[JRoque]

Looks good, Johannes. To cut the PWM noise, I usually add a low-pass RC filter and that works well. Current limiting the signal output and using Zeners also help in case you can't get rid of the noise source.

To clean up your signal, nothing beats hardware like logic or a Schmitt trigger. But I've also used the included comparator on many newer micros as a quick alternative. Not sure if the ST you're using has one of those and the pins are available.... there's never enough pins on a micro, isn't it?

Looking forward to video of your motor spinning.

JR

[Automcdonough]

I'm of the opinion that the DC bus can have whatever ripple/spikes it wants, the micro and signals should be rolling with the punches. In my experience more than 50% of noise immunity comes from careful grounding. You have your signal and power grounds separated? Each sensor's power and ground should have it's own path back to the micro's inputs. Same with the gate drivers. We call it a "spider" connection. that should connect to the power ground at only one place, usually right on the FET source for sake of avoiding false turn-ons. Or IGBT equiv. The CAN/USB communication to outside the box will need some protection/cleanup. I'd prefer to totally isolate it to avoid any sort of ground loop.

This makes it pretty much impossible for it to share the same current path as the power switches and bulk caps. That's like a freakin hazmat zone as far as the micro is concerned, 1000A surges flying around and stuff.
If you groundplaned it then the best you can do is make sure the whole plane connects to the power at only 1 point..

I hope that helps!

[jhuebner]

Well, its about time to report some news:

1. I implemented a simple slip control. The rotor speed is measured with a toothed wheel and the inverter tries to follow that speed. If the inverter has a higher frequency, the motor accelerates, if it runs at lower frequency the motor breaks. Works as expected, but the motor never came to a stand-still. I think there is still some EMI spikes that make it through the filter.

2. I tested the inverter with the newly arrived 18,5 kW motor. It runs smoothly, except at 40-50 Hz it jiggles about. At any higher or lower frequency it is smooth. If I accelerate too quickly, say from 30 to 70 Hz without a ramp, the bridge drivers switch off. Should be enough hardware protection for now.

3. Today I tested the inverter with a 3kW DC source and a 32 kW load resistor (3,5 Ohms). It gets fairly hot after a couple of minutes. Also one driver shut off, presumable due too over temperature. I'll add some fans to avoid hot spots.

The video should be ready soon!
Johannes

[JRoque]

Hello Johannes. Between this and your open source project work it sure sounds like you're keeping busy.

Hey so what happens if the speed sensor goes out and stops working? Would that tell the controller speed = 0 and command higher current/speed? Hopefully the controller will shutdown if there's no signal from the sensor.

Is the sensor cable shielded? I wouldn't expect EMI issue at such low power (nearly stopped) and no load. Maybe your comparator is being too sensitive or needs a bit of padding. Perhaps you can filter at the switch side with a couple of caps to take up the transients.

Could the vibration be self resonance at 50 hz? Lets hope some of the motor gurus jump in here.

A 32kW resistor! wow, never seen a monster like that. Is it a 3 phase load? Would love to see a pic of that thing or a video as you mentioned.

JR

[etischer]

If you run your inverter in simple V/Hz mode (no speed feedback), do you get the stall at 40-50hz?

Could it be your DC bus voltage is sagging and you don't have enough voltage to maintain your V/Hz ratio? Could your speed PID loop just be a bit unstable?

To get the motor to come to a complete stop with simple slip control you can try implementing a DC injection brake.

Originally Posted by jhuebner

Well, its about time to report some news:

1. I implemented a simple slip control. The rotor speed is measured with a toothed wheel and the inverter tries to follow that speed. If the inverter has a higher frequency, the motor accelerates, if it runs at lower frequency the motor breaks. Works as expected, but the motor never came to a stand-still. I think there is still some EMI spikes that make it through the filter.

2. I tested the inverter with the newly arrived 18,5 kW motor. It runs smoothly, except at 40-50 Hz it jiggles about. At any higher or lower frequency it is smooth. If I accelerate too quickly, say from 30 to 70 Hz without a ramp, the bridge drivers switch off. Should be enough hardware protection for now.

3. Today I tested the inverter with a 3kW DC source and a 32 kW load resistor (3,5 Ohms). It gets fairly hot after a couple of minutes. Also one driver shut off, presumable due too over temperature. I'll add some fans to avoid hot spots.

The video should be ready soon!
Johannes

[jhuebner]

Hi!

Originally Posted by JRoque

Hello Johannes. Between this and your open source project work it sure sounds like you're keeping busy.

Hey so what happens if the speed sensor goes out and stops working? Would that tell the controller speed = 0 and command higher current/speed? Hopefully the controller will shutdown if there's no signal from the sensor.

Security stuff will have to wait ;-) Right now it would shut down the motor inherently, because 0 rpm is reported when the sensor connection fails.

Originally Posted by JRoque

Is the sensor cable shielded? I wouldn't expect EMI issue at such low power (nearly stopped) and no load. Maybe your comparator is being too sensitive or needs a bit of padding. Perhaps you can filter at the switch side with a couple of caps to take up the transients.

Well, thats some hints. I have added an RC-filter right before the micro. But the cable isn't shielded.

Originally Posted by JRoque

Could the vibration be self resonance at 50 hz? Lets hope some of the motor gurus jump in here.

A 32kW resistor! wow, never seen a monster like that. Is it a 3 phase load? Would love to see a pic of that thing or a video as you mentioned.

JR

Yep, the resistor is rather large. It has its own trolley and weighs 50kg. Its small though compared to the 1MW resistors used for some tests of the 900kW solar inverters.
Next test is scheduled for Friday next week (17th of June). I hope that I won't forget my camera this time. Maybe I can connect to the motor before that date and make a video of it spinning at 200 Hz.

Originally Posted by etischer

If you run your inverter in simple V/Hz mode (no speed feedback), do you get the stall at 40-50hz?

Could it be your DC bus voltage is sagging and you don't have enough voltage to maintain your V/Hz ratio? Could your speed PID loop just be a bit unstable?

To get the motor to come to a complete stop with simple slip control you can try implementing a DC injection brake.

Yes, the stall is at 40-50 Hz. The DC-voltage is way too low for this motor anyway. Should be 500V but I only have 120V available.
Also right now there is no speed PID loop. Especially with the new motor there isn't even any slip control because the shaft encoder needs to be rebuilt.
I'm simpy running at a fixed frequency with the amplitude controlled by V/Hz.

Cheers.
Johannes

[jhuebner]

More news:

Heres a video: http://www.youtube.com/watch?v=PvUmBNgT1KQ

And some pics:

Fixed the capacitors and the adapter circuit board.


This is the 32 kW resistor. We did a 2nd test run last friday.
At 480V strange noises came from the unpowered inverter. Seems like some distances need to be enlarged to hold up to the desired 600V.

We then looked at the switching patterns:

Text book switching


Text book how-to-not-do-it switching

Around the 0-axis the switching looks really nasty. Besides that, it's beautiful. Anyone ever seen this? Hardware or software issue?

Cheers.
Johannes

[steven4601]

Awesome project!

My guess is false triggered gate signals. If you plot all 3 voltages you'll see the svmpattern for each phase flipping over. If it is synchronous to these weird short blips then you know where to start fault finding.

[meanderingthemaze]

Does anyone know, can you write controller software to work on multiple hardwares? Or does it have to be written hardware specific. Or could it just need to be ported to a different set of hardware rather than rewritten ground up.

[TigerNut]

Originally Posted by meanderingthemaze

Does anyone know, can you write controller software to work on multiple hardwares? Or does it have to be written hardware specific. Or could it just need to be ported to a different set of hardware rather than rewritten ground up.

On different controller families (and I mean microcontrollers, not a Soliton vs. a Zilla), the register interfaces to the timers and I/O ports are going to be different, so that the code would have to be customized at least at the hardware driver level. Above that, it depends on how much you apply the ethic of modular design and configurable code blocks. If the code was written without a view to separating hardware dependency from control algorithm functions, then it would be a ground up rewrite.