PMSM Motor Controlled Using FOC for HiFi Turntable Project

[dellsam34]

For a an experimentation project I'm building a very large and slim BLDC type PMSM Motor that will be controlled using Field-Oriented control, The motor is an outrunner of a size of 30 cm in diameter and 1.5 cm in thickness, the stator uses at least 120 coils, The goal of the project is to reach a near zero torque ripple and a very stable angular/linear speed.

I have few questions:
1- What is the coil # to magnetic pole # ratio ? If I'm aiming for a large number of them, like I said about 120 for the best torque ripple and speed stability.
2- What is the maximum phase # that can be driven by today"s technology? 40 phases sounds great but I know it is not realistic.
3- What should I use to determine the motor outrunner platter position? A high resolution encoder or back EMF, didn't think hall effect sensors will be good for this application but still on the table.

While I can do the mechanical design and fabrication, I have no idea on how to proceed due to the above questions, Any help is really appreciated.

[Dave]

1. There is no set ratio. Here is a reasonably nice explanation on how stators are wound.

2. Depends on how much you're willing to spend on hardware. You could implement hundreds of phases with a massive FPGA, but I don't think there would be any real benefit to it.
6-phases can be done using readily-available microcontrollers on the market (TI Delfino, Microchip SAM, STM32).

3. Sensing back EMF is fine if you don't need full control from 0 RPM.
If you do, you're going to need some sort of absolute position encoder. If you can access one end of a shaft and place a magnet on there, you can use something like this. Hall sensors are absolutely useless for this purpose.

Keep in mind that the number of pole pairs is the multiplier which tells you how the rotational frequency translates to electrical frequency. For this reason, the number of PP practically divides your effective encoder resolution (for example, with 4 pole pairs, one electrical revolution equals just 90° of shaft rotation, so just 2048 digital steps of a 13-bit encoder). Getting good resolution positional feedback is crucial for torque linearity.

[dellsam34]

Thanks for the reply Dave;

What I meant by magnetic pole to coil number ratio is if I have 120 coils, should the number of magnets will be exactly 120, less than 120 or greater? I don't think having equal numbers is good because you want the coils to be pulled or pushed at any position but I could be wrong with my assumption.

While the normal working speed is anywhere from 40 RPM to 80 RPM, I do need one function which is hold still, it basically rotates the platter until something locks and senses the sudden stoppage of the platter and holds it there.

If someone willing to go down with me on the electronic part of it, I'm willing to fully explain what this project is all about and a reward for the time spent will be given.

I can initially use off the shelf parts but at the end the stator coils and all electronic components will be mounted on a 29 cm in diameter PCB.

[Doctorandus_P]

If you have no clue about how many magnets to use for a motor with 120 coils, then you're not ready to build a motor like that.

I recommend to start much less ambitious, and make more iterations.

Do you have some old HDD's lying around? These (nearly?) always have BLDC motors if it's less then 20 years old. Heck, you can buy a 2nd hand HDD for less than a loose BLDC motor, and they've also got a completely integrated BLDC motor driver somewhere on the PCB.

Old washing machines can also be a source for pretty high powered BLDC motors. Some of these motors are regular sized, and have a multi-V belt to drive the drum. Others are "direct drive" and can have a BLDC motor with the same diameter as the drum.

On sites like Ali / Ebay / Banggood / Etc. you can buy small BLDC motors for less then EUR10. Get some experience with motors like that and get back to the bigger motors when you've learned from the small ones what you can.

About the driver electronics:
How do you want to implement FOC? Do you want some pre-built thing, or do you want to build your own hardware and firmware?

You may also want to have a look at https://hackaday.io/search?term=bldc Lot's of projects with BLDC motors and drivers.

If you like video tutorials... There is an excellent series of these (I think 4 parts) from either (or both?) Texas instuments and ST Microelectronics on youtube.

[jbb]

Can people recommend some technical  books on machine design? The design of good electrical machines is quite involved.

It sorta sounds like you want some kind of precision mechanical control? Lowish speed, no gearbox to avoid backlash, low torque ripple.

That sounds like a hell of a project for your first electrical machine design.

I don’t see why you’d need many many phases. 3 phases can - in principle - give nice constant torque. You connect coils in series to make a repeating pattern, which allows your number of phases to be less than the number of coils.

Do you need continuous rotation? Or only part of a turn? Because some kind of rotational voice coil actuator (as used to move the heads on an HDD) could be very good for smooth  movement.

If you don’t need a lot of torque, you could consider an ironless design which will have no cogging torque.

You might also consider more advanced winding schemes to provide the most even magnetic coupling. For quick and easy manufacture, each winding slot normally has only one phase in (eg 20 turns of phase A), but this doesn’t give the most even torque. There are methods to improve torque ripple by having multiple phases go through one winding slot (eg 2 turns of phase C, 16 turns of phase A, 2 turns of phase C).  This is quite complicated, but diligent engineers have worked on the problem extensively since the 19th century and someof them have written reference books.

[Cerebus]

A copy of "Brushless Permanent Magnet Motor Design" by Dr. Duane Hanselman (ISBN 1-881855-15-5) will answer all your questions and then more. Expect to (re-)learn some physics if you're not up to speed on magnetics. The examples chapter has examples of up to 32 pole motors with winding parameters, back emf harmonics etc. etc.

[WattsThat]

Holy bookworms Batman!

I was all ready to buy me a book on pm motors until I found a copy of that thing. Must have gold leaf pages. $349.00. I already flattened my book budget with all three volumes of the AOE earlier this year. Maybe I can put it my Amazon cart for Christmas!

Seriously, thank you for the reference book.

[Cerebus]

You might find it *ahem* elsewhere, if you take my drift.

[WattsThat]

Yes, quite easy if you search using the text of the title rather than the ISBN. Duoh 

Thank you. Oh boy more covid reading

[dellsam34]

Thank you guy for the replies, I'm not trying to re-invent the wheel here, I will be using the exact same technology but with different dimensions and specifications and needed the above questions answered, I can handle the mechanical design from CAD to CNC, I was just hoping that someone has the electronic skills to help me design a custom driver for the motor. Let me try to answer some of the above replies:

Do you have some old HDD's lying around?

Yes but none works for my project

I don’t see why you’d need many many phases. 3 phases can - in principle - give nice constant torque. You connect coils in series to make a repeating pattern, which allows your number of phases to be less than the number of coils.

Yes I could, I jut wasn't too sure if having more phases is equivalent to having repeated same phases, Again the goal is to have as many coils as possible to reduce the torque ripple.

Do you need continuous rotation?

Yes continuous motion, but there is a function that requires the platter to spin about half a turn and then hold its position for few seconds.

If you don’t need a lot of torque, you could consider an ironless design which will have no cogging torque.

Not a lot of torque, 12-14V application, but if this approach is used I can make up for the needed torque by having more coil windings, What materials are used for this?

[Cerebus]

This is beginning to look like a classic "XY Problem" question.

What is your actual problem? What torque do you require, what angular precision, and what is your acceptable torque ripple? If you can't put numeric answers to those then I suspect you're not far enough along in your problem analysis to even be ready to pick a solution, let alone at the point where you need to ask detailed questions about a custom motor design.

[dellsam34]

This is beginning to look like a classic "XY Problem" question.

What is your actual problem? What torque do you require, what angular precision, and what is your acceptable torque ripple? If you can't put numeric answers to those then I suspect you're not far enough along in your problem analysis to even be ready to pick a solution, let alone at the point where you need to ask detailed questions about a custom motor design.

Let me sketch this on Solid Works so we can have something to talk about, You're rights just words won't cut it.

[dellsam34]

Ok, I sketched the stator, The center hole shown is the rotor hub and bearing as well as some other mechanism that is not related to the topic.
As I mentioned above the RPM range is from 40 to 80 where 4 pre programmed fixed speeds are selected via a switch and 4 master variable resistors for speed calibration for each selection, For torque, the load is the platter weight itself which will be cast aluminum plus the Neodymium magnets bonded on its inner edge, spinning around two brass ring bearings, So not a lot of torque needed.

The picture doesn't show the mounting holes for the rotor hub and PCB, Will add them later when I finish the hub design.

So next before I sketch the rotor, will the number of magnets be the same as the coils (which is 120) or less? Some of you who are familiar with motor control should know what is the best coil to magnet combination.

[dellsam34]

MIT used more magnetic poles than coils (42 magnets and 36 coils), it is usually the opposite:

https://youtu.be/Mhxz2Bj2RXA?t=601

[dellsam34]

Seems like 36/42 is a common coil to magnet ratio, so what would be best for 120 coils? Any volunteer who can assist in designing the FOC drive for 3 phase with that many coils? Any help appreciated.
Oh, and this is a turntable record player design. that's why speed and torque ripple are important factors for the design.

[Doctorandus_P]

You probably do not want a BLDC motor, but:
https://en.wikipedia.org/wiki/Reluctance_motor

These do not get much attention, maybe because they are not "fashionable" or maybe because the power density is lower.

They do have some inherent advantages for use in a turn table, and this is the same reason why they are also used in HDD's, and that is that they have 0 torque ripple, and that is in turn because they do not have permanent magnets.

Did you also change the title of this thread?
I responded earlier, but had not noticed then it was for a record player.

There is nothing special about a motor with 120 coils, They are still just 3 phase motors, with a number of inductors in series or parallel, (or combinations of that).

I also am in doubt if it would be useful to cram in as many inductors as you can.
High mass of the rotor, combined with low low friction are probably at least as important.
After spin up, the motor itself only has to deliver enough torque to overcome friction losses and prevent rpm from decreasing.

With the right motor geometry, it can be driven by a perfect 3-phase sine wave. If the motor geometry is not perfect, then it can be compensated for by pre-distorting the 3-phase sine waves as part of the FOC algorithm.

Have you considered air bearings?
"applied science" has a nice ecplanation, and I think even a tutorial for making them:

[ConKbot]

Are there any concerns with leaking magnetic fields from the motor coupling into the record needle's coil? I.e 45rpm, 120 coils would be a 90Hz rate of the poles passing by, and higher for the motor drive frequency.  Noting that couldn't be fixed by adding more steel, but it would drive torque requirements, bearing friction, etc.

[Cerebus]

You probably do not want a BLDC motor, but:
https://en.wikipedia.org/wiki/Reluctance_motor

These do not get much attention, maybe because they are not "fashionable" or maybe because the power density is lower.

They do have some inherent advantages for use in a turn table, and this is the same reason why they are also used in HDD's, and that is that they have 0 torque ripple, and that is in turn because they do not have permanent magnets.

You really should read your own references before making pronouncements.

Reluctance motors can deliver high power density at low cost, making them attractive for many applications. Disadvantages include high torque ripple (the difference between maximum and minimum torque during one revolution) when operated at low speed, and noise due to torque ripple.

A reluctance motor is probably the worst choice of motor technology for a direct drive turntable - stepper motors are [switched] reluctance motors which should make the torque characteristics of a reluctance motor clear. Even synchronous reluctance motors have significant torque ripple.

A BLDC motor is a good choice. The Technics SL1200, the standard turntable in broadcast and recording studios for more years than I care to count, used a 12 pole brushless DC motor - in the earlier models they placed the rotor permanent magnets on the back of the turntable itself, later models opted for a more conventional BLDC motor with the turntable attached to it with screws.

[langwadt]

You probably do not want a BLDC motor, but:
https://en.wikipedia.org/wiki/Reluctance_motor

These do not get much attention, maybe because they are not "fashionable" or maybe because the power density is lower.

They do have some inherent advantages for use in a turn table, and this is the same reason why they are also used in HDD's, and that is that they have 0 torque ripple, and that is in turn because they do not have permanent magnets.

You really should read your own references before making pronouncements.

Reluctance motors can deliver high power density at low cost, making them attractive for many applications. Disadvantages include high torque ripple (the difference between maximum and minimum torque during one revolution) when operated at low speed, and noise due to torque ripple.

A reluctance motor is probably the worst choice of motor technology for a direct drive turntable - stepper motors are [switched] reluctance motors which should make the torque characteristics of a reluctance motor clear. Even synchronous reluctance motors have significant torque ripple.

A BLDC motor is a good choice. The Technics SL1200, the standard turntable in broadcast and recording studios for more years than I care to count, used a 12 pole brushless DC motor - in the earlier models they placed the rotor permanent magnets on the back of the turntable itself, later models opted for a more conventional BLDC motor with the turntable attached to it with screws.

it's older brother but, https://youtu.be/jobfx7wePh4

[Benta]

Seems like 36/42 is a common coil to magnet ratio, so what would be best for 120 coils? Any volunteer who can assist in designing the FOC drive for 3 phase with that many coils? Any help appreciated.
Oh, and this is a turntable record player design. that's why speed and torque ripple are important factors for the design.

The basic thing is to have an odd/even or even/odd relationship between rotor and stator. 2/3, 3/4, 4/5 etc. Otherwise the motor won't start.
Your example of 36/42 is exactly that: it reduces to 6/7.

I'm not sure that increasing the number of magnets will give better a result, it will probably just increase the ripple frequency.

[Doctorandus_P]

You really should read your own references before making pronouncements.

Reluctance motors can deliver high power density at low cost, making them attractive for many applications. Disadvantages include high torque ripple (the difference between maximum and minimum torque during one revolution) when operated at low speed, and noise due to torque ripple.

This may be true when you want to get lots of power out of the motor, but turn tables do not need much power to keep the disk spinning. The reluctance type motors do not have any cogging torque ripple because there are no permanent magnets in them. FOC is designed to get optimize for minimum motor current, and therefore minimum torque.

The whole stepper motor argument is also invalid. Of the 100 or so stepper motors I've pulled out of old machines over the last 30 years, only one is a variable reluctance motor, and it's easy to recognize because the rotor spins completely freely. Just take any HDD motor apart and have a good look at the spindle motor. Almost all stepper motors are of the hybrid type. (Both permanent magnet and steel in the rotor)

Then there is the difference between BLDC and PMSM motors. There is no difference. Except that the motor geometry for PMSM motors is more optimized for linearity and low torque ripple while BLDC motors are more optimized for high efficiency. Motor geometry is very important when optimizing for some parameters. I was surprised at first that the stator magnets in the SL1200 did not have offsets in their lamination's to lessen ripple. Then I realized that the same effect is is probably achieved (cheaper) by slanting the magnetic fields in the magnet ring.

On top of that I have serious doubts about the goal of this whole project. (But still am kind of curious)
You can increase pole count and diameter, and both increase torque of the motor, but this makes them harder to control. The motor diameter of the SL1200 is less than a third of the diameter of the turntable, and with these dimensions it probably still works well as a flywheel and mechanical dampener. Making the motor bigger just makes it harder to control it smoothly and I see no benefits for a bigger motor, except maybe (very maybe) for a reluctance motor, run with mA of drive current.

[dellsam34]

it's older brother but, https://youtu.be/jobfx7wePh4

I have the Technics SL-5300 and surprisingly they look very similar.

The basic thing is to have an odd/even or even/odd relationship between rotor and stator. 2/3, 3/4, 4/5 etc. Otherwise the motor won't start.
Your example of 36/42 is exactly that: it reduces to 6/7.

I'm not sure that increasing the number of magnets will give better a result, it will probably just increase the ripple frequency.

The goal is to increase the ripple frequency and reduce it with FOC control instead of trapezoidal commutation.

[Cerebus]

You really should read your own references before making pronouncements.

Reluctance motors can deliver high power density at low cost, making them attractive for many applications. Disadvantages include high torque ripple (the difference between maximum and minimum torque during one revolution) when operated at low speed, and noise due to torque ripple.

This may be true when you want to get lots of power out of the motor, but turn tables do not need much power to keep the disk spinning. The reluctance type motors do not have any cogging torque ripple because there are no permanent magnets in them. FOC is designed to get optimize for minimum motor current, and therefore minimum torque.

You categorically stated about reluctance motors that "they have 0 torque ripple". But suddenly you change tack and that becomes only under low power, according to you. In fact, as the Wikipedia article states, they have large torque ripple. A google search for "reluctance motor torque ripple reduction" come up with "About 1,260,000 results" because techniques to reduce torque ripple in reluctance motors is a 'thing'. I don't think it would be that way, somehow, if they are "low torque ripple" motors. 

Here's an illustration from "ELECTRIC MOTORS AND DRIVES Fundamentals, Types, and Applications Fourth Edition by AUSTIN HUGHES AND BILL DRURY" demonstrating how reluctance motors develop torque, the thick lines are the steady state torque curve for when the motor is running (a) with zero load (b) with a medium load and (c) with the highest load that can be applied before pull out commences. Each oscillation in torque is down to the salience of the rotor reluctance combined with the angle of the rotor and the advancing stator field. That sure looks like torque ripple to me in all three curves.



Why not just admit you mis-spoke, or even just keep quiet about it and move on - when in a hole stop digging. Don't double down by making silly statements like "FOC is designed to get optimize for minimum motor current, and therefore minimum torque" -  the point of FOC, as far as current and torque are concerned, is to try and get maximum torque for minimum current compared to other methods of motor control - even I know that. Optimizing a motor for minimum current with minimum torque is easy, you don't need any fancy control strategy, you just don't turn it on - zero current and zero torque.

Please, if a relative motor control neophyte such as myself can poke holes in what you're saying perhaps you ought to reassess how you go about making categorical and authoritative sounding statements  as if you're highly knowledgable about the subject, you're misleading people.

[dellsam34]

It's okay Cerebus, All opinions are welcome.  Anyway I came across this publication and into the first few pages my brain started to hurt.

[fmkit]

use any motor made for camera gimbal,  torque ripple canceled out by MEMS sensor feedback/PWM
here is my DIY gimbal using 40g motors(BGM2804  12N14P)
recorded footage                 
gimbal mounted on quadcopter