Pre-charge problems

[Silhouetou]

Hello! I'm trying to test the DCDC, and now there's an issue I'm confused of about the precharge part. Compared to the LTspice Simulation, in which without the TS input but only the 12V traco output, the voltage of the inverting terminal of the U10B still lower than the non-inverting terminal, so the U10B ouputs 12V to open the MOS, the precharge status LED don't light, the phenomenon now is incomprehensible for me(I've solder all the components but the start Cap, doing LV test to the step of switching on the enable switch) that if I switch on the enable switch, there are three LEDs lighting, precharge status LED, 3.8V Step down LED, 12SW LED. And I think they are contradict to each other, because I think the mos should be open now, with 12V appearing on the gate of the MOS and -HV_PC node, down to the TS- in(or HV GND) , so the precharge status LED should not be light. And then I just test with only my multimeter(Lately I still don't have an oscilloscope), the volt between TS- in and the -HV_PC is 12.2V, the 12SW is 12.66V, specifically, and basically check the relative components and don't find other problems. Could u give me some advice for the further step to figure it out? thank u very much!

[Rootthecause]

Hi!
hmmm... this is very weird indeed and should not be possible.
My best guess would be, that there was a measurement error (maybe measured Q4 insted of Q3?).
If you're sure, that your measurement is correct, then I assume U10B pin 7 might be not soldered properly or there is a disconnection somewhere around there.
Where did you measure the voltage of the U10B output? Maybe measure directly on the pin 7 and then on the Gate pin (pin 1) at Q3. Be sure to connect the negative multimeter connection to -HV_in.

[Silhouetou]

Hi! hmmm... this is very weird indeed and should not be possible. My best guess would be, that there was a measurement error (maybe measured Q4 insted of Q3?). If you're sure, that your measurement is correct, then I assume U10B pin 7 might be not soldered properly or there is a disconnection somewhere around there. Where did you measure the voltage of the U10B output? Maybe measure directly on the pin 7 and then on the Gate pin (pin 1) at Q3. Be sure to connect the negative multimeter connection to -HV_in.

Hi! Sincere appreciation for the reply! Actually, I thought over about whether I measure all the components relative to the pre-charge part or not after the question asked, and then I realize that there is something wrong. During the last test before it, there was a problem that TLV61048DBVR was soldered wrong, which caused an short-circuit, but it was not found until I felt the terrible heat from the isolation 12V-12V module, and then the voltage tested was up to 22V! And it was already was the second power-on(first for short time because 4 LEDs Osc power, precharge status, 12VSW power and 3.8V power, which was realized wrong). But for the second time, the Osc power LED turn down for several seconds. Then the short-circuit was solved, and after that, it was the 3 wrong LEDs light we talked.
After receiveing your reply, I just checked datasheets of the conmponents and realized the problem that maybe from UCC21520 and the UCC25600, they may be destroyed. And after the check, now I think it was the 22V that made the U11 conductive between the VCC and the GND, so the 12V net after the precharge was another short-circuit! Now the problem is already solved after the troubleshoot and now 3 LEDs light, but the one is Osc power LED now, rather than the precharge status, which really makes sense. And i think the diodes protect the 12VSW from the damage. And your advice is also really important because measuring directly not only makes the result more precisely but also reduces the risks of short-circuit...
Anyway, thanks a lot again!

[Rootthecause]

The 22 V on the isolation DC/DC (output?) might sound much - but the thing is, that those modules are very dumb converters and quite robust. In fact, for the v9-4 I was testing different gate drive voltages by using a 5 V to 12 V DC/DC and then cranked it up to 22 V on the output (TS side LV supply). Everything stayed intact, but the UCC25600 has a over voltage protection, so it shuts down between 18 and 22 V. Nothing was damaged at 22 V on the TS logic side, so that might not have caused an issue.

Nice that you were able to find the short in your replica! :D Do you know what was shorted?
You're welcome ;)

[Silhouetou]

Yeah, for the test, the UCC25600 was actually protected, and it's the UUCC21520 that caused the short, then I also found its maximum voltage between VCCI to GND is 20V in the datasheet, so the 22V causing the damage may seem reasonable...
By the way, I wonder if you will pay attention to the EMI of different buck or boost modules when choosing different voltage plat for power?
:D

[Rootthecause]

Ah, yes, you're right. I did not check the gate drivers’ max. voltage. I might have gotten lucky that it didn't die at 22V.

IMHO, checking EMI from datasheets alone is very hard. There are ICs which state some features that reduce EMI, like the LM5190, but most ICs do not have such features. More than enough, EMI depends on the PCB layout and used components, so EMI knowledge is key.
However, some ICs are flawed by design. I think the currently used step-down MP4541 might be a flawed IC, as it generates a lot of conducted EMI on the input rail (which is the 24V output rail of the DCDC). The proper solution would be a different IC, but it would require some redesign, which I currently don't have the time for. As an intermediate fix, I installed a ferrite bead in series to the stepdown input on the DCDCv9-4 design, which solved the issue very well in early tests.
Or do you mean larger Step-Up/Step-Down modules on the LV system?
Well, it's similar difficult to check those, but it is more likely to find e.g. output ripple specs. In doubt, try out.