400Hz, three-phase power supply for aviation stuff

[bogdan2014]

Greetings,
I'm looking to build a 400Hz power supply for a piece of military or aviation equipment, some sort of gyroscope, that has a small BLDC motor.
The only information I have about it it's that it needs around 36V at 400Hz with three phases. The voltage can be squarewave.
It needs to be a project that I build, so I can't just buy a power supply for it.

I've investigated several ways of doing it and the trickiest part is the control section.
I need to drive 3 mosfets at 400Hz, two of them having the signal phase shifted by 120 and 240 degrees. I don't think I need an H-bridge (6 mosfets) for 0.8-1A.
I first thought about 555 timers, either using 3 outputs of a 558 or a 555 and a 556 for the shifted signals. I have no idea if it's capable of doing phase shifting.
The second option (which I want to avoid) is using a microcontroller with 3 PWM outputs and doing the shifting in software.
I'm not aware of any 3-phase control IC similar to those used in SMPS.

As a background, I've mostly worked with (and built) audio equipment and power supplies. I also did some microcontroller programming, but just because I had to, I'm no fan of it.
Thanks!

[Kleinstein]

One option to create a 3 phase signal would a counter going up to 6. The 4017 configured as a divide by 6 and than combining 3 inputs each to get the drive signals would be an option.
For 3 phase signal it is more like 6 MOSFETs to make 3 half bridges.

[Benta]

"military or aviation equipment?"
3-phase drive with a 555?

Good luck, my feeling is that you're overestimating your capabilities here.

[Seekonk]

Suspect you might be right if he can't even look on ebay, 39.95 and add a couple amplifiers.

[james_s]

A microcontroller with 3 PWM outputs ought to work, just drive a half bridge of mosfets with each output and use a sine lookup table in code.

Years ago I built a simple driver for a little military surplus gyro that used a two phase motor. IIRC I used an op-amp based oscillator and a simple capacitor to create the second phase, it worked just fine. You may find that a static phase converter will work for your device, it's just some capacitors to create the offset.

[TopLoser]

Before your replies get too complicated....

"The voltage can be squarewave"

[mikeselectricstuff]

Brushless motor driver ( as used for quadcopters etc.) perhaps?

[james_s]

That might be a good place to start. I've used RC ESCs to spin old hard drive spindle motors. I wired one up to a broken 10k RPM server drive with the lid off at at 24V it sounded like it was going to take off. Actually made me a little nervous, there's a lot of kinetic energy in 10 platters spinning at who knows what RPM but it was a lot more than 10K.

[Brumby]

While I am curious as to the suitability of a square wave output, to generate a 3 phase output without going to a microcontroller, I would head down this path...

One option to create a 3 phase signal would a counter going up to 6. The 4017 configured as a divide by 6 and than combining 3 inputs each to get the drive signals would be an option.
For 3 phase signal it is more like 6 MOSFETs to make 3 half bridges.

That's not excluding other alternatives that have been offered which might be a more complete and ready-made solution.

[mikeselectricstuff]

While I am curious as to the suitability of a square wave output, to generate a 3 phase output without going to a microcontroller, I would head down this path...

One option to create a 3 phase signal would a counter going up to 6. The 4017 configured as a divide by 6 and than combining 3 inputs each to get the drive signals would be an option.
For 3 phase signal it is more like 6 MOSFETs to make 3 half bridges.

That's not excluding other alternatives that have been offered which might be a more complete and ready-made solution.

All BLDC motors use squarewave drive - a 400Hz 3-phase motor isn't going to be that different, you may just need to adjust the voltage as 36V square may or may not be comparable to 36VRMS sine, as motor inductance will play some part.
 

[mc172]

Where did you get the information regarding the square wave input?

[BrianHG]

This doesn't sound like you need precision.  I would use a cheap 1$ PIC, with internal oscillator, without the PWM module, I would just program the 3 or 6 outputs to cycle with a timer loop between transitions with some dead time inbetween the gate activation.  You just need to get the motor spinning and I'm not sure if you have a position sensor to deal with, or if you should start at 400hz, or how efficient you need the drive to be.

[rcairman]

Hi,

driving BLDC motors is more complicated than you may think. Forcing 3 square waves shifted by 120 degrees each will not be any good, the motor won't even start spinning unless you are extremely lucky. It is also not going to work with half-bridge, you need full bridge with 6 mosfets.

Every BLDC motor needs to be commutated accordingly to its current orientation. This can be done in 2 ways:
1. Hall sensors or precise encoders
2. Back-EMF commutation

I guess you have a simple BLDC motor with no hall sensors, thus you need to apply current between 2 wires and "listen" for a response on third one. When you get the response, you should switch your mosfets and start listening on a different wire. In BLDC motors there is no common wire, thus you only have 3 wires coming out of the motor. You have to be able to run current both ways through each wire, hence full H-bridge.

What you are planning to achieve has already been done: Every brushless motor controller for RC planes or drones will be just fine for your application. If you cannot rely on off-the-shelf parts, you should easily find at least a few open-source and open-hardware projects online that you can use as a reference.

Just a few projects that I know:
https://github.com/bluerobotics/BlueESC
http://vedder.se/2015/01/vesc-open-source-esc/
http://autoquad.org/esc32/
https://pixhawk.org/modules/pixhawk_esc

[mikeselectricstuff]

Hi,

driving BLDC motors is more complicated than you may think. Forcing 3 square waves shifted by 120 degrees each will not be any good, the motor won't even start spinning unless you are extremely lucky.

The motor we're talking about is designed to run from a 400Hz fixed-frequency supply with no position feedback.
I can't see that the supply being square vs. sine is going to make much difference.
You could use PWM to approximate sinewave drive if it turned out that it mattered.

a 3-channel audio amp driven by a phase-shift oscillator would be another approach - I've had a gyro working like this in the past :
https://youtu.be/AO7pn3uiWA0?t=550

[tecman]

I have run a variety of 400 hz gyros in the past.  Phase shift oscillator (as Mike suggested) driving 3 totem pole transistors.  Works but quite a few parts.  Usually you can use a single 24 volt 400 Hz oscillator, with appropriate drive capacity.  You can use a cap (value needs to be determined by test or calcs) to drive another leg.  It will spin up with the 2 phases and cap as a phase shift.  Not elegant or high efficiency, but it does work.

Paul

[bogdan2014]

That's actually a brilliant idea. Will try it out.

[bogdan2014]

I've simulated the circuit and the only problem is that the signal gets attenuated quite a lot. I thought that I could apply the signal to the gate of 3 mosfets and switch the 30V DC voltage for each phase. But the audio amp seems to be the only option.
Maybe I can include some op-amps?

[bogdan2014]

I later found a buffered phase shift oscillator in an application note and tried to adapt and simulate it. What's weird is that the signal still gets attenuated even though every stage is buffered.

[Seekonk]

What's weird is that the signal still gets attenuated even though every stage is buffered.

What's weird? Every stage is buffered, but it is not amplified. Unity gain.

[bogdan2014]

I found a much more simple phase shifting circuit using op-amps. However, it only works (simulates) with sine waves. Square waves come out distorted.
I also cannot simulate a switching mosfet (right side), seems that the transistor won't open.

[mikerj]

I found a much more simple phase shifting circuit using op-amps. However, it only works (simulates) with sine waves. Square waves come out distorted.

That really shouldn't come as any surprise, this circuit uses RC networks to create the phase shift and the phase will be dependant on frequency.  A square wave contains lots of odd harmonics, all of which will be phase shifted by different amounts.  If you want to work with square waves, use a digital method (e.g. 4017 counter or microcontroller).

I also cannot simulate a switching mosfet (right side), seems that the transistor won't open.

Why are you trying to short out the 36v supply with the transistor?  Put a resistor in the drain circuit at the very least.

[bogdan2014]

Well even if I manage to get the waveforms correctly, I'll still need a driver for the high side and low side transistors. I won't be able to just have three transistors.

[bogdan2014]

I eventually figured out that the motor is in a 3-phase delta configuration and made a circuit that uses a microcontroller and IR2101 + mosfets.
It's a bit too much for this purpose since the transistors stay cold the entire time the motor is running, but it works.
I've attached the schematic of the driver section, the microcontroller simply sends signals on three output pins and the delays are calculated for 400Hz (2.5 miliseconds) + 1 microsecond dead time. In the future, the microcontroller will be included on the PCB.

Here's a video of it running:
https://youtu.be/s1OqEunYx8I

The next step would be to include some kind of RPM sensor in the project. I'm thinking of using an IR emitter and sensor and a small marker line on the rotor, so everytime the marker line passes in front of the sensor, it sends a signal to the microcontroller. Would that work?

[SeanB]

Leave it open loop, it is no good to include compensation for a gyro, as it is not needed. Would be a good idea to provide some low pass filtering on the drive output though, at least a LC filter tuned to the second or third harmonic at least so the drive is roughly a sine wave. Helps if you want to use the synchro resolvers as a position indicator, and these like a nice sine wave, though you might also have to provide a 6.3VAC drive for them as well from the main supply, using a 3 phase transformer to do it, which also acts like a filter as well.

[David Hess]

I would use a Johnson counter.

https://www.allaboutcircuits.com/textbook/digital/chpt-12/ring-counters/