As someone who used to do safety testing on power supplies and electronics (I still do as part of my current engineering job, but my previous job was just electrical safety testing for a Nationally Recognized Test Lab (NRTL) much like TUV Rheinland, the company who certified that power supply there...I didn't work for them, but one of their competitors), I'll walk you through some observations that I see. I can't definitively say what caused your fire, but some things are apparent to me. I'm sorry this goes back and forth a little, but I'm trying to explain why you might look at one thing, then rule it out or push further in that direction.
It looks like most of the fire was from the cables going from the PSU to...a variety of things, but it seems that once wire insulation started going, it continued (which is odd...due to wire insulation non-flammability specs). I'm betting the plastic sleeving is what did the most burning as the actual wire insulation is
supposed to be rated against that sort of thing. For reference, V-0 rated insulation you can actually hold a BIC lighter to and it even if it looks like it's melting or starting to burn, as soon as the flame is removed the fire goes out and the insulation doesn't continue to burn. In the lab it's done with a Bunsen burner and a very specific type of flame, but I just wanted to use an example everyone can understand.
The main area where connectors look damaged is on the motherboard near where your 24-pin connector goes. It looks like that bundle of wires was where most of the heat was, but it's hard to tell what is culprit and what was just in the line of fire.
Knowing that you changed the motherboard recently, I would wonder if something was pinched under the motherboard and over time, poked through the insulation on a wire and eventually caused a short (hence the delay between set up and fire). If there was a slight-short (not a "dead" short), you could have had a situation where a considerably higher current than the wires were rated for was drawn, but not enough to trip the supply. Your power supply can output 54.1A on the +12V line, and 20A each on the +3.3VDC and +5VDC lines. If you were putting 18A through one single wire for example, it might not have tripped, but that's a pretty high amount of current for one single wire. If that wire then heats up to the point of melting the insulation off, it could ignite other more flammable things around it, then the fire can spread. However, if this were the case, I would think this should be limited to one wire, couldn't be enough current to trip the over-current protection, and the fire shouldn't spread to other wires from the power supply because of the insulation being rated adequately. It should just be one wire in that case. Also, the supply should absolutely detect a dead-short, like if +12VDC shorted to the frame of the case. I've seen people do that before and you just hear a click from the supply, it won't even turn on. There's no perfect way to protect from currents less than the rail's maximum, otherwise the power supply wouldn't be able to run as much power as it does, but all that said, you've got a lot more than one single wire burnt here. Because what you'd usually see in this situation is one wire will burn and "fuse" open, basically removing itself from the circuit (not the whole PC going up), I don't buy that as the culprit.
It seems quite odd to me to see this much wire insulation damage unless something else was burning and continuously heating the wires, especially after you removed the power from the situation. You've got a whole lot of wires with burned away (or likely melted away) insulation. If the insulation was rated according to what the regulatory marks on the side of the supply say they should have been (and
also built to specification), that shouldn't happen from just a fire spreading on the wires. You have multiple wires that no longer have insulation. This means there was a great deal of heat/fire in that location for a prolonged period.
Here's another odd thing: the wires look mostly fine down by the power supply. For example, if you had one wire heating up due to over-current, it would heat up through the entire length of the wire. It's not just going to heat up on one end. I would open/fuse at the weakest point, so you would likely see one end a bit worse than the other, or one spot that melted more, but your wires, sleeving, and heat-shrink down on the PSU end have no major signs of heat. So it doesn't look like the "heated wire(s) starts a fire" situation.
So I'm left with a couple non-independent theories (if anybody has followed my ramblings to this point).
1. a device in your computer had some sort of hard failure, resulting in actual fire, not just short-circuit-heat in the wires (GPU, hard drive, or motherboard component), and drew less than what would have been needed to trip the over-current protection of your supply in the process.
2. This fire then spread on the all the mesh sleaving and plastic bits in the cable bundles, particularly above the point of fire. While the insulation is usually supposed to be non-flammable, the sleeving, zip ties, heat-shrink, etc. is not always rated accordingly and if that is what was actually on fire, it could have still melted the "non-flammable" insulation. I think this because of how localized on the wires the burning was.
3. The melted/burned insulation, burn marks everywhere, and heat pattern seem to start right between the right-most end of your GPU and the connector-side of the SSD in the picture. The wires below that look fine. Heat goes up. My
guess from all this would be that fire was literally come out of the back of one of those things and burned everything above it. I'd have some questions about why it happened?, how much current was involved?, did the PSU just keep feeding it even if it was over current limits (i.e. did the OCP work correctly)? etc. but from the pictures and my experience with burning/melting insulation off wires from a variety of fire and over-current, this would be my guess.
Edit: Oh dear. Thanks to a fantastic observation from Tyger (below this), I now realize that the SATA data cables seem like some of the worst damaged wires, end-to-end. I'm currently scratching my head to think about what could have caused all of them to do that (motherboard putting shorting 12V to the data lines somehow?...really not sure), but if the motherboard shorted, the cables would be fine...if there was a short in the SSD, it could maybe draw the kind of current needed to burn the wire, but not trip the PSU...maybe...I can't believe I completely missed how bad those data cables were. I was thinking of CPU power cables that plug in below the 24-pin on some super high end boards (like the newer DARK boards, not this board) and missed that those were data cables burned that bad end to end.
post edited by B0baganoosh - 2022/03/15 13:00:50