Topic: [Setup & Troubleshoot] Bitmain AntMiner S1 180GH/S miner - page 115. (Read 452359 times)

Heat soak on electronics normally works the other way, the device is run over temperature for an extended period to ensure heat induced breakdown of electrolytic material in a capacitor, the corrosive effect of any contaminants, the deterioration of plastics, the oxidation of contacts, joints and solders does not occur.

Just powering them off is good enough.

Well sayed, +1.

That is why i say many time above, just turn the miner off is simple enough.

I'm not an internal combustion expert by any stretch, but my brief understanding is that the problem with heat soak in an engine is that the block is at elevated temperatures, and requires active cooling. When it's shut off, the coolant and fan aren't moving anymore, so the huge thermal mass of the engine block is dumping its heat into the coolant, which causes it to heat up and pressure the system. The coolant actually gets hotter when the engine turns off.

The difference with these electronics is that heat generating elements are the silicon dies in the ASICs as well as the VRM chips. Those are the hottest components, and as you move down the chain from the die to the package to the board to the heatsink, the temperatures drop. However, the silicon die has an almost negligible thermal mass compared to the other components which tend to get larger as you move down the chain. If you pulled the plug and then immediately replaced the air around the miner with vacuum (to stop it from conducting/convecting heat to the ambient air) it might be possible for the PCB temperature to slightly increase as the temperature between the die and board stabilizes, but even that effect would be swamped by the cooling effect of the heatsink even with no airflow.

Not only that, but the hottest thing on the board (the dice) are some of the things most sensitive to heat. Electro caps are an obvious exception, but even if the board did heat up slightly it is less sensitive to heat damage than the dice, and that's doubly true for the heatsink. It's the exact opposite from the engine effect, where you have a high temperature high mass object passing heat into a liquid that is sensitive to being overheated. Here you have a tiny thing generating heat, dumping it into an object that has much more thermal mass and whose temperature won't rise much (if at all), and even if it did it's less sensitive to heat damage than the thing producing the heat in the first place.

Why wouldn't it?

In relation to car engines, yes. I'm not sure how that's applicable here though.

Ever heard of a term called heat soak?

Yeah, but you haven't explained why. The board won't continue to get hotter after you turn it off. It will cool down slower since the fan isn't running, but cooling down slowly isn't a bad thing.

The point is you want to wait until the boards cool down before powering down.

The miner doesn't keep working if you turn off the power at the wall.

The miner will keep working and making heat until there is no more work...the easiest way to stop it from working is to disconnect it from the internet.

Once it stops working then the fans will eventually stop because the boards have cooled down....this is a good time to power them down.

Could you explain that? I usually just switch off the power at the outlet.

The main thing you need to worry about on these things is heat-soak from shutting it down before the boards have had a chance to cool down.

All I have been doing is unplugging it from the internet and then waiting until the fans stop then just simply powering it down with the PSU.

Anyone know how to update the cgminer in this or add bfgminer to instead of cgminer??

zing! 

I'm running into similar issues as you, i.e. 1 of the board sometimes no longer flashes (mine is #1, vs #0), losing some hash power (going down from 180 to 150-160ish), etc. Power cycle has always cleared the errors, but sometimes it needs 2 attempts, and a software-initiated reboot (using the interface or command line) is NOT clearing the error condition. Instead, requesting a software reset makes the whole thing stop and it starts beeping until power cycle. THEN, after a cold restart (about 1 min cool down) it's hashing find again.

I think/thought my issues are/were PSU related, because it's rated at max. 350W for 12V so it's a few percent above max. rated load.

About the shutdown, the cleanest way to do it is probably the usual UNIX shutdown command:


...while SSHing into your box, default user/pass = root/root

I suppose they didn't bother with it in the web interface because they thought it would be useless. They might be right though... IDK

Cheers!

I will take a look again. Psu is a new seasonic 760.  I had other cubes and antminers before if it was psu it would start and stop.  In this case the antminer fan runs on low no green board side lights. Im sure board's arent energized correctly.

World suck to rma and have this out of action for 2 weeks.

Electrical tape does wonders

Try only running one board at a time. The instructions to do so are located in this thread somewhere. At least this way you can possibly rule out a PSU issue which is what I 'think' you might be having.