Topic: Maintenance of rig capacitors (Read 632 times)

Cheers mate.  Honestly it is a tough time for us miners.  Not too many choices of coins to mine in profit.

No problem, glad it worked out for you.  Been mining since 2013 so I've seen so many things I could never claim to remember them all. Just hang in there and be patient... seems like you'll be fine  

Hey thanks for the tips, I basically did the same thing as you suggest by going slow, warming up the rig before doing regular OC and equihash algo.  Luckily that worked well. 

Consider the source.  If your source has 100's of GPUs and you have a dozen or so.. well, he is going to see a lot more of the rare distributions of capacity failures.  But if you are looking for good maintenance practices then I can say:

1) a slow steady burn in of any electronics is always good as compared to slamming it with maximum duty cycle.  When you come back home from vacation, do not unpack... go directly to your basement and turn on the lights so you can see your glorious mining rigs.  Then press the button on your 2-wire power button to your mobo and let it boot into the OS but DO NOT start the miners!  Go unpack.  Come back in an hour or so and make sure all your system is running fine still without any miners running.  

2) Check temperatures at your baseline/no duty cycle.  If there are high temperatures here then sort this out before proceeding.  May need to reboot, check gpu connections, reinstall drivers, or... you may have found a bad component in your build which should be replaced before continuing. 

3) Choose a lower power algorithm such as cryptonite or Lyra2z and then do some test mining on these with NO overclocking.

4) continue watching temperatures.  I recommend a handheld thermal imaging device or a laser temperature gun and point this at various points on you GPUs and the compare temperatures between them.  with equal spacing between cards and a nice controlled environment you shouldn't see very large differences between various GPU components.

5) Stop the miners, activate your MSI afterburner or OC tool you use, make your OC/undervolt/power settings, and start the algo of your choice.

6) Monitor your temperatures and mining performance. Is one card terribly hot or is some other card hashing at 1/2 speed (this happened to me recently)?? If so, I recommend disable/enable GPUs in the device manager and then retrying the OC/mining software.  If it's still weird, then do a reboot and try again.  If it's still weird uninstall/reinstall drivers using the uninstall/cleaning software from GPU guru website.

Good luck!

edit: just joking about "do not unpack"  please don't let the internet control your behavior!

After you come from vacation, you do not need to do something. If your capacitors are in poor condition, they may explode at any time. This happens with very poor capacitors after overheating. Therefore, periodically carefully check the electrolytic capacitors, if they have signs of swelling.

This isnt the 2000s where most Capaciators were blowing up because they were bad quality.

Usually there might be a bad patch on a certain GPU here and there but usually they engineer the capacitors to be able to withstand huge stress.

Mining when undervolted and when properly cooled won't stress the capacitors as much as someone who overvolts and overclocks their GPU inside a badly ventilated case to get the most FPS in games.

Well I am happy to say my rig survived the vacation

Capacitor is the most suffering element in hot temperatures. Especially budget electrolytic capacitors. If anything kills them, that’s probably the temperature (Optimal temperatures - from the dew point (the lowest temperature at which condensation condenses) and up to 21-26C (usually, indicated in the passport).

 Switching it off and back on, can influence other spare parts.

mobos can last for many years  if you run at lower then 100%  and keep the heat down.

maybe your buddy ran gear hard  say at 85% tdp

maybe he had a bad cap to begin with

maybe  his boards did not have good air circulation  so while the card temps were < 70  the board temp was close to 100c

it makes sense if see mining already 1-2 years because not for all capacitor on mobo or GPU can survive for specific times although many things can be influencing it.

caps, the hotter they run shorter lifespan they have, so it is good to keep them cool.

The only thing that you have to do it is don't overheat rigs and sometimes clean them from dust. Also check capacitors for swellings due to the buildup of internal gases.

I think he has a mix of 1070 and 1080.  And that was his mobo that was fired up, to my understanding he has been running it 24/7 for 1-2 years and last month he has to fix something on his rig so he left it off for a week.

Did your friend have old gpus like the Gigabyte Radeon 7950/7970 ?

Because from the LTC days many had issues with those GPUs due to exploding capacitors. Basically they were ceremic capacitors and they had a very high failure rate due to the way they were designed. If you sent your GPU into Gigabyte they basically just desoldered them and the GPU ran fine. This was an isolated incident because it was a design fault.

By turning your rig on and off too many times, its possible to get a slightly higher failure rate. The reason being is because your components heat up and cool off and they experience many of these cycles. One of the problems is that the huge metal heatsink that some GPUs have it expands during heating up and it might stress the solder joints of the GPU chip, and after a while might develop cracks.

However this won't happen after a few power cycles, you would need to do it for months or years.

I think the lesson here is don't make generalizations.

Capacitors of some types are very vulnerable to damage from surge currents ie solid tantalum and aluminum electrolytics. Power on events can, in poorly designed systems, result in large surge currents.


https://nepp.nasa.gov/files/14429/07_270_Teverovsky_%20Reliability%20Effects%20of%20Surge%20Current%20Testing%20of%20Solid%20Tantalum.pdf

electrolytic capacitors come in a range a temperature ratings, commonly 80,105 and 130 C

capacitors are no more susceptible to failure during power up than any other component

Correlation isn't causation. I have 3 mining rigs that over the past year have been shut down for maintenance and/or rearranging several times, including for over 10 days and have yet to have blown a capacitor. I also have a Dell XPS 700 from 2006 that I mined with for months on and off back in 2014 and I still use it as a test bench. Just cause your friend blew a capacitor when he turned the rig back on after an extended time off doesn't mean that's what caused it to go. More likely it was some other reason or was defective.

I agree temp is the killer.  Usually prolonged high heat in most cases.  I did a search on google before posting here, and the results I found weren't conclusive.  One of the 'supportive' argument was the temp change causing capacitors to expand and contract, shorten the life of them.  

But I still yet to see a point explaining sudden failure in my friend's mobo, which happened just 2 weeks ago.

capacitors are rated for 105 C temperatures, probably he dropped his jedi sword on them...

Not sure what you mean?  He sent me a photo of his fired mobo so thats definitely his experience.