Current output regulated Buckboost converter feeds BLDC Motor and its inverter drive

[treez]

thanks, though how high must the bandwidth of the buckboost be in order for it to be able to prevent resonance problems caused by the motor coil inductance (56uH) Ringing with the output capacitance of the buckboost (300uF)...as you know, this resonance will be excited every time the dead time ends (every 2.5ms)

 

[treez]

FvM

The answer to the resonance problem is, if the voltage feedback loop is fast enough, it will damp respectively cancel the resonance.

...this is just the problem...our feedback bandwidth of our buckboost converter would need to be 15KHz in order to be able to avoid the resonance problem of the motor coil and the DC link capacitor.......(the DC link capacitor is actually the buckboost output capacitor)

The motor coil inductance is 56uH
The DC link capacitor is 200uF
This has a resonant period of 665us

The commutation on time interval of each coil of the 3 phase motor is 2.5ms

...there are several LC resonant periods inside that 2.5ms.

This is an LC oscillator, not a motor driver....we would need the buckboost feedback loop bandwidth to be 15KHz to get over this problem..................not likely to be practically doable....at least not in a fast timeframe.
Can you confirm to us that out entire method here, of having an upstream buckboost converter to regulate the speed of a BLDC is entirely bogus...or at least..... having a touch of sci-fi to it?

As mentioned, we are simply commutating the BLDC at maximum duty cycle all the time, and regulating the speed with the upstream buckboost....which limits the current to the inverter/BLDC

 

[FvM]

Can you confirm to us that out entire method here, of having an upstream buckboost converter to regulate the speed of a BLDC is entirely bogus...or at least..... having a touch of sci-fi to it?

Not at all. The topology can be easily implemented with an appropriate DC link capacitor. I guess that it can still work wit the small capacitor you are providing, but that's something you have to test in a simulation or better in a real setup. As already mentioned, also the buck-boost converter dimensioning (e.g. switching frequency) might need adjustment.

 

[treez]

we cannot increase f(sw) as we have little heatsinking room.
We don't have room for any more than 200uF dc link capacitance (= the buckboost output capacitance)
You appear to agree that we have just set up little more than an LC oscillator with our present setup?
If not, then that would mean that the motor is somehow stopping the motor coil from acting like the inductor that it is, and this , surely, is not possible?

 

[FvM]

Did you try to operate the buck-boost converter with voltage feedback or are you just guessing?

 

[treez]

its operated with current feedback...that is, the current out of the output cap of the buckboost, going to the inverter, is monitored and fed to a current error amplifier....the reference to this current error amplifier is the motor speed error voltage....so the lower the motor speed is below required speed, the lower is the reference to the current error amplifier, and more current is called for.

 

[FvM]

its operated with current feedback...

Yes, that's what I read from your block diagram. And it's well understandable why constant current operation causes large voltage overshoot during moto current "gaps" when the DClink capacitor is small.

My suggestion is to operate the converter - at least for test - with pure voltage feedback to find out if the loop is fast enough to limit the converter output voltage to reasonable values. The current sense can be still used to limit the inrush current during motor start.

Intuitively, even if you can't avoid considerable voltage ripple, it can only get better than in constant current mode.

 

[mtwieg]

I'm not sure I understand the concern with "resonance" formed between the DC link capacitor and the motor inductance. Keep in mind the electrical inductance measured by an LCR meter will probably not be equal to the equivalent electrical inductance when the motor is actually spinning (since the measured inductance will depend on rotor position). What do you think your actual current and voltage waveforms are going to look like due to that LC? Have you tried simulating it at all?

 

[treez]

-as you can imagine, current is very peaky to the motor.....its an LC and the resonant period of it is well within the commutation on-time.....so if we have 22 A rms to the motor, then peak current is above 30 Amps , as we have seen in the lab.
I suppose we can make the buckboost current source feedback loop bandwidth very slow, and just put up with the resonance, and at least that way, our input filter will not need be so big, as the commutation frequency will less be prevalent in the input current to the buckboost?.....of course, this depends on our commutation dead times being small, say around 4us, and I think we can arrange that now.

 

[FvM]

I suppose we can make the buckboost current source feedback loop bandwidth very slow.

This means that for short times the inner current mode feedback loop controls the converter, so it will still source constant current in a first order.

 

[treez]

So you mean the input (to the buckboost) will just basically deliver a constant high frequency current, at the buckboost switching frequency, and will not "see" the commutation frequency, which is relatively low, and would need a big filter to filter out?

Incidentally...
Why are simple Fixed Speed Pump Controller IC’s for BLDC pumps not available?

If we wish to pump water to our irrigation system, and we wish the BLDC pump motor to only ever spin at 8000RPM, then this is a fairly simple situation (compared to the requirement for a motor with a requirement to change from one speed to another very quickly whilst laoding might be constantly varying widely).

Why are simple BLDC motor controller ICs for fixed speed pumps (obviously fixed load aswell since they simply pump water at constant speed) not available?

Surely, all that is needed is a controller which …
1….First Aligns the motor into a known position..
2….then Slowly ramps up the bridge transistor switching frequency till the frequency necessary to give 8000RPM has been reached
3…..Then in-built back-EMF sensing could lock the motor coil pulses to the back EMF signals (sensorless control of BLDC) .

..there’d not even be any need to vary the commutation on time of the bridge transistors away from maximum at any stage, since an upstream SMPS could simply be controlled to supply whatever current the motor required to be able to spin at the required RPM?

..as such, you wouldn’t even need to use the BLDC Bridge driver to limit the current in the motor coils.

Why don’t off-the-shelf controllers, with such a simple modus operandi, exist anywhere in the world?