have never used one and never seen a problem, to some extent a linear supply with a transformer offers some inrush protection, if the current is excessive the transformer saturates anyway and limits the secondary current.
This feature is purposely used by manufacturers of voltage regulators such as Sola who saturate the transformer on purpose, guaranteed it runs constantly hot, but these have been in use for years.
I think there is too much hype attached to inrush protection.
Just my 2¢.
Max.

[macona]

With transformer based supplies there is usually no need for it. But with switcher based supplies where the line is directly feeding electrolytics there is always some sort of inrush limiter. Simplest form is a large PTC thermistor that looks like a large capacitor in series with the buss caps.

[MaxHeadRoom]

I have often cannibalized the power supply parts off of blown VFD's, some very high H.P., they all appear to have a 3 phase bridge and large Cap. bank connected directly across the 3ph or 1 ph supply.;)
Max.

[jacampb2]

Max, for kicks, today I tried to energize the main contactor w/ no inrush limiting in place. The PS immediately blew my 70A breaker on that 220 line. Keep in mind that the parts I am using in this came out of a 30Hp VFD, all the wet caps were in that drive, in parallel just as I have them. That drive ran on the same circuit as this PS is currently powered by. The drive never once tripped the breaker. I'm going to have to side with Macona here-- there is some form of inrush limiting, whether you recognize it or not, it is there.

Consider this, the instantaneous current draw upon closing the contactor is limited almost solely by the series resistance of the caps. Most general use caps are under 300 mOhms, if we guess that that is the ESR on my caps, then the 6 in parallel yield a ESR of ~50 mOhms. Solve it for current w/ ohms law, it will tell you the max instantaneous current draw of the PS when energized. 330V/.050= 6600Amps. I know this is not a real world evaluation of inrush current, it is far to simple, but it IS a real world representation of what the inrush is for the first microsecond after that contactor closes. If you don't constantly trip breakers, then you are constanly pushing your components far past what they are designed to take. In rush is a real thing, whether you believe in it or not.

For the record, I wouldn't be monkeying with any inrush limiter either if this were a lower voltage supply w/ a lot less capacitance. My current supply in the Mill, which drives the steppers, is 80Vdc and about 10,000 uF of capacitance. I just flip the switch on, it is transformer isolated, which inherently limits in-rush to some extent, but I have never worried about it. I did cook my bridge rectifier for that supply one time on start up, probably repeated starts w/ little limiting finally did it in.

Anyhow, the update for today is that I got the supply working. The voltage sensing relay resistance still needs some fine tuning, it pulled in at 120Vdc w/ a 10K coil resistance. This is about 120V lower than I wanted it to, but I'll get it fine tuned. At 120Vdc when it engages the main contactor, it has limited enough that the supply doesn't blow the breaker, but it still does dim all the lights in the shop. I think if I get the main to pull in around 200-240Vdc, then it will be about perfect.

I'll work some more on it tomorrow and let you all know what I find.

Thanks,
Jason

[jacampb2]

Well, I have it up and working how I want it. The mains contactor energizes at ~215Vdc. The magic number for the relay coil resistance ended up being ~21.5K.

I needed a power resistor for the additional relay coil resistance. If it had worked out at 10K, the resistor would have needed to dissipate ~9 watts. I didn't want to buy a power resistor and then find it was wrong, so I decided to make a resistor array that would be capable of handling the power, and would be fairly easily modifiable if I needed to change values. I used a little radio shack piece of perf board and originally I started with 47 470K 1/4W resistors. This gave me 10K w/ an approximate power rating of 12W, the layout would not survive that power level for long, but it was good enough for testing. When 10K didn't work out as desired, I dropped a 10K 1/4W resistor in series, I knew it wouldn't last long trying to dissipate 2.5 watts, but I needed an idea where I needed to go w/ the power resistor. At 20K the resistor pulled in at 200Vdc, so I removed the 10K 1/4W resistor, and started pulling 470K resistors out of the array. I left 22 470K resistors in parallel, this gave me right around 21.5K with a power handling of 5.5W or so. With the resistance being higher, the power handling was still in range. I tried it out, and it worked great, so now I have my number for the power resistor I need to order.

Here is a picture of the DIY power resistor, and bellow that is a short video of the PS coming up and running.

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/DIY_Power_resistor.jpg

Servo DC supply

http://www.facebook.com/v/132170750135186

Thanks,
Jason

[Jim Blackwood]

Well that's pretty cool.

These are things that I will keep in mind as I go along. Obviously until I have acquired some servo drives it doesn't apply very much, but in time it will. Thanks for posting such a detailed account.

Jim

[jacampb2]

Hey guys, sorry for the long delay on updates. It has been a rough summer, I have been having trouble with my diabetes, had a piece of crap car I had to work on every free moment and all the other typical family stuff with two little girls and another kid on the way...

Anyhow, I'm doing better and have had a little free time to work on the conversion again. Picking up where I left off, I have the dc supply working how I want it, I ordered proper power resistors and a couple of big single phase bridge rectifiers. The bridges are rated for 80 amps each, I bought a total of four, so if two in parallel is not enough, I have another pair to squeeze in. I fully expect a capability of 160 amps to be enough.

I decided to put the whole thing in an old pc case, I had this old packard bell case, circa 1995 or so, the pc was so far outdated I didn't mind scrapping it... The case is built much better than anything now, when it is all together pretty much every side has steel panels covering it under the plastic, it should offer some reasonable emi shielding and has the benefit of containing the entire 300 Vdc supply and the 24 Vdc supply in one enclosure. The capacitor bank is still external, but that is well protected.

It took a while to get it all squeezed in here and working. The 120 mm fan its only temporarily mounted, but the thing moves a ton of air. It is probably not necessary at all, the heat sink area is already way overkill, but I like overkill! :D

I robbed a din rail barrier strip from a scrapped vfd, that provides 220 Vac entry and dc out. It accepts up to 0 awg, so it is plenty beefy. There are just common everyday screw type barrier strips inside for control voltage and the small dc supply output.

I still need to get a hold of a decent 18 volt or so secondary transformer. The one I have in there was my third one out of the recycle pile and the secondary voltage is a bit high, rectified, even with a load, it does about 36 Vdc. I'm probably going to have to break down and buy one, the first two I tried were shelled, luckily I don't require a big one, just enough to run some fans and release the brakes on the servos.

Anyhow, that is my progress report for today, hopefully I'll get things moving along at a decent pace again. Here are some pictures for your viewing pleasure.

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/ps_in_case_1.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/ps_in_case_2.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/ps_in_case_3.jpg

Later,
Jason

[MaxHeadRoom]

I still need to get a hold of a decent 18 volt or so secondary transformer. The one I have in there was my third one out of the recycle pile and the secondary voltage is a bit high, rectified, even with a load, it does about 36 Vdc. Jason

Have you considered taking a few secondary turns off?
Probably work out to 2.5T-3T/volt.
Max.

[jacampb2]

No, not really, I have never tried to modify a transformer, and being that it is a EI type and shielded, it really seems more trouble than it would be worth. I cant believe a new 18V 40Va or so transformer is going to be much money, especially if I try to find something surplus.

I will consider it though, I just hate to botch a good transformer if it doesn't come apart easily enough.

Thanks,
Jason

[MaxHeadRoom]

I realize EI type are a bit trickier, but if you are careful it usually can be done, the shielding is usually between the primary and secondary and the secondary is wound on last.
With some preliminary investigation it should show if it is viable or not.
If you have room to wind on a small secondary of say, 10 turns, and measure the voltage, this will give you the turns/volt required to modify.
Max.

[jacampb2]

Well, I got a little bit more accomplished today before life intervened and forced me to stop early. Basically all I got to was making an aluminum mounting panel for the fan, rigging up a filter, its in place and I started putting the case back together before I **** down for the day.

I also found and bought a new transformer on ebay. It is a dual secondary model with a 120 v winding which will out nicely, I won't have to run a separate 120 v line to the machine.

Picture of today's work is below:

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/ps_case_4.jpg

Later,
Jason

[jacampb2]

I got a little more done... The case is all buttoned back up, there is now a barrier strip on the back of the case that has input for the control voltage, 110 in for the small ps, I shouldn't need that once I get that new transformer, but just in case, you know? There is also 3 locations on the strip for 24 vdc out.

All in all I'm happy with it. I have a couple of short videos of it, one of it coming up, dang! Is that fan loud. A second video of it shutting down and bleeding off the buss, the yellow led on the case stays lit until the buss is down to safe voltages.

*edit* I'm having technical difficulties, I can't get my pictures uploaded and the two videos are giving me trouble as well, I'll re edit and sort it out in a bit.

Pictures from the work today, the first just shows the metal lining to all the panels, this thing was seriously heavy even before all the crap was inside...

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/ps_in_case_5.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/servo_conversion/ps_in_case_6.jpg

Try again for video:

http://www.youtube.com/v/JMU65NQuiFs&sns=em

http://www.youtube.com/v/hTw_f-T68vg

More later,
Jason