Variable frequency drive design

[ksstms]

Hi!

Storytime!
I'm designing a single phase variable frequency drive circuit for hobby purposes. I have attached the schematic of the bridge circuit.
The final goal is to drive small/medium-sized 230V motors, so V+ would be at around 320V. For testing I have first used 12V on the high side, and was very happy to see that the output waveform is as expected. (Attachment #2  Unfortunately I don't have a proper scope yet, but I was able to get a NI myDAQ for free, so that's what I can use now.  )
After that I increased the voltage to 30V, and suddenly everything got messy. I thought that the problem is around the bootstrap caps, so I started experimenting with it. (Attachment #3: bootstrap cap voltage)  They were 10uF at first, so I've tried going higher, but nothing changed too much. Smaller caps have made it a bit better on higher frequencies.

My microcontroller program is quite primitive, because I thought that would be enough. One half of the sinewave is divided into 25 parts (PWM), and those are divided into 50 more parts. I'm using a lookup table to see how many of these 50 iterations is the output high, and then in the rest of them it's low. Then this is repeated with the other pair of IGBTs for the other half of the wave. There is a safety delay when switching to the other half wave. The PWM value is not 100% at the peaks. The whole wave is scaled down to 90%. The variable frequency of the sinewave is achieved by increasing/decreasing the length of the PWM iterations.

Questions
First of all: how should my program look like? Should I use a fixed PWM frequency instead of this solution?
What size should the bootstrap capacitors be? I've seen that bigger isn't better, but how does it depend on the switching frequency?
I have seen some schematics on the internet, where the gates were pulled down to the emitter with 1K resistors, and there were diodes parallel with the gate resistors to speed up gate discharge. I tried these, but nothing changed. Should I use them?

Thanks in advance.

[Benta]

I'm designing a single phase variable frequency drive circuit for hobby purposes. I have attached the schematic of the bridge circuit.
The final goal is to drive small/medium-sized 230V motors, so V+ would be at around 320V.

Before going into the nitty-gritty: are you talking about 230 V induction motors?
If yes, these are not suitable for VFD control. They are actually three-phase motors with run capacitor (and sometimes start capacitor) to simulate the other two phases, and this of course is frequency dependent (50/60 Hz).

Apart from that, you'll need to look at your drive strategy. Normally, the upper power transistors are switched statically (meaning no PWM, but just according to bridge polarity), and the lower diagonally opposed transistor is PWM'ed.

Benta.

[station240]

Your going to need a snubber network to protect the IGBTs, and some form of output filter to smooth out the spikes before it goes to the motor.
Add those parts then seen what waveforms you get.

[BradC]

I'm designing a single phase variable frequency drive circuit for hobby purposes. I have attached the schematic of the bridge circuit.
The final goal is to drive small/medium-sized 230V motors, so V+ would be at around 320V.

Before going into the nitty-gritty: are you talking about 230 V induction motors?
If yes, these are not suitable for VFD control. They are actually three-phase motors with run capacitor (and sometimes start capacitor) to simulate the other two phases, and this of course is frequency dependent (50/60 Hz)

Better not tell my pool pump or drill press. Both of those are permanent split capacitor motors (single phase with a run cap) and run on vfds. Most single phase motors are two phase, with the second phase somewhere in the order of 90deg out. You won't get as much torque as you slow them down, and you need to be careful if your motor has a centrifugal starter not to drop to or below the speed that engages the starter, but they will work on a vfd.

There are other gotchas, but as a leaning exercise its all good. I run mine on 3 phase vfds, just with active and neutral across two phase outputs on the vfd. Not ideal, but works well.

[ksstms]

Thanks for your input guys!

Before going into the nitty-gritty: are you talking about 230 V induction motors?

Well, the voltage certainly is 230V, but I don't know much about AC motor design. I have thought that brushed and brushless are the main categories here. I have a small but powerful fan that I got from ebay, and it is a brushless one. I have planned to build this vfd for that, because I read that those cheap little triac "dimmers" work only with brushed motors.

and you need to be careful if your motor has a centrifugal starter not to drop to or below the speed that engages the starter, but they will work on a vfd.

A friend of mine lent me a 3-phase inverter to try the fan, and it works flawlessly down to about 6 Hz. 
The switching frequency of this Commander SK inverter is 3 kHz by default, so if I go with fixed frequency pwm, I'm planning to use about that.
EDIT: OFF: Does inverter mean the same as vfd? Or is it just a local thing to call it that way?

Your going to need a snubber network to protect the IGBTs, and some form of output filter to smooth out the spikes before it goes to the motor.
Add those parts then seen what waveforms you get.

I'm planning to work on this in the weekend, so I'm collecting info until that.

[Siwastaja]

Before going into the nitty-gritty: are you talking about 230 V induction motors?
If yes, these are not suitable for VFD control.

Wrong - VFDs for single-phase capacitor-run motors exist on the market. It's not optimal, but it's done all the time for fan, pump etc. control when TRIAC control is not satisfactory and motor replacement with a three-phase machine is not possible.

Another option is to control single phase capacitor-run motor as a two-phase motor with a special two-phase VFD (without the capacitor). I have done this personally.

They are actually three-phase motors with run capacitor (and sometimes start capacitor) to simulate the other two phases, and this of course is frequency dependent (50/60 Hz).

Again, wrong - they are actually two-phase motors. Not saying such a rarity couldn't exist, but I have never heard of a three phase capacitor run motor. It would have two different run caps (or two sets of run/start caps).

[jpittner]

Hi,
look at my design, it's open hardware and open source:
http://www.pittnerovi.com/jiri/hobby/electronics/frequency_changer
However, if you need somethink quickly, get a 3phase vfd from aliexpress
and use 2 phases to drive the motor, development of power circuits
is more tricky than it seems.
jiri

[albert22]

Again, wrong - they are actually two-phase motors. Not saying such a rarity couldn't exist, but I have never heard of a three phase capacitor run motor. It would have two different run caps (or two sets of run/start caps).

+1
they may have two capacitors but still they are 2 phase.

[Benta]

Major brain fart on my side: two-phase of course!
Dunno what I was thinking about.

Well, the voltage certainly is 230V, but I don't know much about AC motor design. I have thought that brushed and brushless are the main categories here.

Now you've got me confused. At 230 VAC, you'd normally see asynchronous/capacitor motors. Brushed would normally be DC, but rare at 300 V unless large. Brushless is a different story altogether.

Which kind of motor do you really want to control?

[ksstms]

Brushed would normally be DC, but rare at 300 V unless large. Brushless is a different story altogether.

I had to look up the correct english term. They are called universal motors, because they can operate with both DC and AC. These are found in many household appliances, and tools like drills and such. They can be easily slowed down by chopping the input sinewave with a triac.
By brushless I meant all other mains AC motors that are not universal motors.
I read, that 'non-universal' motors are not suitable for triac control, but they can be slowed down by vfd. And that is what I intend to do here.
And sorry, I forgot to reply to your earlier post:

Normally, the upper power transistors are switched statically (meaning no PWM, but just according to bridge polarity), and the lower diagonally opposed transistor is PWM'ed.

If the upper transistors were P channel fets, then it would be okay. But for that you would have to 'make' a voltage about 10-15-whatever Volts lower than the positive high voltage supply. It's easier to use all N-channel ones along with high side driver, which charges up a bootstrap cap to the gate's required driving voltage, and connects it to the gate and source (emitter in the case of igbt) when needed. Here is a really nice explanation about it. Of course, you can't drive the gate for too long, because the bootstrap cap's voltage will drop, and then you have to charge it up again. Charging is done in the OFF phase of the pwm signal, and that is why the duty cycles are scaled down to 90%, so there is some charging time left.


jpittner:

look at my design, it's open hardware and open source:

Thanks. I'm going to have a look for sure 

[Benta]

OK, I'm beginning to understand. Universal motors are all over the place, and well suited to triac control. But that's not what you want.

By brushless I meant all other mains AC motors that are not universal motors.

In this case, you'll run into:
Shaded-pole motors: normally very small, like for really old phono turntables and tape recorders. Rare nowadays.
3-phase, shorted rotor induction motors: you see them all over the place, starting from 500 W and up.
Single-phase induction motors with run (and sometimes start) capacitors. I think these are what you are thinking of.

Yes, you can to a certain extent control single-phase motors with a VFD, but only in a limited range.
And the torque characteristics are abysmal. Personally, I'd go for a different motor type.

Concerning the bootstrap, it's not a big issue in practice. Should one of the caps run out of  charge, then it'll be ready again at the next direction change. The motor is much slower than the electronics.

[tatus1969]

i would not recommend using irs2101 because
- it does not have shoot through protection, an error in your code can cause transistors at fire
- they (the whole family) dont like switch node ringing beyond -5v, the high side logic misses control pulses then. if that is the "turn off" pulse you can imagine what happens. 1m long blue-green flames in my case

[ksstms]

Concerning the bootstrap, it's not a big issue in practice. Should one of the caps run out of  charge, then it'll be ready again at the next direction change.

As I have written in my first post, I have cap charging problems.
First of all, I'm going to change the drivers' voltage to 15V. And I should rewrite my program to use a fixed pwm frequency to generate all the waves.

i would not recommend using irs2101

Well, that is bad news. Can you recommend other high side drivers? (high+low side would be preferred)

[tatus1969]

Well, that is bad news. Can you recommend other high side drivers? (high+low side would be preferred)

here's a snap of a HV driver of mine, works fine up to 100V / 40A. It solves some more issues by controlling switch slew rate, and negative bias to prevent parasitic turn on from Miller capacitance.

[Siwastaja]

Now you've got me confused. At 230 VAC, you'd normally see asynchronous/capacitor motors. Brushed would normally be DC, but rare at 300 V unless large. Brushless is a different story altogether.

Brushed motors are VERY widely used for 230VAC - almost every 230V hand power tool has one; so do all vacuum cleaners. Because they are series wound motors, the polarity in both stator and rotor changes at the same time, so the motor produces torque in the same direction regardless of the input polarity - hence they work with both DC and AC.

The reason is that these products need high speed and/or large optimum speed/torque curve sweetspot that 50Hz fixed frequency mains powered motors - synchronous or asynchronous - cannot simply provide; or would be too heavy and big. Brushed motor is the kinda-old-school "VFD" - the windings are provided with variable frequency that automatically scales with the rotor speed, with much higher upper limit than 50Hz, allowing smaller motor design.

[nowlan]

Was under the impression 3 phase vfd would fault if used on single phase.

Ive used an invertek single phase vfd in the past.
http://www.invertekdrives.com/variable-speed-drives/optidrive-e2-single-phase/

[ksstms]

here's a snap of a HV driver of mine

Thank you! It's truly beautiful 
Could you give some info about how you generate the PWM_U signal? What frequency do you use?

[tatus1969]

here's a snap of a HV driver of mine

Thank you! It's truly beautiful 
Could you give some info about how you generate the PWM_U signal? What frequency do you use?

i am using an STM32 for that. frequency is adjustable and depends on motor inductance, the lower the higher. currently it is at 20khz. if i can afford it in terms of efficiency i dont go below that to not let the pwm become audible. the power stage is deliberately made slower than possible to reach a tradeoff between efficiency and emi/ringing/speed. but as you see both dead time and speed are adjustable, for the latter you vary the capacitors from gate to drain. it is tuned here for comparable conduction and switching losses.

[bktemp]

i am using an STM32 for that. frequency is adjustable and depends on motor inductance, the lower the higher. currently it is at 20khz. if i can afford it in terms of efficiency i dont go below that to not let the pwm become audible.

The frequency is often a compromise between current ripple, switching losses, lossed due to eddy currents in the motor and avoidance of audible frequencies.
Most industrial variable frequency drives operate at 1-10kHz. Around 5kHz is the best frequency for most motors (reasonable current ripple, low losses), but it is in the audible range.
I would keep the frequency rather low or add a filter if the motor isn't rated for use with a VFD, because the ringing due to the square wave drive signal puts a high stress to the insulation of the windings.

[tatus1969]

i am using an STM32 for that. frequency is adjustable and depends on motor inductance, the lower the higher. currently it is at 20khz. if i can afford it in terms of efficiency i dont go below that to not let the pwm become audible.

The frequency is often a compromise between current ripple, switching losses, lossed due to eddy currents in the motor and avoidance of audible frequencies.
Most industrial variable frequency drives operate at 1-10kHz. Around 5kHz is the best frequency for most motors (reasonable current ripple, low losses), but it is in the audible range.
I would keep the frequency rather low or add a filter if the motor isn't rated for use with a VFD, because the ringing due to the square wave drive signal puts a high stress to the insulation of the windings.

I think I have to add that I am using the design in a PMSM servo application with FOC. Motor inductances are lower there, and the phase current control loop is already at 5kHz. But of course the power stage is universal.

[Theobald]

Questions
First of all: how should my program look like? Should I use a fixed PWM frequency instead of this solution?
What size should the bootstrap capacitors be? I've seen that bigger isn't better, but how does it depend on the switching frequency?
I have seen some schematics on the internet, where the gates were pulled down to the emitter with 1K resistors, and there were diodes parallel with the gate resistors to speed up gate discharge. I tried these, but nothing changed. Should I use them?

Thanks in advance.

how should my program look like
Many microcontrollers have Motor control unit with complementary PWMs and dead times. Just connect H1+L2 and H2+L1 to the complementary outputs. See also their websites for software examples. Eventually add high speed optocouplers if you want to debug a live ground design.

What size should the bootstrap capacitors be?  From 0.47µ to 1µF. It depends on the switching frequency and Cgate,  Google the formula. If they're too small, the top side voltage Vgate is too low for a good switching.

Pull down resistor on gates: why not, but 15V zener diodes are better to keep Vgate below breakdown voltage, just in case...

Diode in parallel Right, this is a trick to adjust the rise / fall times.

Now you can find high voltage MosFet, much better than IGBTs for small motors: 0.25ohms RdsOn @ 1A= 0.25W, Vcesat = 2V @ 1A = 2W !!! See STP20NM60FD or equivalent in DPAK package.

@bktemp +1: You can drive capacitor run motors with this converter, but take care of the high slewrate on this kind of driver. 100ns typ, with a 600V swing, it's a motor killer ! The standard motors are made for 230Vrms 50Hz sinus, the high slewrate destroys the insulation of the first turns of the coil very quickly. See IEC Technical spec TS60034-25.

Theo

[ksstms]

I would keep the frequency rather low or add a filter if the motor isn't rated for use with a VFD, because the ringing due to the square wave drive signal puts a high stress to the insulation of the windings.

Like an RC or LC filter at around the PWM frequency? Guess I'm googling the wrong things, because I haven't found too much info about it.

Now you can find high voltage MosFet, much better than IGBTs for small motors: 0.25ohms RdsOn @ 1A= 0.25W, Vcesat = 2V @ 1A = 2W !!!

That's a very useful info! I didn't think about that.


Also, my IR2101's are officially dead (along with an igbt). So now I must go with a better solution, and I've ordered the stuff for tatus' driver design.

[tatus1969]

there are two types of filters, both of lc type: emi filters that only reduce the dv/dt and ringing, and sine wave filters like this http://www.schaffner.com/products/power-magnetics/sine-wave-filters/

and you can try rc snubbers to dampen ringing. i suggest to build it up and see how it goes. the cabling also matters, as its inductance and capacitance are part of the game as well as those of the motor.