BLDC motor drive is too simple ..Commutation and PWM needed?

[treez]

Hello,
I would like to do a 3 phase BLDC motor drive (voltage source inverter) for a water pump, about 300W.

Can I simply just commutate the coils from one to the other, etc, with the igbts, and avoid high frequency PWM current control of the current in each coil whilst it's switched in?

-its interesting, I know that nobody will answer this, there is so much secrecy surrounding motor drives that its unbelievable. There is no way of answering this question from the web.

 

[KlausST]

Hi,

to answer your question: I don´t think it´s possible without pwm.
The BLDC acts like a sychronous motor, therefore yu can´t apply a fixed frequency at it´s input and the motor runs.
It needs to ramp up the frequency during start up. But with an BLDC and ramping up frequency it also needs to ramp up voltage.
Voltage and frequency are about linearely dependent.
So with lower frequency you need lower voltage. Even if you don´t use sinusoidal waveform you need to "step down" your supply voltage.
This is why PWM is used. It may be controlled or regulated, often it is combined with the BLDC built in hall sensor signals.
A cheap and good solution.

********
Our current project:
For a measurement equipment we need a motor with exact frequency and precision radial runout (i hope this is the correct translation for constant d(phi)/dt ).
We need to run with low frequency (here a stepper has benefits) and with high frequency (too much for a stepper).
We tried purchased control boards, but were not satisfied with the results, especially at low frequencies.
Therefore we developed our own control board. Our solution uses true sine (filtered pwm) with controlled frequency and regulated voltage.
We do not need the hall sensors for regulation, but we currently use them for stall detection.
Our solution is no low cost solution, but it´s built for high precision, high efficiency and low EMC.

 

[treez]

Thanks.
I also cannot see how it works without the high frequency PWM (about 20khZ).
however, I know of a company who are doing it without PWM....At least in a prototype.
The DC link capacitor is the output capacitor of a buckboost converter.
I cannot understand what, for example, happens if it starts up into a locked rotor, and there's no PWM to limit the motor coil current...surely there will be massive resonance between the "Locked" rotor coil and the DC bus capacitor, and there will likely be a resonant overvoltage on the DC link capacitor?

 

[KlausST]

Hi,

The DC link capacitor is the output capacitor of a buckboost converter.

...then the PWM of the buckboost is used.

It also works fine when the buckboost acts like a current source or has built in current limiting.

Hope that helps
Klaus

 

[treez]

how can you do current limiting from the output of the dc link capacitor to the motor coil.....you can not...there is too much delay..it would resonate.
You are maybe talking about a "current source inverter", that is totally different.

You cannot have instant current turn off unless you treat the motor coil like a buck inductor and run it like that.
If you have an output capacitor of a buckboost and you try to control the motor current from the buckboost converter, then you cannot stop current from going from that output capacitor into the motor coil by turning off the fets of the buckboost converter....the energy is already in the buckboost output capacitor and nothing can stop it going into the motor coil...that's why you cannot do it like that.
You have to pwm the motor coil so as to stop the motor coil and dc link cap from resonating with each other...especially at start up.

 

[KlausST]

Hi,

maybe a misunderstanding. Can you show a picture of what you mean?

What i meant:
I switch mode DC supply ( some 10 kHz) with current limit to supply the FETs.
The FET switches on, the current in the motor coil increases and so the current in the SMPS....until the current limit of the smps gets active and drops the output voltage of the SMPS to the FETs.
In the same time the voltage to the motor is dropped and the current to the motor coil is limited.
In this way you can supply all three halfbridges with one currrent limited supply...

For sure this must be well chosen in the size of capacitors and regulation timing to avoid resonate.
The frequency for a SMPS is some 10kHz and the commutation frequency are some 10Hz, far away, so it is a good chance to get it stable.
On the other hand it does not need to be very fast (some 10 ms), no precsion and short time overshot is no problem also. The most important thing is not to fry the motor coils.

I´m back in some hours..
Klaus

 

[treez]

Yes here is schematic (basic)

 

[KlausST]

Hi,

Your picture is pretty much what i described.
Are you sure about the motor inductance?
I expext it rather at 56uH or 560uH.

If you have the circuit by hand then you can measure with a voltmeter that with low speed the voltage across the capacitor is low and with higher frequency the voltage becomes higher too.
This is one way to get an almost constant torque. But also about constant power loss (heating) in the motor.

Klaus

 

[treez]

The only way that schem could be workable is if the current loop was at least 10 times greater in bandwidth than the LC resonant frequency of the dc link cap and the motor coil.
-that is unreasonable.
The schematic you see is surely totally bogus?

The only way to do a voltage source inverter is to use the motor coil as a kind of buck converter inductor and PWM it like that.
Note that the entirety of the internet contains no info on this as it is an industrial secret to all those who are in the know.

I think you would agree that the question I ask here is a simple and basic one, but due to industrial secrecy, will not be answered, and articles on this subject do not exist.

Surely figure 1, page 2 of the following is the only way to drive a bldc with a voltage source inverter
https://ww1.microchip.com/downloads/en/AppNotes/00899a.pdf

....that sense resistor means the igbt gets shut off the instant that the coil current goes above the limit value.

 

[KlausST]

Hi

I think it could work this way.
The resonant gets damped by: DC coil resistance, drawn mechanical power, and the supply itself and ...
You can add an extra R C damping circuit...

I see it more as a current source inverter than a voltage source inverter....

Maybe you can try the circuit, i think it's worth it.Tell us your experience.
We will make it run...

Good luck
Klaus

 

[treez]

sorry but no, the motor coil resistance is just 16 milliohms...that is not going to have much of a damping effect. If you add an rc damper then that will be very dissipative, and not worth it.
Admittedly when the motor has gotten up to speed, then there is an effective resistance in the motor coil, due to the "work" being done by the motor....but what about when the motor starts, and is stationary or very slowly rotating, -at this point, the motor coil looks like a pure inductor, and this will resonate like mad with the dc link capacitor...especially in the case of locked rotor.

I tell that the schematic in page 2 figure 1 off the above microchip reference is surely the only way to do it.....the current limit resistor there is a fast current limit, and is the only way to do it surely.....

Needless to say, the microchip reference does not reveal why the figure 1 does it like that...its an industrial secret.

 

[KlausST]

Hi,

For sure all depends on the paramerers.
I currently work with a BLDC motor with 1mH and 350mOhms.
Resonant frequency is much lower, damping is more effective...

Why don' t you try it? Come on

Klaus

 

[treez]

with such high inductance, that must be low power?
I cannot try it, I don't work there any more. I only did the power supply. They wanted someone to do the controller but I didn't agree with the way they were doing it......so I got thepower supply working as a power suppy, but did not make it a current controlled power supply as in the post of #7 above.

Surely you agree that with a "voltage source inverter", there are two PWM's...one is the lower frequency commutation PWM, and the other is the higher frequency current controlling PWM....itmust always be like this...surely you agree.
There is only one other way to drive a BLDC, and that is with a current source inverter...and that is where there is an inductor in place of the dc link capacitor, and virtually no dc link capacitance whatsoever....and the current in that inductor is very tightly controlled.

 

[KlausST]

Hi,

[QUOTEwith such high inductance, that must be low power?][/QUOTE]

300W motor.

Klaus

 

[treez]

As you know, since you have 1mH, then you will have high flux density in your motor iron, and surely it will thus heat up with hysteresis losses due to the high frequency pwm?..or are you not using high frequency pwm, -maybe you are just pwm'ing at the commutation frequency only.?
I would have thought the current rise time in a 1mH motor was very high, or is your bus voltage very high?

 

[KlausST]

Hi,

flux depends on windings, current and geometry...

hysteresis loss depends on flux and core material and frequency..

Heat up depends on current and winding resistance...

current rise time is lower with increased inductivity..
(edit: sorry, this is not correct. correct is dI/dt is lower, resulting in higher rise time)

I use true sine input to the motor. generated with 150kHz PWM and second order filter

too bad that you can´t try the circuit, I´d be interested in the results.

Klaus

 

[treez]

The thing is, if it was possible to drive a BLDC with just commutation pwm, and not high frequency pwm aswell, then everybody would do it like that. -Nobody would ever use high frequency pwm.
This is proof that the method I describe is bogus.