REGISTER | LOGIN
Home    Bloggers    Blogs    Article Archives    Messages    About Us   
Tw  |  Fb  |  In  |  Rss
Brad Albing

Baby, You Can Drive My BLDC

Brad Albing
Page 1 / 3   >   >>
DaeJ
DaeJ
2/1/2014 10:31:47 AM
User Rank
Master
Re: Document find
My comments indicates that system would like to control four BLDC, not one BLDC. Depending MCU pin configuration and control scheme, one MCU or four MCU could be used. Or, one master and three slave architectures might be used.  

50%
50%
Victor Lorenzo
Victor Lorenzo
1/28/2014 4:21:34 PM
User Rank
Blogger
Re: Need info on BLDC types
Once again, thanks a lot Dennis for your comments and guidance. I'm also interested on your handouts, Steve and Brad both have my e-mail.

50%
50%
D Feucht
D Feucht
1/28/2014 12:57:15 PM
User Rank
Blogger
Re: Need info on BLDC types
Victor,

You would think that for a device (the motor) that is over a century old and with so many of them around that everyone who teaches about them would have a refined understanding of them by now. Not so. Let me tell you a story about this.

I was blessed to start my motor-drive involvement under a guy who came from GE. What I learned from him (and from Donald Novotny at U of WI, Madison, one of the top motor schools with former GE people such as Novotny and Tom Lipo) was that most schools teach the old steady-state motor theory worked out by Steinmetz and found in textbooks today based on phasors, which are steady-state constructs. What Paul Krause in his book is teaching is the total theory based on vectors. Novotny told me that for undergrad motor courses, they teach (at UWM) the steady-state theory because it is complicated enough for undergrads. However, Krause bites the bullet and does a total theory introduction in the book I cited. At first, it is a bit intensive mathematically (but no vector calculus!) but it includes total motor behavior.

Krause has an older book which is available from Purdue as a reprint and is out of print with McGraw-Hill, called Analysis of Electric Machinery. I consider this the top motor book around. However, it is not the best book for beginners or even intermediate people which is why Krause wrote the second (thinner) book.

If you find Krause and Wasynczuk's book to not be right for you, I recommend retreating to one of the many steady-state motor theory books. The one I like best is by G.R. Slemon and A. Straughen, Electric Machines, Addison-Wesley, 1980.

Historically, steady-state motor theory was adequate because motors were run at a constant speed as a power source for machinery. But the times have changed and they are used (as in Jeff's computer peripheral examples) for "motion control" which can have anything but constant speeds and accelerations. An intermediate motion category is that of "variable-speed drives". These operate at a constant speed but the speed can be changed (quasistatically) and dynamics are not an issue in them. Most motor courses in engineering schools teach the steady-state theory, so be aware that there is a more complete (vector) theory that corresponds to what is called field-oriented or vector control, implemented in motor drives. And invariably, the people who teach it trace back to GE technical ancestors.

50%
50%
D Feucht
D Feucht
1/28/2014 12:26:09 PM
User Rank
Blogger
Re: Need info on BLDC types
Jeff,

I hear your dilemma. It's not just about motors anymore either. Try to get magnetics parts for power converters in the US. Sadly enough, the US is shutting down and Asia is taking over.

As for servo (PMS) motors, except  for the few suppliers left (who usually sell into price-insensitive markets) or high-volume application-specific motors (in cars, for instance) what is left are step motors. They are really the same kind of motor (PMS) except that they usually have two instead of three phase-windings. And to achieve the high performance of computer-peripheral kinds of applications, I can see no other way than to select a motor that has the basic performance and then design a drive for it. Package solutions are a blindfolded dart game.

Indeed (as you previously noted) the drive will need to optimally interface to the motor and this affects drive design, but if the torque-speed curve of the motor is adequate (and with the right winding for supply voltage), then a drive can be designed to achieve maximum possible performance. Bausch and I were achieving from a step motor about 99 % of its theoretical maximum torque using vector (field-oriented) control.The ultimate limitation is motor temperature.

I am overcoming the urge to go on and on about various factors that have to be taken into account (such as armature reaction or the winding series impedance that makes the drive voltage waveform be shifted in phase from the internal winding induced-voltage source.) I once did a three-day course on motor and motor-drive design (both!) for NASA and have the hand-out notebook material in MS Word. If you are interested in a (free) copy, if I could somehow get your email address, I'll try to send a copy. It has all kinds of considerations that affect the design of motor drives (and PMS motors). You would have enough experience to appreciate some of the finer points in it.

50%
50%
Victor Lorenzo
Victor Lorenzo
1/28/2014 3:29:23 AM
User Rank
Blogger
Re: Need info on BLDC types
@Denis,

I found very interesting and useful several comments in your responses.

Thanks for the book reference, although there's plenty of literature about motors on internet I'll try to get a copy of that book.

50%
50%
JeffL_2
JeffL_2
1/27/2014 11:02:43 PM
User Rank
Newbie
Re: Need info on BLDC types
Sorry, now I can tell that your abrasiveness was unintentional, I hope you'll accept my apology. But you're coming out on the other side of the argument from where the industry is today, when I look to procure a "BLDC" all I'm able to purchase in reasonably small quantities is a motor totally integrated in its case with an internal susbsystem which is (usually) wholly inadequate for my needs. I don't think I've ever seen offered for sale a "PMS" (probably because from a marketing standpoint those initials are too likely to suggest a female complaint), but I suppose this is fairly similar to the hytbrid stepmotors available in the US from various subsidiaries of Danaher and others (although to be realistic this market was almost totally ceded to a variety of Japanese vendors many years ago), and yes admittedly accurate application information nowadays is very difficult to come by, maybe someone will get inspired to get this information republished for the motors available today, I don't see the current crop of vendors doing very much research about this any more. At least you're not FORCED (yet anyway!) like in the "BLDC" to feebly submit commands to a chip mounted internal to the motor and put up with what it gives you!

I suppose I was fortunate to have had the experience of working with reasonably high-performance servos fairly early in my career. The company I worked for made high-performance tape and disk drives, I worked mostly on disk but occasionally was "loaned out" when they needed help on the other side. Now our company made ANSI-compatible 7 and 9 track tape drives at many performance levels up to vacuum column. At the high end of the latter range there were drives that could move tape at rates up to 200 inches per second, and (as I recall) the capstan that controlled this motion was a little over an inch in diameter. In order to meet the standards that required that the inter-record gap on the tape had to be no more than .3 inches, that capstan had to accelerate to the specified linear speed within .13 inches of rotation, and be stable enough to be used to start writing data blocks immediately - now THAT'S what I call a high-performance servo, even the resonance of the 3/8 inch shaft that drove the capstan created MAJOR problems with stabilizing the loop at those speeds - we were doing similarly advanced work on the disk side (albeit in absolute terms somewhat more slowly due to the higher masses present in a disk head carriage, which in our designs used a linear "voice-coil" motor). If I could get motors to perform within half an order of magnitude nowadays of what we had then (and remember we did this without $.50 MCUs - or ANY for that matter, I left there in '80!) I could be happy but I guess it's not really "in the cards".

50%
50%
D Feucht
D Feucht
1/27/2014 9:20:40 PM
User Rank
Blogger
Re: Need info on BLDC types
Jeff,

My purpose in responding to your post was to provide a little information that might be of help in guiding you toward answers to the questions you posed. These questions are of a fairly basic nature to me and thus I recommended a motor book and offered some quick aphorisms about motors and drives. I am not, nor have I ever been in the employ of academia, though I have given talks on power electronics at Carnegie-Mellon U., Cleveland State U., and Portland State U.  

You write in response: "Well no, AC synchronous motors were in widespread use for decades before GE started making them available, and they were truly "sensorless" because they had no controller at all to make sensors available to!"

What does this have to do with the nomenclature of PMS motors?

If you survey the history of the development of motor theory you will find that GE has been the source of it in the USA. Steinmetz worked for GE. So did Gabriel Kron. The final refinement of motor theory was made by Paul Krause of Purdue who also came from GE. So did Alan Plunkett who contributed to motor drives.

"whether the manufacturers of motors designed to be driven similarly had in fact chosen a nomenclature that helps the potential purchaser identify the correct "style" of BLDC that would best suit his drive scheme."

What I was trying to say is that motor manufacturers do not always recommend the best ways to drive their motors. It depends on your motion control problem definition.

Example: On one project, Jim Bausch (formerly of HP) and I had the goal of maximizing the motion performance (max torque and acceleration, accurate positioning and holding without jitter) of size 34 step motors. The step-motor makers had various stepping schemes, but none of them would have even come close to max performance. We drove the motors with a field-oriented control scheme to achieve it. None of this kind of control is found in step-motor literature; you have to go to GE-style vector control literature to find it - the kind of control usually applied to induction motors. Yet it was optimal for driving these hybrid step-motors, which are essentially PMS motors with about a 5 % variable-reluctance torque component.

"...  it's actually the very flexibility of a system like that which frustrates me about the current BLDC industry."

Well, keep in mind that the motor-drive industry advances at about a third the speed of the semiconductor industry. You're thinking ahead of where most of them are!

" I can't understand why customers would want to put up with this kind of tyranny."

Non-motor-drive people want a simple solution to their motion problem. The tradeoff is that they have to settle for somebody else's idea of the right or best solution, even though the designer does not know what the customer of the package deal is going to need. In this business, unless the motion system is designed for a particular application with narrow parameters, it is best to "roll your own" solution.

" I don't recall EVER seeing an induced voltage waveform coming out of a motor coil made exclusively out of width-modulated pulses ..."

The PWM waveform average is what is essentially the waveform that drives the motor, and for max torque, it will conform to the induced-voltage waveform. (This is fairly basic theory. I'll elaborate if you want.)

50%
50%
JeffL_2
JeffL_2
1/26/2014 10:18:07 PM
User Rank
Newbie
Re: Need info on BLDC types
D Feucht,

"This is not a subject that can be mastered with a weekend of study."


Funny, I don't remember anything being said that the ground rules here were that unless I post my entire CV for your perusal ahead of time, that entitles you to assume that I'm a complete moron. To what do you attribute your infinite superiority to me, sir, are you a member of that elite group known as academics? I recall from my own education that it's primarily people in that profession who are most susceptible to the exhibition of rampant uncontrollable narcissism. (I apologize if I accidentally elevated you to a standing you're not readily entitled to, it's clear from context that you're not inclined to extend the same to me, but it's not as if I care or anything, given the aforementioned situation what on earth was I expecting?)

"GE uses the best expression for these kinds of motors: permanent magnet synchronous motors (PMS motors)."


Well no, AC synchronous motors were in widespread use for decades before GE started making them available, and they were truly "sensorless" because they had no controller at all to make sensors available to!

"What a motor is and how it is driven are independent considerations." 

Unless if course the manufacturer of said device should actually want to sell large quantities of said devices, in which case he might want to place his design in whichever light makes that device look the most desirable. In pursuit of such representation that vendor might want to specify what method of electrical drive configuration tends to minimize the torque ripple under specified conditions, which is presumably that method which the motor was originally designed to operate with. It's not at all clear that if cost isn't the primary driver, that the OEM purchasing these devices wouldn't seek to recommend driving such a motor in the way that would in fact minimize such torque ripple, as NOT doing so frequently results in introducing spurious mechanical vibrations and other "forcing functions" into the subsystem. All I was asking about was, since the industry by most criteria is a fairly mature one, whether the manufacturers of motors designed to be driven similarly had in fact chosen a nomenclature that helps the potential purchaser identify the correct "style" of BLDC that would best suit his drive scheme. I guess by asking such a question I must have inadvertantly put my entire professional reputation at stake, but I can't for the life of me figure out why.

In a previous career we designed servo systems in which performance was key and energy consumption was fairly unimportant. In such a scenario we had a strong preference to drive any kind of motor coil from a transimpedance power amplifier with lots of available compliance voltage. In such a configuration the current can be made to change in a very short time interval, and the effect of the specific coil inductance can be almost completely removed. Whether or not such a system is practical for other purposes is a bit irrelevant, it's actually the very flexibility of a system like that which frustrates me about the current BLDC industry. I suppose the vast majority of these motors are used in applications that run ONLY at constant velocity, occasionally you'll see one where the velocity can be "tweaked" (frequently with a pot NOT a control voltage) but not over a very wide range. If one indeed DID contemplate a true "dual-mode" (velocity/position) servo application using a BLDC, in most torque ranges there are either no choices at all or the cost is highly prohibitive, and just as frequently the vendor WON'T just bring out the coil leads and let you "roll your own", I can't understand why customers would want to put up with this kind of tyranny. The other advantage of such a system is you can design it to use TRUE velocity profiling, rather than elect to pursue a severe compromise like "microstepping" as would be the likely case if one elected to use a PMS as a stepper motor in order to use this type of motor in a position mode servo the way you suggest.

"This is achieved when the drive waveform is the same waveshape as the induced voltage." 


Again, funny that I don't recall EVER seeing an induced voltage waveform coming out of a motor coil made exclusively out of width-modulated pulses like the arrangement you are proposing, but if you say it exists, then I suppose it does, after all YOU'RE THE EXPERT...


50%
50%
D Feucht
D Feucht
1/25/2014 12:15:48 AM
User Rank
Blogger
Re: Need info on BLDC types
Jeff,

I don't know if I can resolve your dilemma, but when I first entered the motor world, I found much of it confusing too. Part of it is the terminology. GE uses the best expression for these kinds of motors: permanent magnet synchronous motors (PMS motors). BLDC is popular but it is better to obsolete the expressions ac and dc because they are ambiguous and sometimes even incoherent. (Is a dc voltage a current or a voltage?) And a homopolar motor is also brushless but is not a PMS motor.  An abhorrent term still used is "back EMF". This quantity is better called the "induced voltage" in the stator winding. It is not a force but a voltage. Paul Krause put the finishing touches on motor theory and wisely chose good language for this quantity: induced voltage.

What a motor is and how it is driven are independent considerations. A PMS motor can be driven to behave as a step motor and a step motor can be used as a PMS motor. The motor-drive designer can choose how to drive the motor based on its properties. The induced voltage waveform will lead the drive designer to choose a set of drive waveofmrs that optimize desired criteria - usually maximum power transfer with minimum torque ripple as a function of rotor angle. This is achieved when the drive waveform is the same waveshape as the induced voltage. These waveforms are generated efficiently by using switching amplifiers, and there are various alternative ways of designing switch sequencing (or "commutation"), as you have discovered.

The design of a motor drive involves basically two considerations: torque magnitude control and phase control. The 2 or 3 phase-windings of the motor can be transformed into two dimensions, as a vector with magnitude and phase which rotates with the rotor. The goal of the stator drive is like the carrot and donkey: to keep the phase of the stator drive vector ahead of the rotor magnetic vector, ideally by 90 degrees. This maximizes torque and if the phase difference is held constant, the torque is constant (no ripple => no vibration).

The number of pole-pairs of a motor affect the slope of its torque-speed curve. When designing a motor-drive, you will need to know what the mechanical output requirement is, and this affects choice of motor which is another way of saying that it affects what the torque-speed curve is.

Motion control is a world of its own and has many details. I recommend that you obtain a copy of the book Electromachanical Motion Devices by Paul Krause and Oleg Wasynczuk (McGraw-Hill, 1989) and start reading. This is not a subject that can be mastered with a weekend of study.

50%
50%
JeffL_2
JeffL_2
1/24/2014 9:38:49 AM
User Rank
Newbie
Re: Need info on BLDC types
For crying out loud, RTFQ!


Let me ask this a different way, maybe there's still hope. Let's say I run into a whole pile of these BD6922FV chips cheap and I want to use them with motors. It would be a bad idea to try and use them with motors that need multiple PWMs because they don't support that option easily. Does anyone know if there an intelligent way to find motors that have the appropriate arrangement of poles etc. that is at a higher level than saying "I need a BLDC motor that will run with a Rohm BD6922FV chip" which is something practically nobody will say in their data sheet because that term is ridiculously specific and nobody ever heard of the damn part? Am I looking for a PMSM with this kind of poles or a BLDC with that kind?

It seems as soon as I ask anything about "how do I drive an X type of motor" for some idiotic reason all anyone wants to do is tell me "look at the data sheet for motor X and use their recommended circuit". That's not the level I'm approaching the problem from. For these purposes I'm a system engineer, I want to know how the industry categorizes these motors, type X wants sinusoidal commutation, type Y is designed for trapezoidal commutation, type Z doesn't need PWM commutation at all. If I have these categories and there's "motor geometry" names for them then if I have an application and my vendor for type X motors goes out of business then I just look for another vendor of type X motors (and oh yeah, I specify max run voltage, max RPM, winding inductance, all that other stuff too). The way everyone who's responded on this if my vendor fails me and I need a new vendor, I have to start over and design a whole new subsystem because "the specific datasheet changed". That's just no way to do system engineering! Do you REALLY mean if a motor vendor comes out with a new motor the silicon vendor has to design a whole new chip for JUST THAT motor "because the data sheet changed"? I don't think so!

50%
50%
Page 1 / 3   >   >>
More Blogs from Brad Albing
Jobs and where they are is always something we wonder and worry about as engineers.
Researchers at imec are working on new fabrication methods to add germanium and tin layers to a silicon substrate to get a very high speed p-channel FET.
As an analog design engineer, you're used to designing sensor interface circuitry. But sometimes it's easier just to buy sensors as a complete assembly.
Highly integrated analog ICs are perfect for some applications and can be real money-savers. But sometimes, it works out better to pursue your design in a non-integrated manner.
flash poll
educational resources
 
follow Planet Analog on Twitter
Planet Analog Twitter Feed
like us on facebook
our partners
Planet Analog
About Us     Contact Us     Help     Register     Twitter     Facebook     RSS