Servo motion WITH inexpensive electronics?

[Doc Nickel]

The other thread reminded me I wanted to ask this.

I have a manual surface grinder, and I'm looking to 'automate' it. Please don't say "sell it and buy an automatic one", because it took me three years of looking to find this one.

Anyway, the automation should be fairly simple: a servo or stepper to drive the table back and forth, and second one for the cross feed.

The table recip needs to be able to reverse fairly smoothly, but doesn't necessarily need to be too precise. The cross feed need to be able to move in very small increments with reasonable accuracy.

At the moment, I'm thinking a set of toothed belts running from the motors to the handwheel shafts. How much drag would an unpowered stepper or servo have? If I wanted to switch to manual, would I have to remove the belts?

Both axis would need either limit switches (probably easiest and most user-friendly) or to have some sort of programmable limits- Obviously I don't need to cycle the table the full 18" stroke to grind a 6" parallel. And I'd like to be able to let it run hands-off, so a cross-feed start and stop limit would also be handy.

I'm starting from scratch here- I have no controllers, no actuators, no nothing. Which would be better, steppers or servos? Can you recommend a size and power? (Forces are light on a grinder- nowhere near that of even a small milling machine. On the other hand, the table's kind of heavy, and needs to reverse direction constantly.)

What sort of controller could do this? I'm hoping I won't need an actual PC, but arrangements could be made if necessary. Do I need a 'drive' for the servo and then a controller to run the drives?

The recip could almost be done with nothing more than analog switches and fancy wiring, but I'd like to have it a little smarter than that, and to run the cross-feed in sync.

Doc.

[Dawai]

Actually I wish you were close by.. I got three servos, drives, power supply, back panel all wired up. Been waiting on the right rig.

I'd like to find a cylindrical grinder to mount it all on. there was one on ebay with a powered center.

We could have a relationship, I need parts too.

My shop is so full I have to open the door to poot.

[Evan]

How much drag would an unpowered stepper or servo have?

Enough to make it a real pain to try and use. That is, unless you use a particular type of stepper called a variable reluctance stepper. They have no magnets and zero resistance to turning when the power is removed.

With any run of the mill (grinder) type motor you will see a considerable amount of cogging or resistance to turning in the power off condition. I use a magnetic clutch on the cross feed drive on my lathe. A flick of a switch engages it and releases it for zero resistance to hand powered use. They are pretty common devices and easy to use. Give it 24 volts and it transmits torque. Turn it off and it declutches completely.



The black object is the electromagnetic clutch. Gear is input and central shaft on the other side is output. Other types are straight coaxial. Give me a few minutes and I will rummage around for some coaxial units you can have.

[Evan]

Well, I found the clutches. They are from an all electromechanical 4 speed gearbox from the 70's, before easy control via servos or steppers. I haven't seen these in years and unfortunately I don't think they will handle the load. They are a bit smaller than I remembered them. At any rate, this is the sort of thing you need to disconnect the drives instantly.

[Doc Nickel]

Good point on the cogging. Even the little ones I have (ex-copier leftovers) cog noticibly when turned by hand. And that's a definite issue for a surface grinder.

Point in fact; how smoothly does a stepper rotate in actual use? Isn't there a small degree of cogging even when running normally? Would I be better off with a servo, at least on the recip axis?

Actually I wish you were close by.. I got three servos, drives, power supply, back panel all wired up. Been waiting on the right rig.

-Thanks for the thought, Dawai, but it sounds more like you need a small vertical mill for that setup. I'm kind of hoping I don't need a full-blown CNC (as in, requiring a PC) but that I can get away with a relatively simple, if not off-the-shelf, controller.

Let's see...

Two axis. The X axis just reciprocates, no particularly precise positioning. Needs a stroke limiter (switches or software, so the stroke length can be altered or adjusted) and a stroke speed setting (inches per minute.)

The Y axis needs to just step the cross feed, fairly precisely, from about .010" to maybe as much as 1.000" inclusive. I know a stepper can be geared down for better resolution.

The Y axis needs to sync with the X, so it's stroke-stepover-stroke-stepover-etc. The option of stroke-stroke-step-stroke-stroke-step would be handy, and/or stroke-return-step-stroke-return-step (with it grinding both ways.)

Doc.

[Black_Moons]

Steppers can be rather smooth if you do whats known as 'microsteping' where you basicly drive it with a sine wave, or usally, pwm it with a duty cycle that follows a sine wave.

[Evan]

Once a stepper is turning a few hundred steps per second there is no actual "stepping" taking place. Steppers are also used and sold as synchronous motors with almost identical construction. A while back I posted on how to connect a stepper to straight low voltage AC, six to twelve volts, with only a capacitor to provide phase shift to half the coils.

Cogging in operation is no concern at all though. You can use a one chip driver and microstep the motor. It is then running on a synthesized sine wave and is acting exactly like a synchronous ac motor. To control the speed a simple 555 timer circuit to the step input of the driver will do just fine.

[macona]

If a servo is in freewheel mode there is only about as much resistance as the bearings put on it.

I have thought about doing the same to my surface grinder. Mine was originally outfitted for power feed but most of the mechanism is long gone. The clutch remains though. I was thinking a servo motor ought to work just fine instead. I have a bunch of smaller 200 w brushless units. Couple limit switches. All of this could be controlled by a simple PLC. Or go for it and set it up for full cnc with mach!

But in reality for the power feed all you would need is a small three phase gear motor and a VFD. Use an electromagnetic clutch to disconnect it while not in use.

[Doc Nickel]

If a servo is in freewheel mode there is only about as much resistance as the bearings put on it.

-I assume the typical servo freewheels whenever there's no power being applied?

All of this could be controlled by a simple PLC. Or go for it and set it up for full cnc with mach!

-Cool as that'd be, I'm going for simpler-is-better. I'd prefer something where i just have to hit the "on" switch, then press the "go" button. No PCs, no issues with hardening the PC against shop contamination, etc. This is a pretty straightforward application- like both you and Evan note, it ought to be able to be done with simple timers or controllers.

I could almost do it with just some AC motors, a few cleverly-wired analog switches, and some mechanical stops and dogs. The cross feed'd be pretty crude though, so it's worth doing right.

But in reality for the power feed all you would need is a small three phase gear motor and a VFD.

-Well, for the recip, maybe. But I want to sync it with the cross-feed, and make the thing pretty much hands-off. Grinding a larger piece is pretty time consuming, and I'd really like to see it where i just set it up, start it cycling, and then go do something else for the next fifteen minutes.

("Something else" of course meaning only a few feet away or on the other side of the room- not "run into town for dinner and a movie". )

Use an electromagnetic clutch to disconnect it while not in use.

-Minor drag, or even cogging, won't be an issue on the cross-feed. That only gets turned a fraction of a turn at a time. As long as there's not so much drag it can't easily be turned when cranking away to move the table significantly (as in when first lining up a project, or cranking back to the starting point) I could probably get away with a direct toothed-belt connection.

On the recip axis, I'm kind of thinking that a mechanical disconnect would be fairly easy, and possibly more reliable than a magnetic clutch. For example, the toothed pulley could be free-spinning on the handwheel shaft, and have a toothed face, like on a Bridgeport knee handle.

To engage the drive, a disc with a mating toothed face could be slid up to the pulley, and clamped. No big deal if you had to use an allen wrench to undo it.

Anyway, mechanically it doesn't appear to be any problem. But I have no solid idea how to do it electrically. And since I'll have to buy it all (no scrap bin full of servos, sadly) I'm open to suggestions on the type of motor and how to control it.

Sources and part numbers also welcomed.

Doc.

[macona]

Freewheeling, it depends on the servo drive. Some will dynamically brake the motor when the power is off. On my mitsubishi drives it shorts one winding when the power is off. If the drive is on and not enabled it turns freely. DC Brushed Servos like on my CNC mill turn freely when off, at least on the ones I have encountered.

Here is a link to a recent comparison between DC Brushed servo drive.

http://www.thecubestudio.com/ServoDriveReview.htm

The winner of the comparison, the whale3 is only about 75 euros, Thats $110. Thats cheaper than a gecko. Get a few DC motors off ebay and slap some encoders on them if they dont have them already and you have a full servo system for about the price of a stepper system. Heck of a deal.

The place they sell the drives:

http://www.cncdrive.com

Like I have said before. I built my first couple machines with steppers, once I tried servos I was hooked. Great speed, acceleration, closed loop, almost silent, and smooth running.

I was referring to the traverse with the VFD. That can be all set up with limit switches. Magnetic clutches are very reliable. You would never see one fail in a service like this. One advantage of a clutch like this is it can slip if overloaded.

As for the motors themselves you are just going to have to scrounge around. Just about any permanent magnet motor will work as long as you can somehow attach an encoder to it. The encoder will determine the ultimate resolution of the motor. How many pulses it will take to turn the motor one turn equals 4 times the encoder resolution. There are some encoders by a couple different companies, one being CUI, that allow you to set the encoder anywhere from 48 to 2048 counts per turn.

The logic for this can be really simple. Done with some 555's and flip flops. A 555 could send a pulse train to the traverse motor. The frequency would determine the travel speed. Limit switches would reverse the motor with the direction pin at the end of travel. This could be done through a flip-flop or relay logic. The right limit would also signal the infeed. Send so many pulses to move it so far.

Alternatively this could all be done with one of the Arduino boards which are very popular right now. Think microcontrollers for dummies! I have one of the teensy boards from this guy:

http://www.pjrc.com/teensy/

Cheap and probably way easier to deal with than my other method. Little more studying for you!

[Braindead]

Anyone else here running EMC?

[Doc Nickel]

Good call on the Arduino. I've read some on that over the last few years, and it seems pretty versatile. Still don't have a clue how I might set it up, but I'm told it's pretty straightforward.

Would I need an encoder on the servo, or would it be better to have a position indicator of some sort on the table? Limit switches, maybe a lightweight rack with a Hall sensor, etc?

Doc.

[Weston Bye]

I recently did a control system for a production grinding machine. Though the configuration of the grinder was unique for the application, the controls were essentially for a surface grinder.

The table was driven by a hydraulic pump and cylinder as was the part feeder. Grind feed rate was determined by the hydraulic flow controls. A stepper motor powered the cross feed.

A PLC (programmable logic controller) controlled the table grind feed through a directional valve and provided step and direction sgnals to a Gecko 210A stepper drive. I had to use the slightly more expensive 210A rather than the 201A, as the 210A could be configured for full-step operation where the 201A was limited to X10 microstepping operation. The pulse rate output from the PLC was just too slow for microstepping.

The PLC was about the size of a pack of cigarettes, had relay outputs for driving the valves, and transistor outputs compatible with the step and direction inputs for the Gecko drive. I also used an LCD kepad module for the PLC that was programmed so that the operator could select the stepper feed rate and some other parameters for the various products, or directly enter the parameters for custom or experimental grinding operations.

The system included a VFD for spindle speed, also controlled by the PLC.

Now, the PLC I chose was inexpensive, but programming software was required for both the PLC and the LCD keypad. Even these were relatively inexpensive, but an added cost nonetheless, if you only use it to automate one machine.

Over the years I have built half a dozen of these grinding systems, each with updated controls technology. The first were built with relay logic.

[lazlo]

Good call on the Arduino. I've read some on that over the last few years, and it seems pretty versatile. Still don't have a clue how I might set it up, but I'm told it's pretty straightforward.

The Arduino is an AVR development board with a bootloader applique and light version of C to hide gcc from you. It's a great way to get started for non technical folks, and when you want to program in full C, you can ditch the Arduino software and program to the AVR board directly with the excellent (free) AVR development tools.

[Doc Nickel]

Okay, some good data and lots of options. Almost too many options, especially for someone as unfamiliar with this stuff as I am. Can anyone suggest, for want of a better term, a shopping list? A tried-and-true combination of components?

Doc.