How hard is it to drive a 3-ph brushless motor?

Aren't we talking about stepper motors here? I reiterate, what _possible_ failure mode could drive a stepper to max. speed? This would be outside the realm of what I know about stepper motors and controllers.

Please enlighten me.

Thanks, Rich

-jg wrote

There is nothing above that can be used without a micro. Just been onto ON Semi. Looking at the data sheets and the power-up register values, one needs a micro just to enable the drivers...

This turns out to be much more interesting. I didn't even know these things existed.

A3987:

Microstepping, the whole lot, no micro needed.

What I find staggering is that the above chip comes in a miniscule package, yet it is rated at 1.5A and should thus easily cope with this motor

which should draw 0.125A per phase (at 12V). I am really puzzled how such a motor could draw so little current. Is that 0.125A figure meaningful? The coil resistance is 75 ohms so it can't draw that much! The price of the motor is interesting - about US$500!!

This one

uses external MOSFETs and should be a lot more robust...

There is a vast difference - about 10x - between the torque of a stepper like the one above and a brushless motor e.g.

(roughly 10mNm v. 100mNm) which probably explains the different currents involved.

Rich Grise wrote

I think it would only be a fault a long way up the chain, where one is dealing with a straight control voltage.

If one can scale the control voltages such that they are nearly up to the supply rail then not a lot can happen.

If you see my other post, I am moving to a stepper solution. A stepper is much weaker than a brushless (10x weaker for a similar size) but even a weak stepper can produce enough torque to take the gearbox output shaft to its design limit, so I think a brush or brushless motor would be an overkill, and a motor of that size is needed only because a smaller motor cannot be mated with a gearbox whose output shaft is sufficiently thick for this application (6mm).

The only failure I can imagine with a stepper solution would be one which causes the output of the V to F converter to go to a very high frequency. Any V2F is going to use a timing RC of some kind and if the C came off the PCB then you could get a high frequency coming out. The stepper is capable of a few thousand RPM but would normally be running at 200 RPM so there is potential for a high speed under fault conditions.

OTOH if I configure the Allegro controller to do microstepping then that scales down the motor RPM v. input frequency relationship, and a x16 microstepping would limit the max motor RPM to a few hundred, no matter what frequency one applied to the controller 'step' input.

I hope I am understanding this right.

There is also LS7290

which seems to do the same job...

Step and Direction inputs. Needs one of these

for driving big motors, but I can't see I would need that...

How did a V to V get into this?

EXACLTY!

Whew! ;-)

Thanks, Rich

Peter, just got an email from ST and in there was an announcement about this chip, in case you are still considering a stepper solution:

Can supposedly do 128 microsteps, and maybe this could eliminate you gear box? Digikey doesn't carry it yet and if suitable you'd have to call ST about status and samples.

Just thought it might be interesting for you.

Joerg wrote

That's a very impressive chip Joerg - thank you. Unfortunately it needs a processor, which adds a whole dimension to the work involved.

Currently I am working with the Allegro 3987

and to my suprise Allegro are really supporting it, with replies to emails!! I have never seen that kind of support before. But then I am used to Hitachi H8/300 etc ;) ;) ;)

I don't think microstepping is quite what some people (incl myself) expect[ed]. You do get smooth rotation IF the motor is actually rotating continuously, so it cuts out the normal stepper motor noise. But you don't get the angular precision which the microstep size might imply - because the motor has no actual detent in between steps. You get some kind of a fraction of the microstep precision, and there is also less than the normal torque available between the full steps.

At my speeds (200rpm max, maybe 10rpm min) x16 microstepping should produce smooth rotation but I don't need the angular precision. In a nutshell I am using a stepper rather than a brushless (which was the original idea) because a brushless would need a tachometer and feedback, to deliver any speed stability. A tacho is not hard to do (most brushless controllers provide a pulse output which can be lowpass filtered to give a voltage proportional to RPM) but the control loop for the motor rpm obviously involves the usual control loop parameters which will need to be developed with the appropriate margins to ensure stability under all speed, load, temperature and transient conditions. Whereas a stepper gives you implicitly accurate rpm.

It is quite self-contained but yes, you need to feed it the speed and all that via SPI.

They are very responsive, it's just that I've had not so great experiences with EMI and internal noise pollution from the chips. We had sensors on the chassis that would pick up any mechanical shaft noise which the chip generated.

True, mcro-stepping won't net you quite as many extra intermediate positions as there are micro-steps. The only way to do that would be to roll your own controller with a uC. Some day (when you have the time) it may be a good exercise for you to do that because you can use such acquired know-how over and over again.

With a loop I wouldn't worry too much, essentially it gets tuned just like any other regulator such as a PID. But of course not need for that with a stepper.

So, make up your mind. Do you want smooth rotation, or precision positioning? If you want both, it could be very pricey.

What exactly are you trying to accomplish here? What's the goal?

Thanks, Rich

Joerg wrote

Interesting... I will have a 4-layer PCB, with ground planes properly done and decoupled. MOSFETs switch very fast so the potential for muck is substantial.

Why were you sensing mechanical shaft noise?

The thing which concerned me with a brushless motor is that in the standard driver chip implementation the rpm is totally open loop. You have the Hall sensors, and the coil currents are switched purely according to the Hall sensor position feedback. So, if you have zero load and zero friction, the motor will zoom up to an infinite speed :) So, in terms of a control loop trying to hold the rpm, you need a significant margin in there for stability under a wide variety of load conditions. Whether this is done in software or with analog components, makes no difference.

Obviously this problem has been solved - brushless electronically controlled motors have practically taken over the variable speed drives market which is a multi billion sized field - and I would think a more intelligent controller would not just blindly switch the coils around from the Hall sensor feedback but impose a more intelligent regime on the angular speed of the rotating field. And I know, from peripheral involvement through my customers, that a large chunk of the AC drive market uses motors with a shaft encoder because that is the only way they can get any torque at low rpm. In fact I tried to learn something from my project from these people but the stuff they are doing is so different. They aren't using off the shelf brushless controller chips; they have fast microcontrollers (often DSPs) running some pretty complex algorithms.

The brushless controller chips are great for model airplanes and I see some amazing stuff there, but all they want is to stick 50 amps into a tiny motor and get 20000 rpm from it driving a plastic prop :)

But a stepper motor solves this neatly - if the parameters can be met with a stepper which in my case they can be.

Rich Grise wrote

The goal is 0-200rpm (say 10-200rpm in practice).

It has to last "for ever" which rules out a brush motor. Has to be brushless with ball bearings everywhere.

The output angular accuracy willbe done via a gearbox of about 50:1.

If you have the inclination, replace the "standard driver chip" with a small MCU. The R8C family has many chips with built-in three-phase motor control, and can support either hall effect sensors or back EMF sensing. That gives you speed control and a digital channel to talk to it with (i2c, can, uart).

It was an optics measurement setup where the slightest vibration modulated the optical path and showed up in the signal. The Allegro chip was sort of self-polluting. I didn't get very far with the support guys but assume that the internal oscillator became modulated by the power paths on the chip. I am not a fan of chips where the power path is integrated, I rather use external FETs.

IOW the whole EMI dog fight seemed to have gone on inside the chip and there was nothing we could do on the outside, other than roll our own solution.

Yes, your case almost screams for a stepper. 15-20 years ago that was different because the prices were outrageous but that has changed.

Have a look here:

Getting from here to your application....

I can't remember the beginning of this thread: was this a one time solution or are we talking numbers here? For around EUR 100 you can buy 3 phase motor drivers. Why bother developing your own solution?

Meindert

Joerg wrote

I think I will design the PCB with some LC pi-type filtering in the supply rails, to keep conducted muck from getting out.

Have you seen the price of that stepper? :) $500 plus...

Of course it is much cheaper if purchased in Germany where they make it...

Steppers can either come with cut-throat pricing when used in disk drives, cars or the like. Or they come with super-fat profit margins if you pick a boutique size. Shopping around pays big time if your quantities are high.

Order it there and have them send some Niederegger Marzipan along with it. Good stuff but not recommended by the surgeon general :-)

=20

There is one thing to remember about steppers, you need to ramp the=20 stepping speed at the controller input. Even a little inertia can=20 surprise you.