PMP8740: PSFB T4 Primary waveform looks strange at my PMP8740 board

Hello Norman,

I have few questions about your design: does it have the same specifications like the PMP8740, so 2KW nominal?

I ask because 43V @ 8A is 344W, which is only 17.2% of nominal load. With 10uH shim inductor there is a minimum load needed in order to achieve ZVS.

Please refer to UCC28950 EVM calculation tool to derive what is the minimum inductance you need to get ZVS in a programmed load range.

Also the fact that you see spikes on the gate driving transformer voltage, in one transformer only, where ZVS is not active, might depend on the fact that the dV/dT is very high and you get noise in your scope. In order to double check that, please connect the isolated probe (must be isolated because we are talking about the floating driving section) shorted to the floating ground of Q8, which is the source of Q8, therefore connecting the ground and the tip of the isolated probe both on Q8 source. If you see spikes, then they are not real, because you should see zero.

Regarding the gate drive transformers T2 and T3, are you using the parts 56PR3362? These transformer are behaving better compared to the original ones I used, which are the AF4779.

Best regards,

Roberto

Thank you Roberto,

1) I use 10uH shim inductor , and I intend for 48Vbat at 20A design

2) I always use differential probes at this board measurement

3) I can adjust 10uH to decide when > certain load to approach ZVS, however I do have  concern on this abnormal

  ( to me , and it seems real and bother me ) voltage rise then fall on both condition when either gate B off --> gate A on , or  gate A off --> gate B.

because I did not see similar on many others design .

4)Herewith I attach 3 photos I measure today together with current probe ( Q8 drain to source direction )

thank you for any advice from below ~　

Hello Norman,

According to the first screenshot, I see that the waveform of CH4 (V8 Vgs) has a spike up to +42V. This high voltage should already damage the gate of Q8. I am wondering how it can be that you have so high value of spike. Did you make the test I suggested, by connecting both scope terminal and its ground (isolated probe) to the floating node (source of Q8)? If you analyzed this waveform, did you see just zero volt continuous waveform or did you see also spikes? If you saw spikes, they are affecting your measurement. If, instead, you saw a continuous line, showing zero voltage, then theses spikes are real and dangerous. I can only explain these spikes by too high leakage inductance in the transformer T2 or too high value of resistor R73 (higher than 10 Ohm).

Did you use the same FET I employed, like IPP65R190CFD, or are they different?

Thanks,

Roberto

Hello Roberto,

I use the same IPP65R190CFD , and also use 56PR3362 for all gate drive x'former. 

Let me get another main PSFB transformer with lower leakage then try ( current one I need to test

leakage when I have LCR meter , but I can tell its coupling may not good from the very fat  winding.

maybe need to use copper foil for secondary, instead of existing big round  litz wire that far away from primary

winding , it may cause poor coupling ( ie.,  large leakage )

I'm not sure I catch your point  on tie isolated probe ground and scope ground to Q8 source ...

I use the differential probe ( red tip and black tip , plus tip is isolated amplified to its BNC cable to the scope BNC )

and I think it is what you mentioned an isolated probe. But scope BNC gnd = scope's chasis gnd = scope AC plug's earth gnd ,

and I usually don't connect scope plug EARTH gnd  to ac outlet earth gnd,  due to sometimes when non isolated probe BNC gnd tie

to AC primary circuit ( L or N) of DUT board will cause  ac outlet earth gnd short to L or N.  and I treat Q8 source is low impedance to L or N.

that's also why I use isolated differential probe . so I am confused with if  scope BNC gnd been tied to Q8 source , whole scope chassis short

to 400V - a low impedance to L & N , is this kind of dangerous set up ?  

best regards

Norman

Hello Norman,

OK, I see you are using the same FET I employed in my design, as well as the gate drive transformer.

Regarding the isolated probe tied to Q8 source I meant:

Connect both red tip and black tip to the source of Q8. This way, the input of the differential probe should show zero on the scope. If you see spikes, they are not real, but just a noise current flowing from the switch node (*source of Q8) all the way back to the scope. This current can be avoided by winding the cable of the differential probe into a core (used normally for common mode choke).

Best regards,

Roberto

Hello Roberto ,

I ties two tips to Q8 source and a flat zero voltage line , so that spike right after gate B turn off is real . 

Yes, let me have another low leakage PSFB  X'former to try.

Thanks!

Norman