PDA

View Full Version : Electronic QC gearbox and Kermit's offspring



nheng
10-22-2003, 08:35 PM
A while back we had a thread going on electronic quick change gearboxes (a.k.a. electronic leadscrews). It was one of those projects which I may not get to for a while so I sprung for a Frog cheap CNC which someone had mentioned. After tweaking the lathe a bit to reduce friction and tweaking the Frog to happily drive their largest stepper, I started cutting threads. At the slowest spindle speed of my maximat 8 (65 rpm), I cut a 10 TPI thread and could probably do 8 TPI. The upper end is unlimited and the thing will do English or Metric at the touch of a button. This image shows a 1mm thread on the left, 20 TPI thread in the center and a large diameter 20 TPI on the right, all cut just for fun. I love threading now !

I may build a better unit at some point with my own controller and gecko drive but for the near term, this thing does a pretty decent job. The heatsinking and power supply need a little more tweaking and that's about it. Pitch is accurate and thread form is whatever mess I make of it. These were cut with a sharp V carbide insert but the bottoms of the threads are not sharp V so there must be some "jitter" in the timing off the spindle.

http://home.comcast.net/~nheng.gnu/img_qaz_xsw_stf/froggy_threads.jpg

< added > One other thing worth mentioning is that the "microphonic" patterns seen with some gearing and single phase motors is pretty much gone when I use the frog for longitudinal feed. The feed is no longer related to the spindle rpm, gearing or motor "jogging" (is that the right term ??). Note the extremely smooth finish on the first large diameter below the thread in the center image.

Den


[This message has been edited by nheng (edited 03-09-2004).]

Doc Nickel
10-22-2003, 10:06 PM
How does the Frog controller encode the spindle rotation/position?

Doc.

nheng
10-22-2003, 10:17 PM
It has a hall effect sensor and a very strong but tiny (about 0.1" OD) magnet mounted on the spindle. I was skeptical of this approach and was thinking of multi-thousand line encoders and the like but the system works pretty darned good. The frog's software does a good job of resyncing on each pass (but needs a little space, say 0.1",to sync before the cut). Once set up, you just hit cut and the tool cuts to the end of the length you keyed in. You then retract the tool, press go and it returns to the start. When you press cut again, it resyncs and continues the next pass, and so on.

I suppose that if you're in an area where the power line (and motor speed) go up and down frequently, this system might have some hiccups but even then it will probably clean up on subsequent passes.

The ability to cut metric or english at the touch of a button and longitudinal feed with the same controls is pretty cool.

On small lathes with no leadscrew, you could even cut threads by driving the feed wheel used for longitudinal feed.

Den

Oso
10-23-2003, 12:05 AM
Does it seem to "know" in time if the spindle slows in a cut?

If there is only one magnet, the detection delay could be a couple turns, and the threads might be "drunken" if the spindle slowed.

nheng
10-23-2003, 01:18 AM
Oso: I suspect that it may update on each rotation, using some sort of time period measurement instead of rpm. It then could interpolate the required gearing ratio to determine what it feeds the stepper, losing or gaining a step as it maintains sync. This may be where the small flat at the bottom of the thread comes from after multiple passes.

Cutting a coarse pitch in SS might be a good test of the speed variation question. I'll have to try it this weekend.
Den

Evan
10-23-2003, 01:38 AM
I checked it out. It looks cool, nice work. Doesn't look like it would power a SB9 without some fairly major fiddling on my part. I'd have to use a Gecko or something and a bigger stepper. It is just about exactly what I had suggested way back in the other thread. I dunno, maybe I just CNC the whole thing myself. Maybe I should build a mill first. Just WTF am I going to do with CNC anyway? I already spend too much time making computer models of stuff.

John Stevenson
10-23-2003, 03:36 AM
Contary to popular thought a single pulse encoder will work.
A lot of the small educational machines that were common a few years ago like Ortec, Boxford, Emco and the Myford / Connect all use single line encoders, usually just a hall effect switch.
Art Fenarty who's writing Mach2 has got round to the lathe part and although still in Beta it doing very good threads at up to 900 rpm, again with a simple one line encoder.

The Frog whilst being well suited to small lathes is a little too small for the sizes above the micro machines.

For anyone contemplating this it would be cost effective to hang fire until Mach2 is released from beta which could be about 2 months and for a $150 licence fee you can get a full 6 axis mill or lathe controller.

John S.

Doc Nickel
10-23-2003, 05:39 AM
Being able to do metrics at the push of a button would be marvelous. I'm fighting with odd-tooth-count approximation gears and not being able to open the halfnut to thread, and it's rather a pain.

I have a spare lathe, a 9" Logan, that sees minimal use. I take it from the above that this might be too large for the Frog? Why? I thought the Frog was just a controller, not a controller and motor...

I have been toying with doing a full two-axis CNC conversion on it, but I don't even know where to begin.

Doc.

Oso
10-23-2003, 09:12 AM
John.. of course it "will work", but it can't possibly know if the spindle slows down until at least one rotation, maybe two, depending on the timing algorithm.

I think it will make drunken threads, and that the "flat" at the bottom of the thread is due to a timing error, and may have created a threadform error as well. That sounds like it would be an error in the form of the thread, too narrow.

I doubt it is making a standard flat on the bottom if it doesn't on top. How would it know?

Forrest Addy
10-23-2003, 09:31 AM
Hmm. If the gadget makes threads and depends on a single rev pulse for pitch timing it presumes constant spindle angular velocity throughout the spindle revolution. That presume a lot.

Seems to me that if the Frog was going to make good threads it would have to time from the Rev pulse and integrate from the bull gear tooth count to update lead screw angular velocity 80 times per rev or more if "drunken" or staggered lead threas are to be avoided.

I'm not arguing with success just posing things to think about.

nheng
10-23-2003, 05:53 PM
Forest, Oso: I agree fully ... once I cut some tougher metal, the spindle speed is going to drop as the cut is started. Any speed variation is not going to be nice. Not as much of a problem on smaller threads though, especially at my 65 rpm spindle speed.

John S: At this point, it's driving the Emco which is an 8.6 x 36 gearhead. I should mention that the Frog stepping limits only permit traversing a small portion of this length.

I used the largest stepper offered with the Frog, just under 170 oz-in. The stepper seems adequate to drive the Emco quite well although it is a bit beyond the Frog's ability to drive without a little tweaking.

I'll post results after cutting some stainless at 10 or 20 TPI. The thread form is something to look at also. That tiny flat will have to be compensated for somehow.

Den

John Foster
10-23-2003, 06:08 PM
Den, what size stepper are you using? How deep a cut does it take on each pass? John

nheng
10-23-2003, 08:29 PM
John: The stepper is 170 oz-in, NEMA size 34. For thread cutting, I've been using (in thousandths) roughly a 1-3-3...-3-2-1 sequence with multiple, no feed passes at the end. I think I used the same feeds on all but the finest threads so far and they were in 6061 aluminum.

I've been able to feed longitudinally up to a 0.040" depth of cut (0.080" diameter) and a feed of (estimated) 0.004 IPR. Not too sure about the feed settings on the frog since I haven't played with it long enough. This cut was also in 6061 aluminum and at 700 rpm spindle and (I think) on 1/2" stock.

All: The tiny flat at the thread bottom may have been from my half-nut lever slowly heading toward the released position after each back and forth motion. I've gotta tighten it up and test the whole process again.

Doc: You can get a Frog which will drive an external driver like a Gecko but at that point, together with a much larger stepper, you might be pushing the $700 range (one axis). As an only slightly educated guess, I would think that a lathe like a 9" Logan or SB might want 600-900 oz-in or even more. It would be good to hear other's thoughts on this.

Den

[This message has been edited by nheng (edited 10-23-2003).]

nheng
10-24-2003, 05:34 PM
http://home.comcast.net/~nheng.gnu/img_qaz_xsw_stf/frog_303_ss_M12_1.0.jpg

Tried some 303 SS to see how any spindle slowdown would affect the thread. Here's an M12 x 1.0 thread. It looks quite good, even under a microscope, except for a very slight thickening of the first thread. I'm new to threadcutting and without measuring what I see, I don't have a good feel for the impact of this amount of "distortion".

Except for this, the thread was clean and the pitch was spot on.

I still have to agree fully with Forrest regarding the assumption of constant angular velocity. It assumes much.

Evan: Stop beating yourself up and go make some chips ! The only reason I'm looking for an "e-qc" gearbox is because I'm basically too lazy to change gears. I still think gears are good unless you get a very high res encoder on the spindle to match or exceed gear-like performance.

Den


[This message has been edited by nheng (edited 03-09-2004).]

Evan
10-24-2003, 05:56 PM
That looks pretty nice Den. Cool little gadget and decent price too.

I am making chips http://bbs.homeshopmachinist.net//biggrin.gif Won't say what until I see how it turns out... Sure is nice to have the power crossfeed on my SB9.

John Stevenson
10-24-2003, 05:58 PM
I asked a guy today whos running Mach2 on lathe with a single spindle encoder. He says that he's not seen any variaion of pitch on any of the threads he cuts.
Most of the time he's running at 500 rpm with a 1HP motor and feeds of about 10 to 15 thou per cut.
At these speeds and feeds you shouldn't get any discernable slowing down. Remember when cutting a thread on CNC you always have to have a ramp in move to allow the two movements to get into sync before the start of the thread.

John S.

nheng
10-24-2003, 09:10 PM
John S: Threading at 500 rpm is pretty impressive with this type of approach. On my lathe, the leadscrew is a relatively fine 16 TPI and the stepper is a fairly low end unit at this point. I need to run the spindle at 65 rpm to get down around 8 - 10 TPI or the stepper can't keep up.

I'd like to learn more about the system you've described. Does it use a PC driven controller? It seems to me that you would need a substantial driver and stepper also in order to get coarse threads.

Evan: that SB power crossfeed is pretty slick ... had any thoughts about writing it up?

Den

Evan
10-25-2003, 01:43 AM
Den,

I would write it up but I have learned that publishers don't publish articles if the item requires hard to find parts or is in some other way beyond the capabilities of the average reader. I have had a number of articles published in some magazines and more turned down. If I had made the power crossfeed device without a magnetic clutch and used some sort of mechanical coupling/decoupling instead it might be a go. I LIKE the mag clutch. It works really well. I was facing a piece of 4" Al bar stock with it tonight and the finish looked so good I gave it a 15 second polish with Brasso. Holy dog dung Batman, I could shave looking in it. Not gonna though. I'm going to have to rough it up so it will hold some oil.