Strange, PSU started malfunctioning

[riverdip]

I have mining farm and i have 1 PSU going down in every 2 days.

PSU i am using :
Chinese 1800 Watt

things to consider:
My power cuts
Cold whether in night
MCB tripping
Second Hand PSU

UPDATE : I opened 2 PSU and found out a part named ntc thermistor blown ! what could be the reason ?

Out of 200 PSU, 30 have gone down in past 2 months !

What Could be the reason ?

[1714295204]

1714295204

[Steamtyme]

When you say gone down do they outright die, or do they need to be power cycled? I'm curious what you mean about your considerations -
Your power cuts out semi regularly? If so repeated power cycling isn't good for anything so overtime that can cause issues in itself especially when it's restored and is pulling full draw again immediately.
How low do your temperatures drop overnight? Does your area get really foggy this time of year? You could either be dropping below safe operating limits or experiencing a moisture issue depending on your set up.
Are your breakers tripped? You might have more of a load issue with the circuit as opposed to a PSU issue.
Yeah, second hand PSU's can be an issue but so can new ones. You never know how long they ran or what issues they had previously.

We probably need a lot more details to help out. Try giving us the run down on your whole operation and details of exactly what is going wrong. Could be multiple issues.

[riverdip]

When you say gone down do they outright die, or do they need to be power cycled? I'm curious what you mean about your considerations -
Your power cuts out semi regularly? If so repeated power cycling isn't good for anything so overtime that can cause issues in itself especially when it's restored and is pulling full draw again immediately.
How low do your temperatures drop overnight? Does your area get really foggy this time of year? You could either be dropping below safe operating limits or experiencing a moisture issue depending on your set up.
Are your breakers tripped? You might have more of a load issue with the circuit as opposed to a PSU issue.
Yeah, second hand PSU's can be an issue but so can new ones. You never know how long they ran or what issues they had previously.

We probably need a lot more details to help out. Try giving us the run down on your whole operation and details of exactly what is going wrong. Could be multiple issues.

i opened PSU and found a part named : ntc thermistor blown out, what could be reason ?

[wndsnb]

A thermistor is used to measure temperature, in a PSU it would be used as a safety measure to shut itself down if it isn't getting enough airflow or ambient temp. is too high, or to control the speed of the fan, or both. It is possible the part just failed and that's your problem, but it is also possible that something else failed in the PSU that caused the voltage across the thermistor to go too high, frying it.

You could try replacing the part, and see if it works. I would not hook it up to any equipment you care about before measuring the output with a volt meter though.

Are the miners that were using those failed PSUs still operational (ie did you test them with working PSUs after they went down)?

[NotFuzzyWarm]

Thermistors are not only used for temperature sensing but also as current surge protectors. PTC (Positive Temperature Coefficient) thermistors are temp sensors, their resistance rises as temp increases. NTC (Negative Temperature Coefficient) ones are generally surge protectors, specifically, inrush current limiters. They start at certain (higher) resistance when cold to limit startup currents and as they get slightly warm from power flowing through them their resistance goes down to nearly zero to supply full power to a circuit.

In short - where is the bad thermistor located, near the AC power inlet and large bulk capacitors? If so it's a surge limiter.

[wndsnb]

Thermistors are not only used for temperature sensing but also as surge protectors. PTC (Positive Temperature Coefficient) thermistors are temp sensors, their resistance rises as temp increases. NTC (Negative Temperature Coefficient) ones are generally surge protectors, they start at certain (higher) resistance when cold to limit startup currents and as they get slightly warm from power flowing through them their resistance goes down to nearly zero to supply full power to a circuit.

In short - where is the bad thermistor located, near the AC power inlet and large bulk capacitors? If so it's a surge limiter.

That makes much more sense, and if it was a used as a sensor it probably would have been too small for the OP to get the part number off of. The inrush current limiter variety for an 1800W power supply is going to be pretty large.

Any recent lightning storms?

[riverdip]

That makes much more sense, and if it was a used as a sensor it probably would have been too small for the OP to get the part number off of. The inrush current limiter variety for an 1800W power supply is going to be pretty large.

Any recent lightning storms?

so you put it on current surge ? is so then is a device exist that would add startup delay when miners boot up after a power cut, i have my own transformer, 5 METER Earthing for all !

Thermistors are not only used for temperature sensing but also as surge protectors. PTC (Positive Temperature Coefficient) thermistors are temp sensors, their resistance rises as temp increases. NTC (Negative Temperature Coefficient) ones are generally surge protectors, they start at certain (higher) resistance when cold to limit startup currents and as they get slightly warm from power flowing through them their resistance goes down to nearly zero to supply full power to a circuit.

In short - where is the bad thermistor located, near the AC power inlet and large bulk capacitors? If so it's a surge limiter.

Hmm so sruge is causing this? no temperate issue my ASIC runs at 70-75 Range its well ventilated setup !

Are there any way to cause startup delay ? when power hits after power cut ?

[wndsnb]

Yeah, that last photo looks like the size of an inrush limiter for a power supply of that size. I'm not sure what would cause it to fail though. As NotFuzzyWarm said, once the PSU is on and running, the resistance of the thermistor would be very low, so it would require a very large amount of current to smoke it once it had fully turned on. I'd think it would be more at risk during power up when the resistance of the thermistor is high. If you turned the supply on when the input voltage was too high it could fry.

[NotFuzzyWarm]

Ja ^^
Most of the time when I've seen this what actually happened is that the AC input rectifiers shorted and/or one of the big raw DC filter caps shorted which directly overloaded the inrush limiter. In that case, think of it as a last-resort 'fuse' to keep the fireworks to a minimum   And yes, high line voltage surges will take out rectifiers/caps in a heartbeat if the mfgr did not properly de-rate them.

3rd suspect would be the actual switching MOSFET(s) but in even a half way decently built supply that should be rare.

[philipma1957]

He mentioned these are used psu’s 1800 watt.

It could be last user beat on them harder then he does.

It they are three years old and were pushed for years they will show signs of age.

[mikeywith]

so you put it on current surge ? is so then is a device exist that would add startup delay when miners boot up after a power cut...

There is delay start devices you could use, but using it for each miner will cost you a fortune, buying a breaker with delay-on function built-in which is enough for 200 miners is also going to be a bad move, your best option as far as I know is to buy a cheap relay such as >https://www.amazon.com/GRT8-M1-Delay-Multifunctional-Functions-12V-240V/dp/B07L689B3H

and then get a contactor that matches the size of your main breaker/s , another option would be those cheap wireless switches such as https://www.amazon.com/Wireless-Universal-Automation-Compatible-Assistant/dp/B07P335LN3

The product description does not mention that it has a delay-start ,but the app that controls it does have that function, I have tried them, they are also compatible with ALEXA, and the cool thing about these switches is the ability to control them from home , again you will need a contactor behind them, or buy 200 pcs of these (which is a stupid idea) but you might want to be able to control every miner individually.

I also want to add that PSU's quality is crucial, bitmain's apw3/apw7 have been working fine for me, and even in locations where I have no voltage regulators and did have some surges, I have only lost about 2-4 bitmain psus, there are better quality PSUs in china , but the majority are trash, so this should serve as lesson for everyone who wants to save a few dollars on a power supply.

[wndsnb]

It looks like there is discoloration on the heatsink near what might be the input rectifier.

Ja ^^
Most of the time when I've seen this what actually happened is that the AC input rectifiers shorted and or one of the big raw DC filter caps shorted which directly overloaded the inrush limiter. In that case, think of it as a last-resort 'fuse' to keep the fireworks to a minimum   And yes, high line voltage surges will take out rectifiers/caps out in a heartbeat if the mfgr did not properly de-rate them.

3rd suspect would be the actual switching MOSFET(s) but in even a half way decently built supply that should be rare.

[philipma1957]

[...]

Yeah bitmain apw3 = good  97% maybe more still work.  1 dead one  and at least 40 good ones.

An my mine has a low voltage issue  we tend  to stay at 223-229 volts ,but the three phase  transformer does drop  down to 185-191 volts one or two times a month.  that is very hard on the gear.  Causes drop outs  but all my s9's and l3+ units come back on line when power comes up to 224.

I have been  looking into a solution  I found some volatge units that are good at the 3000 watts level.  got 4 of them  to protect the more expensive miners.

Running

an s17  2225 watts

an inno 39 2200 watts

an inno 50  3100 watts

and a t17     2100 watts   these buck boost regulators   are this one

https://www.amazon.com/Norstar-DAVR-5000110-transformer-Automatic-Regulator/dp/B07169Z3XX

and 3 of these

http://www.poshouse.com/item.asp?PID=37538&FID=1236&level=1

note I got a very good price on these units  well under 300 each.  they are far better built then the norstar davr-5000

http://www.smartpowersystems.com/wp-content/uploads/2014/02/AVRPlusUsersGuide1.pdf

[riverdip]

Happened Again, Lost Another one of the PSU.

Things Common : NTC 20D-20 OHM blown out

No MCB tripped, Happened when electricity came back after power cut !

Ja ^^
Most of the time when I've seen this what actually happened is that the AC input rectifiers shorted and/or one of the big raw DC filter caps shorted which directly overloaded the inrush limiter. In that case, think of it as a last-resort 'fuse' to keep the fireworks to a minimum  And yes, high line voltage surges will take out rectifiers/caps in a heartbeat if the mfgr did not properly de-rate them.

3rd suspect would be the actual switching MOSFET(s) but in even a half way decently built supply that should be rare.

makes sense, what can we do to limit inrush current coming from transformer ?

[wndsnb]

Sounds like a voltage spike when the power was restored.

A protection relay with a delayed start might help.

https://www.icmcontrols.com/productdetails/ICM491

https://www.amazon.com/ICM-Controls-ICM491-Adjustable-Protection/dp/B000LEVBF8

Or you could just shut down all of your breakers when the power goes out, and wait a bit after it is restored before turning them back on.

[Artemis3]

Sounds like a voltage spike when the power was restored.

A protection relay with a delayed start might help.

Or you could just shut down all of your breakers when the power goes out, and wait a bit after it is restored before turning them back on.

Lately this is happening frequently in my place. Power goes out, and if it returns quickly, it comes with overvoltage, and sometimes goes undervoltage too, while the rest of the grid stabilizes. It usually means some other area of the city has lost power (when its our area, it does't come back quick, but hours later).

If you don't use these types of power protection devices, you lose your household appliances. Refrigerator being no.1 thing to protect, and next anything else like computers and TVs. It wasn't so bad until this year after the great countrywide blackout from March. And its much much worse outside this city...

When this happens so frequently, you just don't have time go to the breaker panel. Besides breakers are not supposed to be used like switches.

The simplest devices sense the power fluctuation, or the over/under voltage condition and simply cut power until things go back to normal. I have seen some APC ups do the same when they run out of battery (if the battery is good, it uses that as source instead).

Else is transforming the power, they call them voltage regulator here, no matter what input (within a range), output is the same (ie. 220v). I have one of those things and is able to power an S9. Funny thing is that was always fed about 203v. There are also smaller variants of the same thing for computers and electronic appliances, i have 3 of those for the computers.

Variable time delay is important if you have many things you need to turn back ON, so they all don't go ON at the same time, causing a sudden spike load.

[riverdip]

its all good now, i added timer based MCB and its all fine.

let me know if you need the guide !