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

[P-hack]

2 of the ES55242-140M-200AV might work interleaved @ 20khz per inductor to make 70A@510v combined from 144v, not real sure how to compute inductor losses though. Just resistive would be about 12watts apiece, 215 peak inductor current, avg ~107 (triangular). 1.1mOhm.

reduce the duty cycle or the load or the frequency and it gets discontinuous (though the inductor is ramping from ~0-215A as it is). Though I realize I'm proposing %100 current ripple here, so gonna be some AC losses...

[cts_casemod]

Originally Posted by P-hack

2 of the ES55242-140M-200AV might work interleaved @ 20khz per inductor to make 70A@510v combined from 144v, not real sure how to compute inductor losses though. Just resistive would be about 12watts apiece, 215 peak inductor current, avg ~107 (triangular). 1.1mOhm.

reduce the duty cycle or the load or the frequency and it gets discontinuous (though the inductor is ramping from ~0-215A as it is). Though I realize I'm proposing %100 current ripple here, so gonna be some AC losses...

Ive been doing some calculations using the micrometals simulator.

http://www.micrometals.com/downloads..._March2010.exe

Ripple translates to very high core losses. They go down with frequency (100KHz++). To reduce them one could increase the inductance, but that would increase copper losses. Add to that that the inductance itself goes down very quickly at high currents and this might be a harder challenge than originally envisioned.

[MPaulHolmes]

On their calculator, what do they mean by the inductor on and off voltage? If input voltage was 200, and output was 600, would I use those 2 values?

edit: I guess inductor ON voltage would be the pack voltage. And off voltage would be the voltage across the inductor while the switch is off? How awkward.

I was looking at some SiC half bridges on Mouser. They had a 300amp 1200v 5mOhm RdsON mosfet half bridge, but it was expensive, and the switching losses weren't small. They had 200KHz listed as a sample switching frequency, but then looking at the losses when it was an inductive load, each switch would have about 1500w of waste heat. And the peak power dissipation was 1600w.

The CH-200 I bought awhile back is under $300 in quantity 10. Maybe that is good enough. 200amp continuous

[cts_casemod]

Originally Posted by MPaulHolmes

On their calculator, what do they mean by the inductor on and off voltage? If input voltage was 200, and output was 600, would I use those 2 values?

edit: I guess inductor ON voltage would be the pack voltage. And off voltage would be the voltage across the inductor while the switch is off? How awkward.

I was looking at some SiC half bridges on Mouser. They had a 300amp 1200v 5mOhm RdsON mosfet half bridge, but it was expensive, and the switching losses weren't small. They had 200KHz listed as a sample switching frequency, but then looking at the losses when it was an inductive load, each switch would have about 1500w of waste heat. And the peak power dissipation was 1600w.

The CH-200 I bought awhile back is under $300 in quantity 10. Maybe that is good enough. 200amp continuous

I am using the PFC calculations. Which one are you using?

Just make sure the maximun input voltage is in AC, so for DC divide by 1.41 (144V ~ 100VAC).



Those values for the mosfet seem correct. I'm assuming they tested at 600V with a 100A load or something. Whats the RDS(on)? IGBT is probably best in terms on conduction losses at high currents, but switching losses are a bit higher as it is slower.

So, we have conduction losses and switching losses:

Assume 200Amps for the inductor

PD(conduction) = VCE/VGE(sat)*Io*Duty(on)

So for a 200A and a voltage drop of 1.7(assumed) and a duty = 50% that would be 170W

Now switching, depends on the frequency and the RDS(on)

PD(sw) = (Vd*Io*F(sw)*(t_rise+Tfall))/2
PD(sw) = (144*200*20KHz*200ns)/2
PD(sw) = 28.8W

here I'm assuming a rise and fall time of 200ns

At 200Khz, that jumps to 280watt! But the resistive and core losses on the inductor probably will go down twice as much.

[cts_casemod]

Bad news. I tried to get the inductor on the link I posted earlier, but the minimum postage option to the UK was $65. Add to that another much on customs!

If anyone's thinking in buying and can post me one let me know. It surelly turns out cheaper. I can paypal the payment or something!

Also related is the option to wire my own. I have been looking at the T600-52 core from micrometals. But the problem is getting some litz wire. Cheapest I found (AWG10) was $100 for 30ft!

Paul, that flat copper sheet you were going to use to for the E cores, have you found a supplier for it?

[z_power]

Hi cts, you might want to check this supplier in Poland: Feryster
Mike

[P-hack]

despite it's resemblance to a certain motor controller, at the multi-hundred dollar price point it is probably worth revisiting multiple high voltage mosfets @100khz, perhaps each with its own inductor and diode but one significant gate driver (or interleaved for $$). But getting the switches/inductors down to the 60 amp range could be good for costs.

Also it seems like litz would need a certain length to be effective, don't know, but plain old 28 strands of 26 gauge seems doable on a nailboard.

[cts_casemod]

According to this table:
American Wire Gauge table and AWG Electrical Current Load Limits with skin depth frequencies and wire breaking strength

For 10KHz the maximum conductor size would be AWG16 or 1.3mm, so that would be 4 strands

For 100Khz it would be AWG26 or 0.40mm, 41 strands. At this point makes sense to get the litz wire, doing it manually would yield a very large diameter conductor.

[MPaulHolmes]

For copper sheet, there's basiccopper.com, and nimrodcopper.com. They both have 6" x 0.020" x VERY_LONG. The E827 part is a little over 7" wide, which would be perfect for a 6" copper sheet.

[MPaulHolmes]

I just talked to Signal Transformer, and they sent me a graph of the CH-200 inductance vs. current graph.



I have used it 24/7 at 10KHz (even at night it's trying to track max power. talk about lazy programming on my part!) over the last couple years in a maximum power point tracker, running at about 90 amps during the peak sun hours of the day, in a shop that gets over 120degF. Not a dent in the fender! haha.

Notice the graph shows that the inductance is still 120uH at 300amp input.