Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[cts_casemod]

The line is 213 Vrms phase to Neutral, yes. But for an inverter what we're looking for the line to line, which is 415V, which of course, was proven.


By the way, was reading one phase with the wrong polarity, looks much better now:



Almost feel tempted to just feed the induction motor with this, in fact it's funny that I just happen to do that last week:





The results weren't that bad considering it was just a square wave!

File attached, if someone wants to see what I've done

[P-hack]

the 213v is phase to phase, 176v (peak) phase to neutral, purple is what one motor coil would see if you ran this in reverse (resistor is a 300v battery, diodes are igbts switching red blue cyan 120 degrees out, the lines to the voltage source are the motor leads.

Originally Posted by cts_casemod

That would allow 415V motors to be operated from a 300V Battery with no conversion losses. Anyone interested in this?

I don't think so, you can do the same reverse maze thing with a full wave rectifier also.

pretend the voltage source below is actually a single phase motor, and that the diodes are doing igbt switching a-la red and blue lines (300v is 100% duty cycle upper leg, 0v is lower igbt on, upper off) and the resistor is a 300v battery, you will get 600v peak to peak out (purple again, it is the source below), which is 300v peak which is 213vrms... My spidy sense tells me that is all you can get out of a 300v battery and a full bridge.

[MPaulHolmes]

Dang you people have been busy! I understand a word every now and again. Very fun reading!

I did some control modifications that let you get closer to hugging the line of maximum voltage without overflowing the error from the integral term of the PI loop during the "field weakening". So, now it ramps up to max rpm faster than it did before. Also, I optimized the SVPWM so now it takes 17 cycles rather than like 23-31 (depending on which "if" was the one).

[P-hack]

Originally Posted by cts_casemod

But for an inverter what we're looking for the line to line, which is 415V, which of course, was proven.

I did notice the author in one of those papers mixed up rms and peak (I mix this stuff up all the time), and/or maybe 415 is based on peak to peak, which isn't how you do rms unfortunately.

But another reverse engineering an inverter from a rectifier hack, to show you the same switching waveforms you are describing, is adding a capacitor. I prefer the leaving one leg on the rail 1/2 (or 1/3 for 3phase) of the time (rotating unipolar I guess) since it is less switching losses and less opportunity for shoot-through (and it is simple enough with a sine table, and processing power is cheap), but you can have bumpy lines on top and bottom and still get 600v peak to peak output (213v rms, i.e. 220v single phase motor at rated voltage) with a sine wave between the leads.

the green lead goes from 300 to 0 as the red goes from 0 to 300 and cyan shows the result. pink is the bus to ground (or the battery in reverse engineering mode). 150v is 0% duty cycle. 300v is 100% duty cycle for a top leg and for the catacorner bottom leg. same only different for 0v.

Sorry, I don't see where 415vrms is gonna come from with a 300v battery.

[cts_casemod]

Originally Posted by P-hack

Unfortunately that 300v peak purple line 2 line is still ~213vrms (i.e. 208v industrial motor at rated voltage).

P-hack,

We are talking about three phase machines and 3 phase waveforms

You're calculations, are showing the right results, you're just not adding all the information together.

3 phase equipment is rated in V(RMS) line to line.

A 3 Phase 208V motor has a Line to neutral voltage of 208/sqroot(3) = 120V.

With that in mind, your single phase 213V RMS has a line to line voltage of 230*sqroot(3) = 370V(RMS) when used as part of a 3 phase waveform

DC_LINK voltage for 213V RMS = Peak Sine amplitude is 213*sqroot(2) = 301V, assuming perfect IGBT's and no over modulation.

The concept may not be that simple to digest, and my theory could be initially wrong, but unless matlab is providing the wrong results the proof is there.

Give it a try, simulating a similar system, rectifiers work differently.

[P-hack]

Originally Posted by cts_casemod

With that in mind, your single phase 213V RMS has a line to line voltage of 230*sqroot(3) = 370V(RMS) when used as part of a 3 phase waveform

You need a full bridge to get 600v peak to peak from a 300v battery. The single phase examples show that. You cannot swing a leg more than 300vpp with a half bridge.

The purple line above shows 600v peak to peak between 2 half bridges (a full bridge). That is a line to line voltage, i.e. 213v rms (purple here is the difference between the red and blue lines, ignoring cyan for the moment). Thats it, there aint no more that I can see. You are getting the full 600vpp swing line to line (full bridge) with your 300v battery.

176/1.414 ~ 120v rms sources here.

I *think* the diode circuit is telling us the whole story here (sans overmodulation, etc).

If you took a six lead motor and gave each field a full bridge of its own (i.e. pretend it is 3 single phase motors 120 degrees out), you still won't get more than 600v peak to peak per field coil from a 300v battery.

I see lots of agreement here, just not understanding where 415v rms line2line came from with a 300v battery using only switches. It'd be nice if I was wrong, I guess I'll just grab my popcorn

Edit: added some annotations, with 300vdc, there isn't any more room to make a bigger line to line waveform and still have a sinewave.

[cts_casemod]

In regards to your previous question, getting 415V(RMS) from 300VDC, thats not possible. What i meant was, with a 300V Battery attain 213V line to line or 370V V(RMS) from the resulting output drive a 415V motor, which is mostly rated 380-440V.

How did I get to these values: Using the matlab simulations. I posted the file so that everyone can have a look at it. I also have a few more, if needed, with a bit different modulation.

On that simulation I've used a 10V DC supply and the resulting peak to peak was 20V. What this differs is that on a regular inverter, the PWM is center aligned and Vpeak = Vbus, hence sqroot(2) applies to find the maximum phase to phase RMS value, ignoring over modulation.

Another alternative would be a boost converter with a capacitor, two IGBT's and two diodes (or 4 IGBT's to be able to regenerate). I also did that (In real life one would need a pre-charge circuit as well)

The values given are:
Input Voltage = 400V
Duty cycle = 45%
Frequency = 10KHz
Capacitor size = 1000uF
Load = 10Ohm (About 78Amps current)

[P-hack]

Originally Posted by cts_casemod

How did I get to these values: Using the matlab simulations. I posted the file so that everyone can have a look at it. I also have a few more, if needed, with a bit different modulation.

Yah, I'm not gonna mess with thousands of dollars for matlab, so I couldn't really see what you are getting at.

so 370v rms is suspiciously like (213 line to line) to (213 line to line) with 120 degrees separation. Still at a loss how you combine the two (three? four?) though (easy as an abstract equation, not sure how to in the real world). If you could then you could just keep repeating the process indefinitely. Ah well, good luck

[P-hack]

one thing that does occur to me, in lieu of a boost converter (and maybe this is what is in your matlab) is to repeatedly reverse the polarity of the battery pack itself via full bridge a-la square wave when feeding the inverter bus (with a capacitive divider). Extra losses but 1/2 the batteries for a given bus voltage, and regen is pretty wonky.

[cts_casemod]

Originally Posted by P-hack

Yah, I'm not gonna mess with thousands of dollars for matlab, so I couldn't really see what you are getting at.

so 370v rms is suspiciously like (213 line to line) to (213 line to line) with 120 degrees separation. Still at a loss how you combine the two (three? four?) though (easy as an abstract equation, not sure how to in the real world). If you could then you could just keep repeating the process indefinitely. Ah well, good luck

I'm guessing you could do the same on LTSpice, haven't tried it tough.

Ill try to run a simulation using arduino and see where it goes.