Big difference in S9 temps - same room

[boontis]

I just got 2 S9's setup and running - but am seeing what I would consider wildly different chip temps between the two:

First one - 82/76/79

Second one - 100/96/96

Those were temps without the units hooked up to my 5" ducting - when I hooked them up to that the first one went up to 88/82/84 (which I think is still well within normal range) but the second one went up to 118/105/109 which is when I took the duct off and it returned down to the ~100 range.

I have not altered any of the settings and actually did a factory reset on the second one to see if the temps would come down but they do not.

Any ideas on why the one is running so hot?

One thing I did notice is that I never saw the fan 6 get above 6000k rpm with the second one...whereas with the first one it regularly gets up above that.

Any help would be much appreciated...for now I'm only running the first one just in case

[1714111028]

1714111028

[1714111028]

1714111028

[1714111028]

1714111028

[Geronimoe]

You could try swapping the fans from your first S9 to the second S9 to rule out faulty fan unit

[SGMPhil]

You might consider cleaning the one that is running hot - it could have some dust/dirt accumulation if it is not a new unit. Also, you could have a failing fan.

BTW, I have one S9 that consistently runs cooler that all my others. They are all configured the same and I have been trying to figure out why that one runs cooler. I have found that the rear fan on that one is different than the others but I have not taken it apart to see what model fan it has on the rear. I am going to do that and consider replacing the rear fans on my others to match that one as it runs ~10 deg cooler that the others.

[OgNasty]

Are they both running the same firmware version?

[Raymond_B]

Also they might be in the same room, but is one picking up the hot air from the other or is it not getting any fresh incoming air?

I had a PSU slightly pulled forward maybe 1"-2" on one of my miner's fan. Sliding it back changed the miner temp by 2 deg C. They need clear unobstructed intake areas.

[ccgllc]

Please provide a photo of your setup so we can visualize what your doing. 

This is a classic case of a photo being worth a thousand words.

[boontis]

Thanks for the help & suggestions!

Here are a couple pics of the setup:

https://www.dropbox.com/s/cjbuodbsxumilpo/2018-02-20%2020.52.44.jpg?dl=0 (no ducting)

https://www.dropbox.com/s/judzemft27ci9jr/2018-02-22%2021.12.12.jpg?dl=0 (partial ducting just so you can see what I was doing)

I've been doing some testing today with NO ducting and here is what I found:

-ran them both with no duct just free air ambient temp below 20c
-swapped power supplies - no diff
-swapped locations - no diff
-tested fans on the hot one - both working normally
-tried the hot one on 110 - miner started but the power cable I had was getting so hot I couldn’t let the unit boot up completely - like it got real hot to touch and started to smell like burning so I unplugged it
-both are on 16.8.1.3

Couple things to note:

1 - the cool one’s fan seems to spin up and then slow down every 30-60 seconds - up to 6120rpm on fan 6 then back down to 4680. Does this consistently always.  The hot one sits at 4680 both fans all the time.

2 - the cool one when I was moving it around sounded like there was lose cards in it - like if you jiggle it it sounded loose. The hot one doesn’t make any sound.

So I’m no further ahead or behind where I was yesterday 😭

Any other advice would be much appreciated

[ccgllc]

Photos are good thanks!

1st photo - air is mixing and the miner on top is eating hot air from the miner below.  Heat rises, so the lower miner gets the cooling room air, the upper miner gets a mix.

2nd photo - back-pressure from the Y.  If you removed the Y and vented both outdoors I think you would achieve your objective.

[boontis]

Photos are good thanks!

1st photo - air is mixing and the miner on top is eating hot air from the miner below.  Heat rises, so the lower miner gets the cooling room air, the upper miner gets a mix.

2nd photo - back-pressure from the Y.  If you removed the Y and vented both outdoors I think you would achieve your objective.

I thought of that too - and put the hot one on the bottom - temps stayed the same though. (with no duct hooked up at all)  The hot one, even on the bottom, runs ~20c hotter than the other one.  I also tried only running the hot one by itself and the temps are still at ~100.  Room temp is below 20c all the time (there's alot of airspace and concrete floor so that room is quite cool all the time).

[philipma1957]

Photos are good thanks!

1st photo - air is mixing and the miner on top is eating hot air from the miner below.  Heat rises, so the lower miner gets the cooling room air, the upper miner gets a mix.

2nd photo - back-pressure from the Y.  If you removed the Y and vented both outdoors I think you would achieve your objective.

I thought of that too - and put the hot one on the bottom - temps stayed the same though. (with no duct hooked up at all)  The hot one, even on the bottom, runs ~20c hotter than the other one.  I also tried only running the hot one by itself and the temps are still at ~100.  Room temp is below 20c all the time (there's alot of airspace and concrete floor so that room is quite cool all the time).

switch the fans

And screen shot the firmware so we can see if it is the same

[BitMaxz]

Photos are good thanks!

1st photo - air is mixing and the miner on top is eating hot air from the miner below.  Heat rises, so the lower miner gets the cooling room air, the upper miner gets a mix.

2nd photo - back-pressure from the Y.  If you removed the Y and vented both outdoors I think you would achieve your objective.

I thought of that too - and put the hot one on the bottom - temps stayed the same though. (with no duct hooked up at all)  The hot one, even on the bottom, runs ~20c hotter than the other one.  I also tried only running the hot one by itself and the temps are still at ~100.  Room temp is below 20c all the time (there's alot of airspace and concrete floor so that room is quite cool all the time).

Can you check the firmware of both miner because there is miner with a fix frequency and manual settings for frequency and fan...
And sometimes its firmware issue or sometimes its hardware issue where the heatsink is not properly glued or dropped off from every chips.
Can you try to switch the psu from the one miner to another one.? and check if there are changes..

Also just add this bitmain trouble shooting guide if ever you need it .
https://support.bitmain.com/hc/en-us/articles/220872007-Possible-problems-for-Antminer-Troubleshooting-for-S7-S9-L3-

[ccgllc]

I thought of that too - and put the hot one on the bottom - temps stayed the same though. (with no duct hooked up at all)  The hot one, even on the bottom, runs ~20c hotter than the other one.  I also tried only running the hot one by itself and the temps are still at ~100.  Room temp is below 20c all the time (there's alot of airspace and concrete floor so that room is quite cool all the time).

In that case, I'd say your having a cooling problem with the hot unit.  It could be:

1) A dead or underperforming fan  (what speeds are being reported?)
2) Dirt on the boards clogging the heatsinks or otherwise obstructing air flow
3) Missing or lose heatsinks

Suggest focusing this conversation on that hot unit for now.

If this unit is still under warranty, you MIGHT violate that by opening the box.  If its not:

A)  Take off the fans and plug them into a fan port on any CPU motherboard and see what happens.
B)  Open the case and pull the hash cards to visually inspect them

[Raymond_B]

2 - the cool one when I was moving it around sounded like there was lose cards in it - like if you jiggle it it sounded loose. The hot one doesn’t make any sound.

The hash boards are only in slots in the aluminum case, so if you jiggle the miner you SHOULD hear the cards move a bit. Sounds like something is in there or some heatsinks fell off and are wedged in between boards. My S7 sucked in a giant moth which made it run quite hot. I kick myself for not taking picks, but he was BBQ'd on one of the heatsinks.

Now fully disassembling the unit will void the warranty, but I believe Bitmain has advice on how to check the heatsinks.

[boontis]

Thank you all for your comments/feedback/suggestions!

I looked again and realized that the firmware - which I thought was the same - was actually the hardware version. (16.8.1.3)  I flashed firmware on the hot one and after it rebooted I saw that the file system version date is now the same as the cool one. 

And now the temps are coming in the low 70's across the board!

I'm sorry for missing such an obvious thing - and appreciate the help.  Considering both of the S9's were part of the January batch I just assumed they would be on the same firmware?  I guess I just learned my first lesson in mining...never assume anything

[HoleShot]

I noticed they added a fix in the latest firmware to correct an error. They were looking for a middle temp sensor which doesn't exist on these hash boards. The fix applies an offset to that missing chip to take the variable to minimum instead of maximum as before.

Oh, the warranty sucks so I fully disassemble every new miner before I power it up. Failures are just too costly to risk it.

[SGMPhil]

Thank you all for your comments/feedback/suggestions!

I looked again and realized that the firmware - which I thought was the same - was actually the hardware version. (16.8.1.3)  I flashed firmware on the hot one and after it rebooted I saw that the file system version date is now the same as the cool one. 

And now the temps are coming in the low 70's across the board!

I'm sorry for missing such an obvious thing - and appreciate the help.  Considering both of the S9's were part of the January batch I just assumed they would be on the same firmware?  I guess I just learned my first lesson in mining...never assume anything

Good that you got it figured out and that your problem is solved. Interesting how firmware can affect these things.

[boontis]

Ya I would have thought temps were temps regardless of firmware?  Makes me wonder now - was the unit actually overheating or running hot previously or just *reporting* hot?  I only let it run for a couple hours at a time as I didn't want to risk permanent damage so either way I'm hoping it's all clear sailing from here on!