BLDC motor driver

Hi everyone,

we need to design a 3phases BLDC motor driver with optical encoders as a position feedaback and the spec on the motor interface is something along these lines:

Rmin = 12 Ohm Rmax = 18 Ohm L = 15mH Vmax = 20V (+/- 25%!!) Imax = 0.85A (+/- 15%!!)

and that's it. I guess they still have no clue about what type of motor they are going to have therefore they are trying to take some lousy tolerances on some values (Vmax and Imax).

But on top of that isn't this spec incomplete? Shouldn't I need to know the motor constant to know how much power I'm going to deliver? Shouldn't I need to know how fast the motor can spin? What about the control stability? Shouldn't I be wondering what kind of torque ripple is accepted?

Am I missing something here?

Any comment is appreciated.

Al

--
A: Because it messes up the order in which people normally read text.  
Q: Why is top-posting such a bad thing?  
 Click to see the full signature
Reply to
alb
Loading thread data ...

It looks more like "they" have a range of motors in mind, like someone is in the midst of a custom motor design, and no one is sure exactly how it'll turn out, or like someone is trying to test in some performance, again without knowing exactly how it'll turn out.

The motor constant, by itself, doesn't tell you that unless you know the final intended motor speed.

What is missing is the motor's no-load current: with that, and solid numbers for Vmax and Imax, you could pretty much predict the maximum output power at the motor's ideal speed. 'course, you'd need the motor constant to know that ideal speed.

Probably yes on most of these. You want to know how fast the motor _must_ spin, and how many poles, so that you can determine whether the motor inductance is going to have much affect on performance.

I would assume that control stability is a given -- what conceivable reason would someone want an unstable motor controller?

Torque ripple is more slippery, because that depends on the motor's magnetic circuit, and the interaction of how the motor is wound and how you drive it (but the fact that it's BLDC implies a trapezoidal back-EMF, and a constant-voltage drive).

There's not much you'll be able to do about cogging torque induced by variable reluctance, at least not unless you're working on your PhD and not on a practical motor control project.

Probably. I don't know of anyone who never, ever screws up -- I just try to design as defensively as possible, so my screwups generally have mild impact.

--
Tim Wescott 
Wescott Design Services 
 Click to see the full signature
Reply to
Tim Wescott

I have no particular issues with them being still wondering some details, but it means that I expect them to *fix* their spec once the motor is defined. Is it too much to ask? Or that tolerance range is actually doable?

yep, it comes after :-)

Thanks for the 'no-load current' hint, I'll need to look it up, since I'm not really sure what it is needed for (but I take that as part of my 'homework' ;-)).

Talking about Vmax and Imax, if I understand them correctly, they represent the maximum voltage and current I can drive the coil (for each phase). Is that correct?

Could you elaborate here? I have not completely understood how the number of poles is linked to the inductance. Is it because a certain number of poles determines my activation sequence and therefore, for a given speed, how the time constant of the inductance is going to affect my effective torque?

I'm not sure what performance means in your terms.

Yeah, it was badly stated. We are driving an interferogram and precision in the position stability is a key element, therefore I believe the controller shall be specified accordingly in order to achieve that. The controller delay also may play a rather important role since a big delay between the feedback and the countereaction will have an impact on the driven current (therefore speed, therefore position).

We sens the current in order to control the current we drive the motor with. What about sinusoidal back-EMF? Why you say is a trapezoidal back-EMF?

what is 'variable reluctance'??? BTW, no PhD here, still to decide if fortunately or unfortunately!

That is not very reassuring though ;-) I guess it would be wise to get from our customer at least a model of their motor in order not to miss the boat completely...

Al

Reply to
alb

The tolerance range would be weird for a finished motor.

No-load current is a consequence of all the motor losses that happen just because it's spinning. It's a combination of friction, magnetic effects (eddy current and hysteresis in the iron, mostly) and (sometimes) windage (air resistance).

Yes and no. Yes to Imax -- it's the maximum current that you can put through the motor without things overheating. Vmax, less so -- normally, small motors are rated to work at a certain voltage, but that rating has more to do with the voltage needed to drive the motor fast enough that mechanical damage may occur (usually the first thing to go is the bearings, from overheating, but sometimes it's the armature flying apart from centrifugal force).

The spec _is_ weirdly stated.

More poles = higher number of electrical cycles per revolution = less time to charge up inductances.

In this case I mean achieving full torque at full speed.

Ah -- I should have connected the name to the project.

It's unlikely that the electrical dynamics of the motor by themselves will come into play as a performance limitation. The motor's rotor inertia and cogging torque (the variable reluctance affect, about which I'll speak more below) may have some pretty devastating effect, but I doubt that the motor's inductance will come into play.

Since I now know that this is for a precision servo application:

MAKE SURE IT IS CORELESS!!!!

Cogging torque is a royal bitch, and can be a complete performance- killer, if by "performance" you mean the ability to quickly settle in one spot or another repeatably. If it _isn't_ coreless, and if the motor manufacturer or your boss leans on you to use it anyway, then make sure to do an analysis of how much the cogging torque will screw you up, and share that with your boss.

Industry terminology seems to mostly use "BLDC" to apply to a motor with a trapezoidal back-emf. I can't remember all the terms for motors with sinusoidal back-emf, but I think the most common one is "brushless AC". If it's an issue, you ask the motor manufacturer. They'll probably act like you're a bit gormless when they answer, because everyone in their world (that is, them) just _knows_ the terminology.

You may want to get yourself a copy of a book on motors. Mine is "Electric Machines" by Mulukutla Sarma, which over 30 years old and is really about GREAT BIG motors, generators and transformers used in power generation and heavy industry, but the basics are the basics, and it has a chapter on small motors.

Reluctance is the magnetic equivalent to resistance -- given a certain amount of magnetomotive force (MMF, usually expressed in Ampere-turns; it's the product of turns of wire and current), the reluctance determines the amount of flux that is generated in a magnetic circuit. Magnets attract metal because the metal reduces the reluctance of the magnetic circuit, which reduces the flux and hence the amount of energy stored in the flux.

In a motor that has a moving iron core (or moving magnets), as the shaft turns the iron gets, on average, closer to or farther from the field magnets, varying the reluctance. The magnetic attraction between magnets and iron makes cogging torque.

Oh, it's reassuring if you hold your mouth right. For instance, you never have to worry about the answer to "will I screw up?" because the answer is always "yes". "Will I screw up badly and have it noticed?" is a worrier, but not the simple "will I screw up?"

Don't expect to get a model of the motor. Motor manufacturers don't do models -- they hand out specifications, and expect you to make mathematical models from them.

Fortunately all DC motors -- even BLDC motors with correctly-done drivers

-- have pretty much the same mathematical model; you just need to plug in the right constants.

--
Tim Wescott 
Wescott Design Services 
 Click to see the full signature
Reply to
Tim Wescott

Hi Tim, sorry for the delayed reply... I was strangled at work with an inevitable series of deadlines all crammed together.

Tim Wescott wrote: []

I thought so as well and they confirmed they are trying to get the interfaces to 'converge' in the near future...

But on top of adding an extra loss to the power I'm delivering I do not see it as a big element to be added in the spec for the controller...uhm but it's true that my PID or whatever is implemented may skip the extra effort to compensate a feedback which is not linear (until the no-load current limit is reached there's no displacement at all).

We are regulating the current as it is in our spec, therefore I do not see how the Vmax could really impact us. On top of it Imax*Rmax is way below Vmax, therefore I do not see the reason of the requirement.

What happens instead when I'm breaking? Isn't the current inverted and therefore back-EMF inverted as well. adding to the applied voltage with the possibility to exceed Vmax?

Is there a chance that Vmax is indeed related to an insulation of the windings? ot to some other insulation?

Ok, that means I'm not alone! Doesn't proove anything, but yet gives a sense of reassurance ;-)

[]

Ok, this is what I had in mind but your explanation is definitely clearer than mine!

Can you really have full speed and full torque? I thought maximum torque would be with 0 speed (max current), while maximum speed would be in a no load condition and 0 torque

formatting link

Well, I'll be sitting on this project for the next 3/4 years, I guess you'll have the time to get used it ;-)

I do not need to specify the motor. Indeed is our customer who has a selected team of specialists to pick up the best fitted motor for the application. I only need to make sure they are not screwing us up with constraints that will never make the system work together.

is a Fourier Transform spectrometer (Mechelson interferometer), with the moving mirror needing to displace at constant speed during observation and reversing the direction at the end of it (reversal phase): _ _ _ _ _ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ ... / \_/ \_/ \_/ \_/ \_

The reversal phase does not need to be extremely precise, but we need to brake and reaccelerate in 80ms, while the constant speed part is 800ms. I do not know if that corresponds to your idea of 'precision servo application'

I only want to make sure that if the guys who pick up the motor do not generates us too much trouble for the control. I'm not in the position to decide what is the need of precision or sensitivity, but I want to avoid to end up in the situation where we cannot meet the spec because they screwed up!

[]

I got one, but is not covering brushless dc motors, so I guess I need to look for another one...

[]
[]

we are building the driver, I'd rather ask the guys you are building or buying the motor to provide us with the model. On top of it the controller we are building has an algorithm which is 'designed' by our customer, we *only* need to implement it.

So you are suggesting to get the model verified against the real artifact ASAP in order not to be doomed later. In a previous project, also about an interferogram, the customer got it completely messed up and we (well, I was not yet part of this company) had to redesign the interface from ground zero. Definitely something that I'd like to avoid.

Al

Reply to
alb

Later in your response you indicate that you'll be going at a constant speed -- the worst of this set of nonlinearities won't bother you, then.

It's traditional.

The back EMF is a function of the rotor velocity (speed and direction). It doesn't change when you reverse current. In fact, just shorting the windings together will provide some braking.

Not at the mentioned voltages, no. I think you need to be over 100V before it starts to matter at all.

It depends on who's specifying the motor and how. You CAN achieve full speed at full torque on most motors, but you may have to exceed the "rated" voltage to do it. Some motors have the top right corner of their safe operating area lopped off, because it can't stand the heating from both the bearings and the coils.

Oh you poor bastard.

UH, I MEAN -- HOW FORTUITOUS!!!

Is the selected team of specialists competent? At something other than retaining their jobs and looking good to da boss? (Am I cynical?)

If there, the cogging torque will act like a periodic disturbance torque, speeding up and slowing down the thing when it's running. Depending on how fast it needs to go and how much acceleration/deceleration you can stand, that's either a no-matter or a deal killer.

It may -- the precision part matters more for how tightly positions must be held.

Make sure that the motor, when driven at maximum current, can reverse speed in 80ms. This will depend on the maximum torque and the armature moment of inertia (coreless motors have much lower moments of inertia, and thus can accelerate much quicker if you don't hang humongous flywheels off of them).

See my comment about fortuity, above.

If it covers brushed DC motors you've got 90% of the information. A BLDC motor is just a brushed DC motor turned inside out (so the magnets move and the coils don't), with the commutation happening electronically.

That's a good idea.

--
Tim Wescott 
Wescott Design Services 
 Click to see the full signature
Reply to
Tim Wescott

nevitable series of deadl> []

ls, but it means that I expect them to *fix* their spec once the motor is d efined. Is it too much to ask? Or that tolerance range is actually doable?

aces to 'converge' in the near future...

I'm not really sure what it is needed for (but I take that as part of my 'h omework' ;-)).

st because it's spinning. It's a combination of friction, magnetic effect s (eddy current and hysteresis in the iron, mostly) and (sometimes) windag e (air resistance).

see it as a big element to be added in the spec for the controller...uhm b ut it's true that my PID or whatever is implemented may skip the extra effort to compensate a feedback which is not linear (until the no-load cur rent limit is reached there's no displacement at all).

esent the maximum voltage and current I can drive the coil (for each phase) . Is that correct?

through the motor without things overheating. Vmax, less so -- normally, small motors are rated to work at a certain voltage, but that rating has m ore to do with the voltage needed to drive the motor fast enough that mech anical damage may occur (usually the first thing to go is the bearings, fro m overheating, but sometimes it's the armature flying apart from centrifug al force).

see how the Vmax could really impact us. On top of it Imax*Rmax is way below Vmax, therefore I do not see the reason of the requirement.

erefore back-EMF inverted as well. adding to the applied voltage with the possibility to exceed Vmax?

ndings? or to some other insulation?

Anything is *possible*. Enamelled wire is usually good for at least 500V so that isn't all that likely.

nse of reassurance ;-)

number of poles is linked to the inductance. Is it because a certain number of poles determines my activation sequence and therefore, for a given spee d, how the time constant of the inductance is going to affect my effective torque?

ss time to charge up inductances.

r than mine!

would be with 0 speed (max current), while maximum speed would be in at n o load condition and 0 torque

rve-01.jpg)

Torque is essentially controlled by the current through the motor windings. If you've got a high enough supply voltage to drive the maximum current th ough the windings, against the back emf and and the winding inductances you can - in principle - generate maximum torque at any motor speed.

If the rapidly changing current in the motor windings generate inconvenient eddy currents in the rest of the motor, life can get more complicated.

u'll have the time to get used it ;-)

Sounds like a fun project.

will come into play as a performance limitation. The motor's rotor inertia and cogging torque (the variable reluctance affect, about which I' ll speak more below) may have some pretty devastating effect, but I doubt t hat the motor's inductance will come into play.

ected team of specialists to pick up the best fitted motor for the applicat ion. I only need to make sure they are not screwing us up with constraints that will never make the system work together.

ler, if by "performance" you mean the ability to quickly settle in one spo t or another repeatably.

he moving mirror needing to displace at constant speed during observation and reversing the direction at the end of it (reversal phase):

brake and reaccelerate in 80ms, while the constant speed part is 800ms.

ns on you to use it anyway, then make sure to do an analysis of how much t he cogging torque will screw you up, and share that with your boss.

enerates us too much trouble for the control. I'm not in the position to d ecide what is the need of precision or sensitivity, but I want to avoid to end up in the situation where we cannot meet the spec because they screwe d up!

r
s

from our customer at least a model of their motor in order not to miss the boat completely...

models -- they hand out specifications, and expect you to make mathematic al models from them.

buying the motor to provide us with the model. On top of it the controller we are building has an algorithm which is 'designed' by our customer, we * only* need to implement it.

rs -- have pretty much the same mathematical model; you just need to plug i n the right constants.

artifact ASAP in order not to be doomed later. In a previous project, also about an interferogram, the customer got it completely messed up and we ( well, I was not yet part of this company) had to redesign the interface fro m ground zero. Definitely something that I'd like to avoid.

Separating the design of the driver from the selection of the motor seems l ike a really bad idea.

Back in 1978, at EMI Central Research, I had to cope with some mechanical e ngineers who through that you couldn't run a stepping motor faster than it' s natural resonance. The people who were selling us the motor knew that it was done all the time, which left me free to get on with doing it - the acc eleration sequence to take the motor through resonance had to be coded into a PROM back then, but it wasn't all that complicated, and worked fine.

--
Bill Sloman, Sydney
Reply to
Bill Sloman

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.