Step Motor Drives

[Mariss Freimanis]

OK, so I posted this to a private message I received but it puked because the reply was too long. I think it has general interest so I posted it here instead. Scroll down past (5) to a "Friendly Rant" to get to the non reply-specific stuff.

I've thought long and hard about my reply. It is because the answer is complex in ways you may not have considered. Here goes:

1) It is the easiest thing in the world to design an all n-channel MOSFET dual bridge in the 5A to 15A range, 100VDC max rated. Use 4 International Rectifier Inc. IR2104 half bridge MOSFET drivers and it practically designs itself. If you like, I can send you the schematic of our drive's power stage.

2) It is much harder to do a printed circuit board layout for this power stage. MOSFETs are so fast and carry such high pulse currents that wire (or pcb traces) has to be considered inductors in series with resistance or it flat-out won't work. The same design layout done right will work, done wrong it won't right and may even blow up MOSFETs. High current power stages cannot be hand wired or breadboarded for that reason.

3) Your goal is to tack-on a high power output stage to an L297 or some Allegro chip. In my opinion that is a misguided goal. The Allegro device is an integrated solution where the controller and power stage are in the same IC. This means the signals of interest, those going to control the power stage, are unavailable. You cannot retrieve these signals from the presented motor outputs without a great deal of difficulty. The L297 at least gives you the A,B,C and D outputs needed. Still, the L297 device is a dog and the result will be a crippled dog.

4) Don't assume Allegro or ST know what they are doing or are good at it. They really don't and they aren't; they are silicon IC vendors. Both vendors put out chopper-type controllers. If you love the hissing, squealing and whistling of a chopper at 2A, just think of how you will like it at 6A.:-) The Allegro part is particularly hilarious to me with its "mode switching on current slope" design. It introduces a discontinuity in the motor current on "mode switching" that makes the motor resonate like all Hades. The idea is completely naive because they didn't consider the consequences of this seemingly good idea. Again, if you like, I can send you a design that will accept the L297 as a controller.

5) I don't know your age, electronics background or the level of the fire in your belly. If it's "young", "some" and "could melt iron", then consider learning a lot. Make the effort; become an expert on step motors and use what you learn to design really good drives. That effort will bring you a lifetime's worth of very good income if you meet the criteria.

A friendly rant:

Choppers really suck; there are far better techniques. There are 5,156,023 step motor drive mfgs and vendors. Their designs all breakdown (worst to best) in:

a) $0.99 L/R drive full-step drive rated at 20A, 5VDC sequenced from the parallel port. Software and schematic included.:-) Usually an ASCII based 48-point font website in red and green.

b) $9.99 L/R full-step/haf-step drive rated at 20A and 100VDC if you use 2kW current limit resistors, 2 per axis. Usually 1, 2, 3 or 4 axis per board. Toned down 24-point website.

c) $19.99 L/R microstepping drives. Same ridiculous ratings, same ridiculous megaWatt current limit resistors required. Usually has some really cool name and the website owner wants to share his whacked-out philosophy. White background website, tons of multicolored boxes, most often blue background, yellow text. Kits and soldering instruction supplied, of course.

d) $29.99 L297/L298 or Allegro based chopper drives. Usually 1, 2, 3 or 4 axis per board. Why someone would have more than 1 axis per board is beyond me but that's for another day. You blow up a drive on a 4-axis thing and you are out 3 perfectly good drives; go figure. Drives are rated at their absolute limits (ST's L297/L298 already over-enthusiastic limits) They are good for 1A at 24VDC, the product is usually rated right up to the limits: 2A and 46VDC; good luck. The website more professional.

The exception is Xylotex. They have a good design and they don't mess with max limits. Lots of other good stuff about them like support.

e) Big gap here. Prices really jump and you get into industrial drives. The players are also whittled down, API, Parker Hannafin, Anaheim Automation, etc. Maybe a few dozen others. These drives are characterized by full-size MOSFETs (TO-220 package), 80VDC or more voltage and at least 7A of rated current. The drives include essential features like mid-band stability, protection against short-circuits, etc. These are drives designed for industrial 7-24 usage and they cost from $100 to way up.

Let's take a break here. Some in group (e) can also turn out crap just like (a) thru (d) so how can you recognize a crap drive from a good one? Let's take a look at how a low-power, hobby drive can be recognized. The most important thing is the drive picture. So take a look at a picture of the drive.

If you count only 4 power transistors (3 legs, a black body 0.4" square and a metal tab with a hole in it), you are looking at an L/R drive. Discard it unless $$$ is of paramount importance. If it is, get ready to live with severe performance limitations.

If you see something mounted to a heatsink with 15 leads about 1.5" wide, you are looking at an L298 chopper drive. Cavetat emptor.

If you are looking at a metal heatsink in the middle of a printed circuit board, you are looking at an Allegro device. Again, Cavetat emptor.

What you want to see is at least 8 three legged, black body 0.4" square devices with a metal tab with a hole in them. Then you are looking at an industrial rated drive.

Mariss

[poordumbbastard]

Mariss Freimanis,

What is stopping anyone from copying or building your 8 three legged, black body 0.4" square devices with a metal tab with a hole in them?

Since you have made the best drive, other manufactures might just buy one of yours and reproduce it.

PDB

[irving2008]

Mariss,

I don't know who your reply was to, but considering some of the posts I have seen and my inbox recently I can hazard a few guesses. I hope they understand and accept the value in the advice and experience you give.

By way of adding weight to your arguement on points 1 and 2 my background, many years ago, was the design of high-power VHF/UHF transmitters for aerospace applications and I worked on some of the earliest MOSFET devices for that purpose. I haven't done much in the way of design for some years, except as a hobby thing and MOSFETs have become incredibly cheap and easy to use since those days, but PCB layout is still the critical factor and getting it right is still as much art as science amd hasn't really got any easier. In those days two seemingly identical boards would behave entirely differently and from what I have read on here sometimes things dont seem to have changed that much!

I am researching the design and build of a decent high current driver. Even though I know that it will take more than one attempt to get it right (and it'll probably cost more than buying something prebuilt in the long run) but the experience and lessons learnt and the challenge is part of the fun. I'm not out to build a CNC device to make money from it, its a hobby, so time, and to some extent cost, is not an issue. A bit of me baulks at buying something I know I can design and build; my experience however tells me to not underestimate the complexity of the task...

In answer to 'poordumbbastard', basically from a pure hardware perspective, nothing is stopping that, although there are laws against it assuming they are enforcable. The fact is that there are companies in the far east who specialise in doing just that... therefore you protect yourself by using something proprietary - a gate array, software or whatever, thats not so easy to reproduce (but nothing is impossible to replicate - witness AMD's green room re-engineering of the Intel Pentium). However price isnt the only reason we buy from one company or another... support, quality, positive feedback all come into the buying decision.... I know where I'd put my money....

[Mariss Freimanis]

irving2008,

1) Board layout is incredibly critical. Example: When I did the G250 prototype board layout (REV0), the hand-built drive was audibly noisy. The noise was traced to the Kelvin ground connection at the current sense resistors and the sheet resistance of the ground-plane copper. I thought it was good enough when I laid the board out; it wasn't. The entire power section layout needed to be completely redone, this time with insane rigorousness regarding current loops and the Kelvin ground pick-off point. The biggest part location move was 0.2" but it made all the difference. MOSFETs require great care in pcb design.

2) There is nothing in the MOSFET bridge circuit design that even remotely constitutes proprietary information. It is brain-dead simple so copy away. Intellectual property considerations apply to the circuitry that drives the bridge inputs; that's where all the magic is. It is naturally protected because it resides in a CPLD which has a readback security bit set. The only way to reset this bit is to erase the CPLD. All the external circuitry is analog and services the CPLD.

Mariss