probability of only 5V rail remote power cause PCB damage on welder?

[coppercone2]

So I am troubleshooting the inverter board on my welder, and this question comes to mind:

Is it generally safe to power a 5V switching reg on a daughter card directly from a power supply, a LM2575-5.0WT ? I never tried powering a card up by itself that had power electronics on it.

Do you think its possible to have any kind of interweaving in a device that applying power to a single rail would cause a problem somewhere.. like if something is biased from that rail in a HV power circuit?

I am in a expensive problem, I have about 15 seconds of runtime from my miller dynasty 200 welding machine, before the transistors in the inverter telemetry cascode? (like some kinda signal reciever), that are transformer coupled, get to 400F in about 10 seconds and parts far away start getting overloaded. I tried to diagnose it a few times but I am scared to even go in there with a multimeter while its squeeling, I feel like a cap is gonna blow.

I have the option to replace about 20 transistors (all suspect in the cluster), then leave probes attached and do a start, to get a single readout on the condition of the 5VDD rail that is interfaced with the HV transistors. It's kinda wasteful though, I am paying 15$ each time I need to replace these suckers. I don't know exactly whats going on because they appear directly connected to a HF transformer that goes off board to the inverter, i think its the signal coupler, but that port is actually the HV-measure/feedback one.

And I also feel that powering it like this is eventually gonna stress other components, I am sure I am shocking something unrelated.

So I have seen BATTERY backup circuits which are really unhappy without the battery, but how about a board found in a inverter supply being freed from seemingly irrelevant rail circuits?

My thought is that maybe some impedance is controlled by another rail and then its low and not high when that rail is off, overloading the regulator in a way not seen in the actual system. I suppose this is illogical right, since this circuit directly connects to like 74 series logic. Does anyone see any obvious traps before I confound the problem? my intuition says I won't see anything more then a load switch that I need to bypass or somekind of sequencing crap, that is high Z and not a risk

to clarify, I am suspicious also that the mosfet replacements I got have bad voltage rating, and its the HV through the transformer, thats either not snubbed properly or too high voltage for the crappy transistor to handle, but also these are OOS on digikey so I need to deal with shady third parties. If I knew the thing is not blowing out on the bench with voltage at the input of the power resistors that are changing from 2.7 ohm to 4 ohm in 30 seconds from heat because of the local regulation, I know I can focus my efforts on testing the interface electronics rather then the card itself.

The measured load on the regulator is also only 99 ohm, which is a little weird, but measured with the vreg in place, but thats only 50mA, but I assume thats a guarantee that another card not connected to the board is required to maintain minimum load resistance parameters.

[coppercone2]

well I just went for it, with 5V directly to a cap it was drawing substantial current, so what I did is then hook up 5V into the input of the switchmode regulator IC and I saw it was just passing current without regulating at all, and what happened is a BCN4013 took out the rail.

Now I see clearly when you go above 5V on the 5V rail, the

-the BCN has the lowest rating of 15V
-the 4000 series parts are higher voltage, so are the other chips, like op amps

I am guessing the rail voltage is above the 15V limit of the BCN so it takes it out, it leaves the other chips unharmed, but since that 5V reg is connected to the mosfet, it leaves em at a higher level, and on a power supply when I ramp it up past 5V I see a huge increase in current draw, by 7V its drawing at 2W but at 5V its running at 0.5W and the chips that are warm I can try to replace one last time before redoing the 5V test them (they might be alright, but I suspect maybe slight gate damage on 4 chips) trying to analyze the circuit around the switching regulator to see why its not regulating to 5V.

So the feeder to the board is screaming because what I guess should be a ~50mA digital rail was drawing several amps, and the massive thermal flare from the fets was hiding the damage on the BCN chip.

Now the question is, is the sense circuit still being pulsed by something that can re-do all this damage on next powerup. I probobly will scope the pins to see whats going on near the transformers.