E3634A Mystery Explosion

I thought I would open up (once agin) the E3634A and see what damage was done and what I found is a real mystery!

[JackZimmermann]

Electronics video blog with suspense! Part of me wanted a second bang on tape, but in my heart, I would feel bad with such nice equipment.

[markobutter]

When the FET failed the small gate PCB trace may have failed causing the sound/flash and subsequent intermittent trace problem. That's my guess

[ickipoo]

I don't think I would have had the guts to try the load on it in that configuration again - good one! Check those FETs carefully - sometimes the smoke-liberating cracks are very fine. I wonder if the fault on the gate of the preregulator caused the origial problem as well.

[gerrysweeney]

I found the smoke source, it was the pre regulator MOSFET, I have added photos' to the blog article (link in the video description above)

[theharbinger]

I am years late, but I am trying to repair my 3646A and came across your videos. Regarding the loud sound you heard. If during the failure it reached saturation current on the power transformer, a laminated core transformer can make a very loud bang when it ejects the field. The flash would have been a separate thing, probably the trace and gate failure at the FET.

[gerrysweeney]

Yeah I got to thinking that too, although that would still not explain the current shunt resistor destruction, that to me still looks like a faulty component. Anyways, the pre-regulator was definitely not working last time around because the PSU makes quite different noises under load now. I guess time and some use will tell. I think if it fails one more time I might give up with it - should have my own design done by then :)

[gerrysweeney]

It does amaze me just how big a visual show was created through such a small crack in the device. That is certainly a plausible scenario although I would expect the PSU linear regulator to handle that condition, although I would not be happy pulling 7A through a single Power MOSFET, it just feels wrong :) Its been reliable ever since the last fix though

[rogeronslow]

GerryI would have had a good look at the gate drive waveform of the pre-regulator. It could be that the initial explosion caused the gate to short to the drain damaging the series gate resistor making it high in value. This is common with switchmode  power supplies. I am assuming it has a series gate resistor of course.

Hmm, I do 10A through very similar cables when charging/discharging batteries. Sure they get lukewarm but never a sign of trouble.

[gerrysweeney]

These wires are high quality so not really, perhaps a small few degrees above room temp but nothing too outrageous. Gerry

[gerrysweeney]

No problem, it was bugging me that it was sitting there broken, I had to fix it! Then when I fixed it, it was bugging me even more that I could not find the source of the flash and pop so I had to find that too - now I am at rest :)

[Arek_R.]

This is why I hate small traces, even when its small digital signal.
Trace below 12mils is just mechanically weak.

[MIX]

Is there a video of this explosion you talk about?

[jpmorgan]

why didn't you solder those huge caps back on the pcb?

[hdmalpas]

Now you know why its called magic smoke, I had a led go up in smoke on a radio and to this day I have no clue why, I just replaced it expecting the same thing to happen again it never did, go figure it? M3KQW.

[ickipoo]

A graphic demonstration of just how fine a smoke-liberating crack can be!

My thought was the pre-regulator failed due to the gate track problem (or was just intermittent), then the series pass regulator failed short circuit due to excessive dissipation, which cooked the shunt resistor (and probably destroyed the load as well!)