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

On a sunny day (Thu, 10 Dec 2009 00:50:15 +0200) it happened Niklas Holsti wrote in :

Ah, yes, makes sense, thank you for the correction.

his

e.

A micro-stepped stepper would be ideal: smaller, lighter, cheaper, more rugged, and mechanically more reliable than something with a gearbox turning 8k rpm.

There are plenty under 35mm. Do you need super-smooth rotation or super-precise shaft positions? If not, the world's yer oyster. If so, maybe even then, too.

-- Cheers, James Arthur

I haven't heard of them being serviced, either.

"Scuttling" of the satellite is a bit tricky. Actually de-orbiting it (and having it burn up in the atmosphere) would require a huge amount of delta-V - far more than the remaining fuel in the station-keeping thrusters is going to be able to achieve.

The current approach seems to be to move the satellite to a non-geostationary "graveyard" orbit, a few hundred kilometers above or below the geostationary altitude. This requires a much smaller delta-V than a de-orbit operation.

In message , snipped-for-privacy@yahoo.com writes

The original poster specified high reliability over a period of years; I don't see a 600:1 gearbox lasting that long without maintenance or replacement. It will also exert a significant frictional load on the motor that will change over time due to wear and lubricant expiry, causing more control problems down the line. In addition a DC motor capable of running at up to 8000rpm continuously for years on end will be very expensive; what are the MTBF figures for such motors? 10,000 hours, 50,000 hours?

A stepper motor gets rid of the requirement for a gearbox; fewer moving parts are a good thing, but the long-duration operation means whatever the prime mover is it will have to be a hi-rel part and they are never cheap whether it is a stepper motor or a DC brushless motor.

That would be a clear sign to move on and find another manufacturer. At least if this goes into production. Just imagine when a supply issue creeps up and you are afforded the same sort of "courtesy".

If you can't trust the controller or control signal and bad things can happen when it's over-revving it is almost a must to do that.

can you define 'small' ?

The ideal candidate would seem to be micro-step controlled Stepper motors.

See Allegro and OnSemi, for chips that do all the drive, and current ROM handling for Micro stepping.

Steppers will give very precise average speed, but will still step, so you need to define how much non-linear phase velocity is tolerable.

-jg

Oh well, we have turbo pumps that do 50k RPM :)

Perhaps a BLDC motor, like the one found in any old hard drive might do the trick ... and the back-emf commutation method should work ...

Cheers, Rob Sciuk

Sneddon

Yep. Not cheap, but cheaper than a high-rel 8Krpm motor + gearbox.

Speed control? Trivial, of course.

-- Cheers, James Arthur

For BEMF commutation, the motor's got to turn some minimum speed, which suggests you'll need to spin faster and gear down.

That's more parts.

If the OP can tolerate the stepping motion (torque ripple), many-pole motors (steppers) really shine here. They've effectively a lossless gearbox built in (the many poles), with zero moving parts.

-- Cheers, James Arthur

-jg wrote

OK, some very good input here, many thanks everyone.

The gearboxes seem to be rated at say 5000rpm so the motor would have to slow down a bit anyway.

Here is a bit more info:

This application used to be fulfilled with a standard brush motor running at 1000rpm. It is believed this motor was custom wound. The gearbox then didn't need such a big ratio.

The motor was being driven with a botched PWM circuit which used to blow up. However the motors used to wear out anyway after a few hundred hours of operation. None of these motors seem to have a life spec (despite their high price) but dismantling one showed the commutator to be 1/3 worn through. It is clearly a crap solution. That's why I was going to attack it with a brushless motor.

I suspect that the above comments about gearboxes are why the brush motors were specially slow versions of ones which normally run at ~5k rpm. So the manufacturer must have learnt something.... even if they botched the rest of it.

It seems clear that if I go brushless I will need to run the motor slowly (maybe 1000rpm) for the sake of the gearbox, and have a separate rpm limiter for the safety angle.

I think a stepper would do it, but I have two problems with a *direct drive* stepper:

2) I have no way of determining the worst case required torque (yes this seems a big issue but I was going to deal with it by speccing a gearbox similar to the one currently used, and a motor at least as powerful as the one currently used, which is easy)

My guess is that a stepper with a 10:1 gearbox would do perfectly.

The motor originally used was made by Globe USA and was very similar to this one

So 100 ox-in and 13rpm are the parameters.

I am happy to look at steppers. Are there some widely used stepper motor controllers which would give me +100% to -100% rpm with a -10V to +10V control voltage?

With a 10:1 gearbox I would need 130rpm on the motor.

There will have to be an rpm limiter but I can design that.

I see there are 1-phase steppers and 2-phase steppers....

Most stepper controllers are either SPI, or Step.Dirn types. If you wanted Linear control, then a V to F converter could give a micro-less solution. (or, you use a small uC with a ADC and a reload Timer, to build the V to F)

eg AMIS-30542 has Step.Dirn control and also a SPI to configure the modes.

or, something from here ?

-jg

Most miltary and avionics has 20+ years which spares have to be made for the Nimrod is a 60's airframe and some of its avionics is still the same design. Some production runs for avionics I have seen being planned last year for 2015. You need to make exactly the same thing to avoid having to recertify the whole aircraft.

In some cases to ensure making parts the manufacturer or their production supplier of sub systems, literally do last buys on WHOLE Wafers of chips to ensure supply for 20 years.

...

As I have seen in other posts you may be doing lower RPM than that. If you are wanting RPM limiter and speed control then maybe a sensored motor would be easier as you will have feed back for RPM directly, three of them for redundancy.

I know of at least one ASIC for sensored with control signals, available in UK, email if you want details.

Just a few quick comments because I'll have to head out soon:

This can be muffled a great deal if the controller is able to do what's called "micro-stepping", meaning it can steer the motor in steps much, much finer that its actual pole number. Once we did that, the same stepper that would emit a loud GRRRRRR was now doing it's job without vibration and with just a faint hiss. _Then_ the marketeers wanted the vibration and growl back (!) because the customers could otherwise not hear that it was actually running. IOW we had made it too smooth for them.

Allegro makes lots of stepper controller chips but I found them too noisy electrically (EMI) because we often deal with low noise electronics right next door.

You could measure the torque of the old solution and then see if a stepper sans gearbox can be found that rivals that torque. As others have said a gearbox can become a reliability concern if high MTBF numbers are required.

-jg wrote

The 30512 is one of a few in a hand-solderable package

but it isn't clear if it can be used without a microcontroller. It has the NXT/DIR inputs so can take a V2F converter output, but if any of the config registers need initialisation, then a micro will be needed for that.

Joerg wrote

This is another concern. This thing must not make much noise. Can't quantify it though... I do know a brushless (or brush) motor would be fine, especially if running at say 2000rpm max (to make the gearbox last a long time).

The old motor+gearbox has loads more torque than is needed. The gearbox itself was rated at 0.7Nm torque, max, and it turns out that most similar gearboxes are similarly rated e.g.

the 43:1 version is rated at 1.2Nm so a stepper which can deliver 1/43 of that (plus a bit for losses) should be big enough.

I am looking at the AM2224

which at 200 steps/sec (to give me the required output rpm) has a torque of 0.3Nm which is about 50% of the max rating of the gearbox itself, so give or take a bit, this stepper should not be able to strip the teeth off the gears :) That is with a voltage drive; with a current drive you get loads more torque but I think the controller will be a lot more complicated - unless I misunderstand how the controller chips e.g.

actually work. It seems that they just switch the supply voltage to the two motor coils i.e. a "voltage mode".

I have heard of micro-stepping - a friend designed a linear motion thingy for a microscope many years ago, using the old Hitachi 64180 micro to do microstepping, but I am trying to avoid a microcontroller. I have done some AVR programming recently but I am not up to speed on it, and also even simple chips like the one above seem to do so much more than just switching the coild around...

Yes. That's why for the final product you PWM it to control the speed.

Exactly. What I just described are two separate processes. First, determine at what DC current the motor drives your motor at at max. speed.

Then, use that current value to design a constant current source/sink that can be PWM'd, and the speed, at that current/torque, will be proportional to the duty cycle.

Hope This Helps! Rich