1kW Gen2

[ChocoHolic]

... up to now I do see just two drawbacks:
-High ripple current load in the rail e-caps. But with proper choice that is no problem.
-Slightly higher no-load-losses. No problem, but of course not nice.

 

[Eva]

Choco,

Why I seldom see so low inductance output filter (+big capacitance) like you use (5uH)? You've given your reasons before, what are the drawbacks of this approach?

Try a second diode to force a lower supply voltage to the low side

 

[ChocoHolic]

Hi Eva,
I don't get it.
Under which conditions do you expect issues?

Low supply voltage of the low side driver?
But that's always trouble and I am avoiding issues with the UVLO of the drivers.

Inbalance of the supply rails?
From my understanding the supply pump effect would act stabilizing, not destabilzing.

 

[Eva]

From the datasheet note, one thing can be done to lessen the heat of the chip. To put external bootstrapp dioda. This dioda contributes to significant heat of the whole chip. But this external dioda has to be ultrafast with Vf<0V65.

Sorry, I was replying to this post from lumanauw. It seems that I checked the wrong "quote" box.

A way to bypass UCC27200 internal diode is to use an external dropper diode to reduce the low side supply voltage by 0.65V and connect an external bootstrap diode before this diode.

 

[ChocoHolic]

In fact I think that's a very nice idea.
It allows easy bypassing of the internal diode and does
match the supply voltages of upper and lower gate drivers.

 

[ChocoHolic]

Hi

Small and simple...go for it

I examined it and decided not to use it.
The overlap can be serious.
Especially if the pulses are short...

If you feed this into the inputs of the driver...

Upper trace: High Side Input with 2V/Grid
Lower trace: Low Side Input with 2V/Grid

 

[ChocoHolic]

...then you do not just get slightly longer ON pulses at the output... it may happen that the High Side Driver does not notice at all that it should move down, while the LOW Side Driver is delivering an enlarged pulse....

Upper trace: High Side Output with 10V/Grid
Lower trace: Low Side Output with 10V/Grid

 

[ChocoHolic]

If the data sheet is correct, then the UCC27200 should be better.
But data sheets are patient....
Hope to find time to organize and test some UCC27200.

 

[ChocoHolic]

The HCPL9030 is really looking nice, and up to now I was not able to cause any seriously wrong output signals.

Here a screen shot with a short HIGH pulse.
Upper Trace: Input, 2V/Grid
Lower Trace: Output, 2V/Grid

 

[ChocoHolic]

...and the HCPL9030 with a short LOW pulse.

Upper Trace: Input, 2V/Grid
Lower Trace: Output, 2V/Grid

 

[kASD]

Nice