Question about: fan speed and temperatures (S17)

[addalol]

Im wondering a bit about fan speeds and temperatures, on a antminer S17.

So I have had my miner running for about a month.

The miner has been stady on 80/80/75C with fan speeds of: FAN1: 2600-3000 FAN2: 2600-3000 FAN3: 3500-4000 FAN4 3500-4000. Temperature in the room where the miner sits is about 17C, but temp can fluxuate(MAX 23C), which only affects the fan speeds, the chip temps always stay the same. The colder it gets, fans just adjust down RPM so that the miner sits at a stedy 80/80/75C.

So my question is, are 80/80/75-C to high temperatures? what are good operating temperatures and fan speeds. Also wondering which fan is front and back? is it fan1 and fan2 is intake, fan3 and fan4 exhaust?

I experimented with customized fanspeeds @55% fan speed and @50% fan speed

https://imgur.com/a/7Bsdk8e

Should I customize the fanspeed to keep chips lower temps - or - dont run fans to hard, and get a more silent miner. (Im finding it a bit to high db at fan speed @ 50% but)

Can i run the fans at even lower speeds 40-45% range and still have "responsible" temperatures?

Regards

[1713580610]

1713580610

[1713580610]

1713580610

[1713580610]

1713580610

[1713580610]

1713580610

[philipma1957]

try to see if you can do 3000 rpm and stay under 80c   75c  for instance is okay.

[mikeywith]

Can i run the fans at even lower speeds 40-45% range and still have "responsible" temperatures?

This is quite a subjective topic, some people think that as long as the miner runs under the maximum operation temperature you are good, others say the cooler the better and that heat will shorten the life-span of any chip, another group agrees with the latter but say that the chip's economical life-span is shorter than it's technical life-snap anyway.

I personally think the cooler the better  , now according to Bitmain these are the max temps for S17

Code:

S17/S17 Pro/T17

PCB max. 75

you have not reported the PCB temp, but you can compare it against 75c, if it's a bit far then you are safe,however, if noise is not an issue I would put the fans on 90% (this is what I do to almost every miner i have). Fans are dirt cheap, hash boards are not.

Also wondering which fan is front and back?

The once that run at higher RPMs are most often the intake, so in your case fan3 and fan4 are exhaust, you can also check the fan's wire, on the control board it states which number the fan belongs to.

[thierry4wd]

One point of corse ...

More temperature:
- More power consum
- Down life of chips

Down temperature:
- Down power (more efficiency)
- Up life of chips

So , on my experience, my T9+ run above 105c at all time arround 2 years ! no problem ! is just info but i thinks is not recommended... but for me, no problem ! lol

[mikeywith]

More temperature:
- More power consum

Are you sure about this? I know that more power consumption = more heat , but I don't think the opposite is necessary correct.

To my understanding, heat increases resistance and assuming Volts and Watts are static, only amps will change but Wattage will remain the same as long voltage is since W = V*A

So i doubt that an antminer S17 the runs at 70c consumes more power than another S17 that runs at 60c!!

please correct me if i am wrong.

Thanks.

[judypug1956]

You could be right or wrong.

Enough heat and the psu becomes less efficient.

Bitmain has been better with heat then most other gear.

But at a certain heat it will lose some efficiency.

I think I can find a chart. On bitmain website.

https://shop.bitmain.com/product/detail?pid=0002019072316001724716dkNtX50679

the psu has an op range of 20c to 60c

I know there are some efficiency charts

found them for the apw7

https://shop.bitmain.com/product/detail?pid=00020190918160626197jS1ejsgA06DC

as you max out the  power efficiency drops from 95 to  93  chart is on the left.

most psu's have this issue as you approach max power bitmain has done a decent job with many of their newer psu's
I have found that the apw3++ are bulletproof at 1200 watts but when you push them close to 1600 watts they have issues.

[mikeywith]

This is a bit off-topic, you are confusing PSU operation temps and chip temps, to make this a bit clearer, imagine that psu sits in a different room from where the miner is, now you have the following

Miner 1:

Antminer s17 running at chip temp of 80c

Miner 2:

Antminer s17 running at chip temp of 60c

again, both PSUs are far from the miners, same distance, same ambient temperature, same frequency, same voltage, same fan RPM and same just about everything, but for whatever reason one miner runs hotter than the other.

Will miner number 1 consume more power than miner 2 just because it runs hotter?

[philipma1957]

No I am not confusing them the odds are if you have a  hotter pcb  the s17 psu is laying against the pcb boards thus causing more watts usage due to a hotter psu.

Though the question is more complicated then this.

I get back and write it more clearly.

[mikeywith]

No I am not confusing them the odds are if you have a  hotter pcb  the s17 psu is laying against the pcb boards thus causing more watts usage due to a hotter psu.

Though the question is more complicated then this.

I get back and write it more clearly.

I understand your point, I think it would be best to consider using another miner for an example we can use S9, I think the reason why you said what you said is because of the S17's design (Which i forgot about completely!), the psu is too close to the miner itself, and it makes sense that if the chips run at higher temps, heat will transfer to the PSU, make it less efficient and cause it to draw more power.

That is correct, but remember the miner will consume more power not because the chips got hot, it's because the miner did, in other words if you could move the psu away from the miner, say like in all other old gen miners like an S9, say you have a meter long 6pin connectors whereby you leave noway for the heat coming from the miner to transfer to the PSU, will that PSU consume more energy when the miner runs hotter?

[philipma1957]

ahh  it is complex  I  would say  that  each pcb  board would be stand by itself.

They vary 2-4% in most cases.  But your points are valid if everything was perfectly binned  you would be correct all the time.

When I first got into this  everything was gpu  and they could be same make and model  and one would down clock great  the other would over clock.

I know I have 16x s17 or s17 pro  here are 3 in a screen shot.
All in a room with very good air flow all set to  turbo
nine boards all same firmware all set to turbo

one at  575
one at  590
one at  591
three at 595
three at 605

lot of variance for identical gear at  same settings.

one 605 board is hotter 77c  fans are faster  miner address of .163

two 605 boards are cooler fans are slower   miner address of .160

suppose his gear  has 3 hotter  boards  like that one 77c   605 freq faster fans

suppose his gear has 3 cooler boards like that one  70 c   065 freq  slower fans

[mikeywith]

I am interested in seeing the power consumption on the two gears that run at different temps but same fan speed (the top and bottom images), the only problem here is that they don't run at exact same frequency which renders the results inaccurate.

a perfect test would be using fixed frequency and voltage on two gears, setting all fans at static % , and then force one miner to run hotter by slightly blocking the air circulation or any other method, and then see how exactly heat causes the miner to consume more power, it would be great to know if the difference will make up for running all fans at 100% to achieve the following:

more power for fans > less heat > less overall power consumption due to temp reduction.

[thierry4wd]

Of corse, i confirm this !

for exemple , my T9+ at 10.5TH

chips at 105c = 1170w arround
chips at 80c = 950w arround

And bitmain confirm this !!! just look exemple for T17

Power efficiency on wall @25°C, J/TH     55.00    　    58.85
Power efficiency on wall @40°C, J/TH     58.26    　    62.34


so , just try this !!! run S9 with fan full speed , look temp after 3min (stabilise temp) and run with same config but config the fan at low speed (to force up temp) ... and look power is more !!!

Or for test, run S9 with fixed fan arround 20% ... at startup ... just look the power up every segonde !


So ! but this theorie is only true with fixed fan ... at autofan, the chips temp is regulated, the ambiant temp is 10 or 40c , just the fan speed change but the chips temp is arround same value! so, the power consum is only difered by fan power...

Is valable at all miner, if you run this on lowtemp environement, run this with fixed fan (upper than original) for force down temp chips, and seed the efficiency, is better

[mikeywith]

Of corse, i confirm this !

for exemple , my T9+ at 10.5TH

chips at 105c = 1170w arround
chips at 80c = 950w arround

What is the frequency ?  I assume the 105c runs at high freqncy this is why it runs way too hot and consumes more power, not just because it's hot, please confirm the frequency and let me know.

And bitmain confirm this !!! just look exemple for T17
Power efficiency on wall @25°C, J/TH     55.00    　    58.85
Power efficiency on wall @40°C, J/TH     58.26    　    62.34

That is PSU related, the cooler the environment, the cooler and the more efficient the PSU is.

Is valable at all miner, if you run this on lowtemp environement, run this with fixed fan (upper than original) for force down temp chips, and seed the efficiency, is better

it makes sense, only if we are sure that lowering the temps will save more power than the fans at higher speed use, have you actually tested it this way?

[thierry4wd]

No , No , and NO !!! lol

T9+ at 10,5th is original freq, so 575mhz, but at 790mv (with my firmware) ...

The power efficiency is not full link by PSU ...

I assure my say is correct and confirmed ! i have habit this ...

If you want test by easy methode ... juste make your hand on "intake" fan of miner... monitoring the temp chips, normaly is higher every arround 10s , and see your wattmeter

Please confidence of me


PS : honestly i thought everyone knew that!

[mikeywith]

Please confidence of me

I do,  I really believe you thierry4wd , I know you would not "lie" about something like that, my concern was that you might have been getting these results due to a mistake and not by intention, but now that you confirmed both miners run at same frequency and voltage then the debate is over and you won  .

Ok, another question since you seem to have done a good amount of testing regarding this subject, what would be better in terms of power efficiency.

1- Run fans at high speed > consume more power by fans > less heat > save power because of less heat.

or

2- Run fans at slow speed > consume less power by fans > more heat > more power because of more heat.

[thierry4wd]

No problem ! 

So , my test is approuved by S7 / D3 / T9+ and S9s ... i thinks is same this with all miners...

for the T9+ and S9 of corse the best efficience is force litle up fan than stock ... i thinks arround 80% (at arround 20c temp ambient)

[Ripster]

How do we change fan speed with the latest bitmain firmware on the T17's, im not seeing any settings fans staying very low like 2800-3200 rpm chips hit 80 some times 85 degress, fans wont speed up at all, anything to worry about? Will it speed the fans at a higher temp?

[mikeywith]

How do we change fan speed with the latest bitmain firmware on the T17's...

I was actually trying some keywords on Google regarding this, I wanted to check if my topic would show up on the first page and to my surprise, yours showed up and mine did not so allow me to answer your question by posting a link to my topic which has a simple guide on how to change fan speed for Antminer T17 and all other Antminer models from S9 to S17.