New Build caught fire - help find the culprit

dethsw · Nov 21, 2017

All new componants for new build. built bare-bones to boot to bios to ensure major components were working. First power on, white smoke cause me to immediately pull the power (Smoke started puffing out instantly after power on). In total the machine was on probably 10 seconds.

Smoke coming from the CoolerMaster HotSwap board SATA Power cable connector, which was connected to the PSU modular peripheral - IDE/SATA connector.

Pulled the hotswap card to investigate. Then followed the cables from the SATA POWER connector on the hotswap card, through the SATA data cables ultimately connected to the Mobo SATA express cluster.

All cables and connectors from the hotswap board to the Mobo are burnt/melted, this includes two SATA data connectors on the hotswap board, 2 SATA cables (on both ends), and the SATA data connectors on the Mobo. The most sever damage is the power cable on the hotswap board, but obviously the current flowed through the board, and through the data cables into the Mobo connectors.

After removing the hotswap board, I powered the machine back on and it booted to bios.

What is the likely culprit of the failure? The hotswap board from CoolerMaster? The PSU from SeaSonic (80+ Bronze rated), the PSU to SATA power cable?

Everything Connected:
PSU: SeaSonic 620w Bronze+ Certified
Case: CoolerMaster HAF XB EVO (included hotswap card)
Mobo: Gigabyte GA-Z270P-SLI
CPU: Intel i5 7500
RAM: G.Skill Ripjaws 16gb DDR4 3200
CPU Fan: CoolerMaster 212X
Monitor: AOC 23" via HDMI to DVI

FYI - No Disk Drives were connected, I just wanted to boot to bios.

HotSwap SATA Power Connector:

HotSwap SATA Power Connector Unplugged:

Mobo SATA Data connector Fried:

Mobo SATA Data Connector unused:

Darkbreeze · Nov 21, 2017

The power supply is unlikely to be the culprit. Especially seeing as it's a quality brand and model, regardless that it's group regulated.

The hotswap components from Cooler master, or anything from Cooler master for that matter, are, in my opinion, the most likely culprits.

Hotswap hardware has a spotty history to begin with, with the majority of those that are built into cases having issues in one way or another. The fact that it's a cooler master product, a company I have history and take exception with due to the high number of problems I've seen on a heck of a lot of their hardware, makes it even more likely.

The fact that you are able to boot with no further problems with that hardware disconnected makes it even more suspect.

It's certainly possibly that something else is to blame. I am however wondering if perhaps one of those wires in the hotswap connector did not get inadvertently installed in the wrong place on it's connector. I don't know if those are seated by a machine that does the exact same thing every time, or human hands, but it could also be that one of those wires going to the HS board were soldered in the wrong spot as well. Something had to have created a direct short in order for that to happen. I'm assuming that there is no chance you accidentally plugged something in wrong since those are keyed connections?

dethsw · Nov 21, 2017

DarkBreeze - Yes you are correct, every connector is keyed and I was very particular taking it apart to look at every connector. Not to potentially shift blame, I do not believe any of the connections were plugged in correctly.

I also believe its the hotswap card - (and I wished I trusted my gut) but when I unpacked the case, I saw the hotswap card and honestly it looked cheap. the board has residual flux smears, and where where someone nipped off the through-hole component ends, they scuffed the board in several places.

I might research the sata power connector pin-out because the board is clearly marked what each of the 5 wires is (+12V, GND1, +5V, GND2, 3.3V) - I'm curious if maybe they soldered on the connector backwards.

[UPDATE] A quick search and it looks like the power connector is soldered on in the correct orientation with the 3.3v pin 1 going to the 3.3v board pad.

Not sure if its important or not - the part of the connector that melted was the GND1. Additionally All the SATA Data cables are burnt on the Pins (1,4,7) all of which are ground pins. The only place in the series that isn't melted along the ground is the power connector from the SATA Power cable to the PSU. I don't even smell the smoke.

The Mobo is grounded through the standoffs to the case, the case is connected to the PSU via 4 screws from the case, and the PSU is grounded to my house ground via the plug.

Is it possible the PSU in that plug has a bad ground... so the system wanted to dump the angry pixies down to ground, but because the plug didn't ground at the PSU - it started melting backwards up the chain progressively getting less sever to the Mobo???

tsnor · Nov 21, 2017

FWIW, I googled to see why the Seasonic's overcurrent protection PSU didn't trigger and found this forum append "..That Seasonic M12II Bronze EVO Edition 620W PSU only has a single +12V rail despite what is shown on its label. If it did have two +12V rails then the PSU would have multirail OCP. That PSU doesn't have any OCP (Over Current Protection)..." Guess that explains why the short pulled enough current to melt the insulation.

Surprised the MB powered up after you pulled the parts. The MB would also have pulled a good bit of current.

If you have a multi-meter check the resistance between the 12V input (and maybe the 5V and 3.3v) to what's left of the ground pins. Think you'll find the short that way.

Darkbreeze · Nov 21, 2017

I've had, and seen, MANY issues with the front panel mini-boards on Cooler Master cases. None of them were models that had hot swap units, but due to the high failure of those mini-boards (And you could see poor soldering and electrical trace damage on most of them) I'd be surprised if the hot swap boards were any different.

Since you have the schematic, I'd make sure that EVERY one of those connector matched what the SATA power connector schematics are supposed to pin out as.

I'm assuming that there is no chance you had the SATA modular connector in the wrong spot on the PSU, as I believe those too are individually keyed and that you cannot insert a PCI connector into, say, the 12vEPS connectors spot. I have seen some power supplies in the past that had poor design in this regard, with more than one connector keyed similar, but never on a Seasonic unit.

Anything is possible though, always. I'm wondering if you have another SATA device, something older that isn't going to be used like an old optical drive, that you can plug into the PSU in the same socked you had that one connected to, with the PSU OUT of the computer, just to see if perhaps it IS a PSU internal problem? Seasonic is great, but not perfect, and sh#$ happens sometimes.

dethsw · Nov 21, 2017

Thanks so much, both of you (Tsnor & Darkbreeze).

I actually do have a HDD that I cant seem to get recognized by Windows due to a crappy format. I will pull the PSU and connect to each of the Peripheral SATA connectors.

Also - the SeaSonic has each type of connector (IDE/SATA, M/B, PCI-E) keyed so you cannot insert them incorrectly. The SATA power I was using is a 6 pin layout, and each of the 6 keyways is arched differently than the larger 8 pin PCI-E connector. So I physically cannot even jam it into the wrong modular connector.

Tsnor - Id like to try your recommendations, but I apologize that I don't fully understand your comment. I do have a multi-meter, am I checking the pins at the motherboard? or down the circuit to the hotswap card, or PSU?

Darkbreeze · Nov 21, 2017

tsnor :

Yes, testing is probably a good idea. I usually don't instantly go right to the DVOM because most of the visitors we have would become immediately jittery or flat out admit they are scared to test anything since they are unfamiliar. But that's the best way to figure out what's going on.

tsnor · Nov 21, 2017

One suggestion, pull the MB from the case and inspect the traces on the top and bottom for signs that the MB PCB melted. The traces that carry the current from the MB 4-pin / 24-pin to the Sata port also likely got some serious current.

IF it was not clear above, my guess:: looking at the burnt HotSwap SATA Power Connector, test the resistance from the 12V (or 5 or 3.3) input pin to the burnt ground pins. There should be resistance, not open. Think you will find an open short. It looks like a short pulled current to ground through the hot swap card, and the pins that took the most damage were the ones with the worst soldering job and thin connector so highest (albeit low) resistance - that's what heats up the most. The signalling wires were not meant to carry a lot of current. The connector at the sata power end to the PSU was thicker wire and solid connection so low resistance, therefore low heat so you do not see damage there "..The only place in the series that isn't melted along the ground is the power connector from the SATA Power cable to the PSU. " .

dethsw · Nov 21, 2017

Pulled the PSU - got a new SATA Power cable and connected the HDD. In each of the 4 SATA modular connectors nothing seemed to happen, no smoke or anything.

This PSU is new to me, because my existing Thermaltake Toughpower 750w has an LED that shows its in standby mode, etc. This SeaSonic doesn't show anything and the fan doesn't even kick on when plugged in and turned on.

Out of Curiosity - because this is the first time ive had component failure. Do I RMA each piece separately and the companies will just send out new ones, even if its not their component that failed? Do I have to prove failure to get coverage, and if I can prove failure, does that manufacturer reimburse for all other components that were damaged? Or am I just out the cost of all the components?

Darkbreeze · Nov 21, 2017

You are likely out the cost of everything except for the case. They should, hopefully, replace that since clearly there was a problem with it. Most these manufacturers have explicit disclaimers stating that if damage is caused to other hardware because of their hardware, they will only warrant the cost of their own hardware. But you can certainly try.

Many hardware manufacturers will simply replace a damaged item unless there is CLEAR evidence that something intentional was the cause. It's just cheaper for them to replace than to argue or try to fight it. I've yet to have a power supply, memory, motherboard, graphics card or case manufacturer refuse an RMA if something is wrong, caused or otherwise. CPUs I don't list because I've never had or seen a CPU failure that didn't involve another component or was simply due to age. CPUs just don't generally fail until they are either old or you ruin them somehow.

You can certainly bring it up with CM support though, and see what they say.

tsnor · Nov 21, 2017

dethsw :

The store that sold you the parts is going to be more helpful that the manufacturer. Contact them. Show them the pictures. Ask for RMA on any parts you no longer trust. See what they can do. Your case is much better than the average "MB doesn't work" because the builder didn't know to plug in the 4/8 pin -- you know they take those back all the time.

Second approach is to contact CoolerMaster and see if they want the burnt card back for PD. The company I work for would be overnighting it back to project engineering so they could find how the card escaped testing. Not sure a low cost company does that. IF you go this route try to get to their "social media" team if you hit problems. I'm sure they dislike it when their products show up burnt in photos on the internet.

(And because my earlier post was unreadable, with the mult-meter try to find where current can go into the card through one of the 12v, 5v or 3.3v pins and come out the burnt ground pins with low resistance. When you find it on that hot swap card you likely have your root cause. There is a chance that the short melted open before you pulled the plug, so failure to find it does not rule out the hot swap card )

dethsw · Nov 21, 2017

Pulled the Mobo and inspected both top and bottom traces, with extra review of the ones to the power connectors. I cannot see anything, nothing, not even discoloration, flat solder color, nothing. I even pulled the manual and found each ground connector and reviewed both bottom pins soldered to the board, and the pins in the connector. I don't see anything.

I submitted an RMA/Troubleshoot ticket with CM and supplied pictures, etc. We'll see how this works. I will put an RMA for Gigabyte because I'm sure the DATA side of the two SATA connectors are fried at the very least, and possibly more I am not supper willing to keep testing by adding more working components that can potentially be the next victim. I will also RMA the powersupply, and then just buy a better one, with special consideration to OCP!! Any recommendations?

Everything purchased from Newegg, along with a major rookie mistake. The Mobo - PSU and Case all had mail in rebates - which I instinctively mailed in immediately. Needless to say I am now wiser to the fact Newegg will not RMA anything that doesn't have the UPC.. So now I have to work with each manufacturer specifically.

Darkbreeze · Nov 21, 2017

Yes, Newegg is terrible to work with in that situation. Always make copies of the UPC if you have to send it for MIR.

IF you go this route try to get to their "social media" team if you hit problems. I'm sure they dislike it when their products show up burnt in photos on the internet.

This is 100% completely and unavoidably true. They don't like that and will usually work hard to spin it in their favor with positive support. I'm afraid I can't agree about the seller being more helpful than the manufacturer though. I can't say that's not true, might be in many cases. To my experience though I've had a lot more success going straight to the OEM than dealing with the seller, although if it's been within like 30 days they usually DO want you to deal with the seller first.

I think that might be more complex in a situation like this though.

As to the power supply, any of the more modern units that are not group regulated will have DC to DC conversion and MOST of them will have multi-rail OCP. Nothing wrong with your choice of Seasonic, you just happened to choose a model that has used the same platform for a very long time and while it was an excellent platform back in the day, or when fitting everything into a budget is absolutely necessary, it certainly isn't the greatest of offerings from them these days. All of the S12II and M12II units are group regulated.

Also, pretty much every 620w unit you see out there is based on this same platform. Antec and a bunch of others sell units with this platform inside and are Seasonic built models. I don't know of any other company making 620w units anyhow, although there might be, but I doubt it.

Seasonic and Super Flower built units are usually going to offer you the best build quality. That means, for US purchases anyhow, that you probably want something in a newer design from Seasonic, XFX (NOT the XT models. I'm pretty sure all the rest of XFX supplies are Seasonic built), EVGA (Gotta check model. Avoid the B models. Stick with G2, G3, P2 and T2 models which are all Super Flower built).

Personally, if it's not one of the higher end Seasonic's, or a Super Flower gold or higher Leadex based model, I try to avoid it unless of course it's budget oriented and unavoidable.

dethsw · Nov 22, 2017

Initiated RMAs.

In the mean time bought an EVGA SuperNova G2 650W Gold. Since I want some kind of protection.

Hopefully Gigabyte and CoolerMaster will work with me

Sincerely thanks for all the help!!!

Darkbreeze · Nov 22, 2017

Of course. If we can help with anything else just let us know.

Search

New Build caught fire - help find the culprit

dethsw

Prominent

Darkbreeze

Darkbreeze

Retired Mod

dethsw

Prominent

tsnor

Splendid

Darkbreeze

Retired Mod

dethsw

Prominent

Darkbreeze

Retired Mod

tsnor

Splendid

dethsw

Prominent

Darkbreeze

Retired Mod

tsnor

Splendid

dethsw

Prominent

Darkbreeze

Retired Mod

dethsw

Prominent

Darkbreeze

Retired Mod

TRENDING THREADS

Latest posts

Moderators online

Share this page