High current bipolar stepper drive

[wsc]

Hi guys, first post here.

Seems I'm in a bit of a pickle. My current project requires the use of bipolar steppers with 2.8 amps per phase. This application demands super smooth, super quiet low speed operation. I've been researching for a few days and have looked at LMD18245, L297, L6506, etc. They each have their own flaws that I believe rules them out for my application.

I think the best solution is to design a discrete controller based on a processor (AVR) and MOSFETs. Ideally, similar to a linistepper for bipolar motors, which isn't quite a direct correlation. Right now I've got a few questions as I'm pouring over documents and racking my brain for ideas.

- Anyone working on such a project at the moment?
- Any ideas for sensing current through each winding in a more elegant fashion than a sense resistor?

I'm open to any comments or suggestions.

[pminmo]

Take a look at Alan's PICstep. The LMD18245's will do 3A. While it's not AVR based, it something you can get up and running in a short time. Construct your own code if you want to different functionality.

[wsc]

Well, heh.. here is what I have in mind.

The 1, 2, 3, 4 and enable lines are all pulse width modulated by the AVR. The filters after the gates convert the digital pulses to an analog voltage. The idea being that the MOSFETs are smoothly driven from saturation to cutoff, instead of just slammed on and off. The only reason I'm considering the filtering is because I've got to reduce noise and get the motor to run as smoothly as possible. I'm hoping I'm not totally off my rocker here...

Please let me know what you think guys!

[wsc]

Phil the only reason I don't want to use LMD18245's is that eventually we will be moving to even larger motors in the future. Might as well just skip right ahead to a design that's more future proof.

Edit: and we have space and cost constraints that rule out something like a geckodrive.

[jerber]

On this site, http://www.sb-microstep.com/ , there is a board that uses two LMD18245's in parallel to get 6 amps.

Maybe a modular design is the way to go in order to get higher amperages (3,6,9...).

What kind of AVR are you using ?

[Garfield2]

Originally Posted by jerber On this site, http://www.sb-microstep.com/ , there is a board that uses two LMD18245's in parallel to get 6 amps.

Err I'm pretty sure from my reading of the LMD18245's spec sheet that parallel LMD18245's on a common coil ain't such a good idea. Their MOSFET based current sensor and limiting circuit will all be screwed up with devices switching the same load in parallel. In fact I remember the datasheet saying that you can't do this (unlike the L298 which can happily do this as pminmo has skillfully shown).

Designing a high-powered microstepping bi-polar driver that can do > 3 amps is possible, but you get to a point where you are spending more time designing and using expensive exotic parts on something you're not 100% sure will work, if it doesn't just blow all those said nice expensive parts up in smoke. You'd almost be able to justify an off the shelf driver from a reputable manufacturer with a proven track record for less.

Building a MOSFET based bi-polar microstepper shouldn't be to hard, but you need some form of current sensing system, which for my money I hate using a big fat nasty resistor. Plus at the current ratings you're after it'll need to be a fairly hefty resistor too!

Laws of diminishing returns and all that come into play for my money.

Cheers,
Alan.

[jerber]

Originally Posted by Garfield2 Err I'm pretty sure from my reading of the LMD18245's spec sheet that parallel LMD18245's on a common coil ain't such a good idea. Their MOSFET based current sensor and limiting circuit will all be screwed up with devices switching the same load in parallel. In fact I remember the datasheet saying that you can't do this (unlike the L298 which can happily do this as pminmo has skillfully shown. ... Cheers, Alan.

But how do they do it/get away with it then ?

From the schematic from their site, "Building instructions 6A" under document (It's a flash site so I can't link ), I understand it's 2 LMD18245's in parallel.

Maybe I just buy the card and see what it does.

[wsc]

Originally Posted by jerber On this site, http://www.sb-microstep.com/ , there is a board that uses two LMD18245's in parallel to get 6 amps. Maybe a modular design is the way to go in order to get higher amperages (3,6,9...). What kind of AVR are you using ?

Those are L298s, not LMD18245s. I'm thinking of using a mega48 or mega88. Hopefully I can squeeze control of two motors into one chip, I'm not sure though. Here's an updated schematic.

[pminmo]

couple of things. #1 if your planning on implementing the bipolar drives in COTS logic, you diagram won't work. #2 I think your going to put to much real time activity on the avr. You talk of smoothness, what is your real need?

[wsc]

Phil, I dont see it, how would it not work? I think the AVR would be loaded down, but these chips have instruction rate as high as 20Mhz. Thats alot of processing power. The application is film related, really demands smooth low speed operation. Has to be as silent as possible too.

Edit: mega48/88/168 have 6 channels of hardware PWM. No overhead there.

[pminmo]

wsc,
Your pullup bipolar is biased through the inductive motor coil. How do you intend to close the loop for pwm?

[wsc]

Yeah, the circuit has alot of problems as drawn. Too many exams right now, will post majorly redesigned circuit in a few days.