problems of forming a sinusoidal current

Post Reply
yaroslav
Posts: 176
Joined: Tue Aug 13, 2019 4:32 pm
Location: Russia, Stavropol
Has thanked: 14 times
Been thanked: 5 times

problems of forming a sinusoidal current

Post by yaroslav »

Hello JOHANNES. I built a new 135 volt inverter. Faced with the problem of the formation of the sine itself. I searched for the reason for a long time, connected various three-phase bridges, changed the voltage, it did not give any result. I shoot with an oscilloscope connected to the current sensor in the phase, you can see the sine with distortion. Maybe you have encountered such a problem?

I also noticed that when the motor with the encoder is spinning, there is a very large floating of the phase current.
Attachments
sinus.png
sinus.png (10.55 KiB) Viewed 2605 times
yaroslav
Posts: 176
Joined: Tue Aug 13, 2019 4:32 pm
Location: Russia, Stavropol
Has thanked: 14 times
Been thanked: 5 times

Re: problems of forming a sinusoidal current

Post by yaroslav »

This distortion is present at slip frequencies, if the encoder is connected and the motor is already rotating it is not there, the signal already becomes similar to a sinusoid, but floats on the current
User avatar
SciroccoEV
Posts: 369
Joined: Thu Oct 10, 2019 1:50 pm
Location: Luton UK
Been thanked: 15 times

Re: problems of forming a sinusoidal current

Post by SciroccoEV »

Do a search online for deadtime distortion.
johnspark
Posts: 264
Joined: Fri Apr 12, 2019 10:42 pm
Location: Adelaide, South Australia
Has thanked: 59 times
Been thanked: 48 times

Re: problems of forming a sinusoidal current

Post by johnspark »

you have at least 3rd harmonics, they are the distortions at the zero crossings
collin80
Posts: 110
Joined: Sun Aug 30, 2020 3:28 pm
Location: United States, Michigan
Been thanked: 4 times
Contact:

Re: problems of forming a sinusoidal current

Post by collin80 »

The answers can be complicated. To some extent it depends on the motor you are using. The only way that PWM output is going to look like a sine wave is if something with capacitance or inductance smooths the PWM into a smoother shape. It is possible that your motor doesn't quite have enough inductance to smooth the waveform into a sinusoidal shape fully. Also, some motors really do have trapezoidal windings and are really supposed to more or less look like that waveform you are seeing. Lastly, as SciroccoEV said, you seem to have deadtime around the zero crossing. Deadtime is meant to be there and doesn't really hurt anything - it merely causes a small period where you aren't really getting power output.

As for floating phase currents, yes that does sound like slip. Naturally, the goal would be to control slip accurately so that the output has only just the right amount of slip to get the maximum power. It sounds like in your case the dynamic load of not being at steady state is causing variable slip that affects the current output. Lots of motor controllers do all sorts of complicated things to try to manage this. Usually the answer is space vector modulation with field oriented control. So far as I know the OpenInverter stuff tends to avoid doing these calculations and instead does V/Hz (correct me anyone if I'm wrong there?!) If so, it has limited control over slip frequency relative to the way your specific motor works so you should expect that dynamic loads would cause "current draw" weirdness temporarily. Keep in mind that commercial controllers coupled to their own commercial motors are very, very carefully tuned to each other.
yaroslav
Posts: 176
Joined: Tue Aug 13, 2019 4:32 pm
Location: Russia, Stavropol
Has thanked: 14 times
Been thanked: 5 times

Re: problems of forming a sinusoidal current

Post by yaroslav »

I checked on this engine at the factory inverter, the sine is normal, but it is at a lower voltage, so I put more dead time in mine. Perhaps because of this.
yaroslav
Posts: 176
Joined: Tue Aug 13, 2019 4:32 pm
Location: Russia, Stavropol
Has thanked: 14 times
Been thanked: 5 times

Re: problems of forming a sinusoidal current

Post by yaroslav »

I turned the engine without load, could this be the reason for the floating current?
collin80
Posts: 110
Joined: Sun Aug 30, 2020 3:28 pm
Location: United States, Michigan
Been thanked: 4 times
Contact:

Re: problems of forming a sinusoidal current

Post by collin80 »

yaroslav wrote: Thu Nov 11, 2021 6:05 pm I checked on this engine at the factory inverter, the sine is normal, but it is at a lower voltage, so I put more dead time in mine. Perhaps because of this.
Dead time is merely to prevent shoot through, not meant to lower the voltage.

Think of it this way: in a motor controller there are four big transistors (or four sets of smaller transistors that act as 4 large transistors). At any time two are turned on allowing current to either flow one way through the motor or the other. Dead time is only there because transistors usually turn on and off at different rates. If one set of transistors turns off slightly slower than the other set turns on then you get four transistors all on at the same time and shorting the battery pack and/or motor. That's obviously not good. So, to prevent this situation you include some dead time where you hope no transistors will be turned on. This obviously does lower the apparent voltage a bit but the goal is for such a situation to be happening only for perhaps 1 microsecond. So, usually dead time is very short and not readily apparent even on a scope unless you're actively probing the PWM lines.

However, there is another aspect of dead time that you can see and you do see on your output. Near zero volts the output tends to flatten out more than you would expect. This is because you'd need to turn on the transistors very, very briefly to get such a low voltage. In fact, so briefly that dead time might remove any chance of transistors firing at all. Added to this, sometimes motor controllers will intentionally just output 0 PWM when the value would otherwise be small. This doesn't really affect the end result much at all because you aren't getting any power out of that part of the sinusoidal wave anyway. It's low voltage and low current. So, blanking out there doesn't really do much but prevent the transistors from trying to turn on for hundreds of nanoseconds or only 1-2 microseconds at a time. Instead you wait until the transistors were going to be commanded to turn on longer and then start doing it. This causes little flat spots near the zero crossing but nobody really worries about that.

Dead time is something you want juuuuuust enough of and no more. Generally what happens is that you set the dead time somewhere and then lower it little bit by little bit until you can see shoot through with a scope. Then you bring it back up about 2 steps and call it good. The needed dead time is mostly a function of the transistors and the transistor drivers and so a given motor controller can pretty much tune that value and then leave it alone. As OpenInverter does not send you the transistors you would be responsible for ensuring that the dead time is correct. But, for common things you can probably count on someone having already found settings that work well. Increasing dead time just isn't a good idea unless you really have a good reason.
User avatar
johu
Site Admin
Posts: 5684
Joined: Thu Nov 08, 2018 10:52 pm
Location: Kassel/Germany
Has thanked: 153 times
Been thanked: 960 times
Contact:

Re: problems of forming a sinusoidal current

Post by johu »

collin80 wrote: Sat Nov 13, 2021 1:39 am However, there is another aspect of dead time that you can see and you do see on your output. Near zero volts the output tends to flatten out more than you would expect. This is because you'd need to turn on the transistors very, very briefly to get such a low voltage. In fact, so briefly that dead time might remove any chance of transistors firing at all. Added to this, sometimes motor controllers will intentionally just output 0 PWM when the value would otherwise be small.
Quick correction here: 0 volts on a 3-phase bridge is accomplished by setting equal duty cycle on all 3 channels, typically 50%. So the effect merely happens at high amplitudes. Therefor a roughly 98% on time is translated to "always on"
Support R/D and forum on Patreon: https://patreon.com/openinverter - Subscribe on odysee: https://odysee.com/@openinverter:9
yaroslav
Posts: 176
Joined: Tue Aug 13, 2019 4:32 pm
Location: Russia, Stavropol
Has thanked: 14 times
Been thanked: 5 times

Re: problems of forming a sinusoidal current

Post by yaroslav »

I set a dead time of 2 microseconds, at a shorter time there was a through current. Transistors cost IRFP4668. HCPL3120 Drivers
Post Reply