Quote from: CryptoCoin2015 on April 30, 2015, 08:23:17 PM

Quote from: JaredR26 on April 30, 2015, 07:58:01 PM

Ok. So why did you make claims that it might have been 100C to deform it? You can practically deform it with your hand. By your own admission you must have known this. Why make such claims?

There is a huge difference in a material being easily able to be bent by physical force and material sitting in a shelf being deformed by heat. I will put a 1.200 watt blowdryer on the weekend against one of my S5s sides (gotta pick one) and we will see the outcome. I will measure the heat at impact, ok?

Fair enough, but that doesn't rule out physical deformation.

In addition, having recently done this with a different type of plastic, heat deformation of plastic generally leads to very different results than your pictures show. Since the deformation itself is caused by the expansion and contraction of different sections of the plastic at different rates, what you end up getting is generally edges aren't straight anymore and/or that the material will bulge, bow out, or sag at certain points.

The pictures you've shown show perfectly straight edges on all of the plastic, with the possible exceptions of the middle one(rear, left side) and the far left rear one, but both of that could be camera angle too. I don't see any bulges or sagging, nor any deformation of the antminer/bitmain logos. All of the units have bends near the screwholes, sometimes much sharper than what heat alone will achieve, implying that they were bent either on installation(at bitmain), during transportation on one of the several trips, during setup, or during the removal of the covers. The largest bend is the far rear right one, but A. the bend is sharper than what most heat damage would cause, and B. both the edges and the lettering on the logo are still straight.

I honestly think the plastic sheets are a red herring completely.

Quote

Quote from: JaredR26 on April 30, 2015, 05:34:32 PM

I know this to be at least partially untrue, but it is all secondhand information. However you and I can definitely agree that things weren't handled right, and that this was not how a hoster should treat their customers.

Yes, and no hoster should have the customer get aid from the police to retrieve his equipment.

In this particular case I agree this could and should have been handled better.

That said, if a customer hasn't paid their bills, they can't retrieve their equipment. Hosters have big electricity bills and other bills to pay, none of whom give a damn if your customers haven't paid their bills.

And clearly your hardware did run up its portion of an electricity bill, otherwise how could it have damaged capacitors? There is no temperature in an operational bitcoin mine, no matter how poorly cooled, that can damage a device that isn't powered on(Short of fire, not relevant here).

Quote

Quote from: JaredR26 on April 30, 2015, 05:34:32 PM

Right, and it will be good to have their assistance; keeping in mind that they also have a very strong impetus to avoid both the liability for the hardware and the reputation impact that it would have if they admitted any fault such as automatic shut-off temperatures not working. Bitmain are good guys, I know them, but they are also very clever. If Bitmain can avoid reputation impact by pinning 100% of the blame on another party, they will likely do so even if a failure on their side contributed to 20% of the problem or more.

Agreed. I am suspicious of Bitmain as well, but to have such an error quote in such a short time frame under the same conditions, with only me claiming these errors seems to statistically less relevant and only an option to pursue after all other reason have been ruled out.

Quote from: JaredR26 on April 30, 2015, 05:34:32 PM

50% of what boards? 50% of the pieces of plastic? I'd say that's rough handling or the result of running in a BTC mine. 50% of the circuit boards have a capacitor blown? Now this would be some useful data. Yes, I would agree that if you have 50% of S5 boards with visible capacitors blown like that I'd agree that you either have a major electrical problem(more likely, although I know the electrical is pretty solid there) or a major heat problem(heat affects chips more than capacitors).

50% of S5 boards with visible capacitors blown or similar damages.

Do you happen to have more pictures you could post and/or send me? That is actually very very suspicious. I know of bitcoin mines that had spots of up to 135F(57C, some with GPU, some with bitfury hw). The bitfuries had few failures, and the GPU farm had mostly PSU failures. I've personally run AM hardware at 45C stably, and know of other mines that ran (for a time) at upwards of 76C with bitmain hardware and afterwards ran at 49C for a long time, and had a lower failure rate than that.

All that said, I've been in that DC a bunch and know their cooling pretty well. I'm sure it got hot, but I really have a hard time believing that it got above 57C without a major cooling failure(Which, conspiracy theories aside, doesn't seem to be the case). And that's where my background is important, as I've designed and built the cooling for 4 or 5 bitcoin mines now. The record day was 27C at peak, so that would be a 30C temperature rise, which is crazy.

Even if you doubt me or my background, I'd be really careful about making the central point of your investigation/case a > 57C ambient temperature, because they may have/gather other data and if they prove you wrong, your entire case falls apart.

If the failure rate is truly that high, my best guess from what I know is:
A. It was definitely too hot in the mine, though lower than the highs I have seen in other mines(57C)
B. S5's are particularly sensitive to heat damage, moreso than most other hardware brands and even previous generations of antminers.
C. The S5's did not apply any temperature protection, if they have such.

On another thought, what power supplies were used with these S5's? Link or image? To me those are more likely to blow capacitors(likely in combination with the heat), if they don't deliver clean voltages.