Bitmain S5 Replacement Fan

[Evan]

Push Pull? or stacked?

Just 1 delta as a push. It's specs were close to the stock fan, so I was hoping it would do the job. After about 1h of testing it doesn't look awful, however one S5 is overheating only on one side (the other one is proportionate). Not sure what could be causing it - the fans are identical and were installed exactly the same way.

Can I see Pictures

I had to remove the black screws to get the Delta to fit flush. One unit running 60C/60C, the other 63C/55C, still not sure why it's uneven. The heatsinks got slightly bent while I was taking it in/out, but I got them almost back into the stock position. Not sure if the heatsinks would be causing such a temperature imbalance?

Running open case?

[1714065160]

1714065160

[1714065160]

1714065160

[1714065160]

1714065160

[amelen]

Yes, open case. Going to try the Delta as push and stock as pull, but I think that will use more power than I wanted and will be louder than stock.

[Evan]

Yes, open case. Going to try the Delta as push and stock as pull, but I think that will use more power than I wanted and will be louder than stock.

No idea, then Closed case will make them run coolest as that forces air where it needs to be

[amelen]

Yes, open case. Going to try the Delta as push and stock as pull, but I think that will use more power than I wanted and will be louder than stock.

No idea, then Closed case will make them run coolest as that forces air where it needs to be

What's the cheapest/easiest way to close the case up?

[Evan]

Yes, open case. Going to try the Delta as push and stock as pull, but I think that will use more power than I wanted and will be louder than stock.

No idea, then Closed case will make them run coolest as that forces air where it needs to be

What's the cheapest/easiest way to close the case up?

Cardboard and Duct tape......

[amelen]

Yes, open case. Going to try the Delta as push and stock as pull, but I think that will use more power than I wanted and will be louder than stock.

No idea, then Closed case will make them run coolest as that forces air where it needs to be

What's the cheapest/easiest way to close the case up?

Cardboard and Duct tape......

Won't catch fire? Has anyone tried this and seen a decrease in temps?

[MarkAz]

On the type of screws, the fans are these (pick whatever length you want, this is what I use with a spacer):

http://amzn.to/1FLWefa

and FWIW, these are what the rest of the case are:

http://amzn.to/1AYeera

As far as temps - the first and easiest thing to do is add a spacer between both the push and pull fans and the case - doing this dropped my temps by about 2c.

As far as enclosure, I made one this weekend that is very nice on my laser cutter, and did a bunch of testing with my thermal camera to see how it changed temperatures, etc - and I will tell you that it won't drop your temperatures anything really in the standard configuration.  What you do gain are some alternative configs - so for instance, when you look at the thermal characteristics of the board when running, the front (where the fan is) is substantially cooler than the back - which makes sense, the front is getting fresh cool air, while the back just gets what it gets.  When you tweak it to push/pull, then it gets better overall, since the case is so open the rear fan sucks in fresh air from all the gaps in the case.  If you put it in a full enclosure though, now you might see better indicated temperature from their onboard sensor due to better airflow - but the overall board temps go up slightly.  I also found that push/pull becomes far less effective in this case, and actually found pull-only to be better than push-only... This has the added bonus of making the intake exposed, so easier to clean the grill and stop debris from building up (that I've seen on several used S5's I've purchased).  I'll be posting a thread at some point in the next few days with a bunch of pictures for anyone else who is interested. 

I should also say I'm doing all my testing using the Ultra Kaze:

http://amzn.to/1cH0rta

My fan of choice for the S5 - I think it's a good balance between noise and airflow, and I prefer an always on fan than a PWM-controlled fan.  That's not to say this fan is even remotely close to the performance of the stock fan - they're hard to beat, or more accurately, I haven't been able to beat them yet.

[amelen]

On the type of screws, the fans are these (pick whatever length you want, this is what I use with a spacer):

http://[Suspicious link removed]/1FLWefa

and FWIW, these are what the rest of the case are:

http://[Suspicious link removed]/1AYeera

As far as temps - the first and easiest thing to do is add a spacer between both the push and pull fans and the case - doing this dropped my temps by about 2c.

As far as enclosure, I made one this weekend that is very nice on my laser cutter, and did a bunch of testing with my thermal camera to see how it changed temperatures, etc - and I will tell you that it won't drop your temperatures anything really in the standard configuration.  What you do gain are some alternative configs - so for instance, when you look at the thermal characteristics of the board when running, the front (where the fan is) is substantially cooler than the back - which makes sense, the front is getting fresh cool air, while the back just gets what it gets.  When you tweak it to push/pull, then it gets better overall, since the case is so open the rear fan sucks in fresh air from all the gaps in the case.  If you put it in a full enclosure though, now you might see better indicated temperature from their onboard sensor due to better airflow - but the overall board temps go up slightly.  I also found that push/pull becomes far less effective in this case, and actually found pull-only to be better than push-only... This has the added bonus of making the intake exposed, so easier to clean the grill and stop debris from building up (that I've seen on several used S5's I've purchased).  I'll be posting a thread at some point in the next few days with a bunch of pictures for anyone else who is interested. 

I should also say I'm doing all my testing using the Ultra Kaze:

http://[Suspicious link removed]/1cH0rta

My fan of choice for the S5 - I think it's a good balance between noise and airflow, and I prefer an always on fan than a PWM-controlled fan.  That's not to say this fan is even remotely close to the performance of the stock fan - they're hard to beat, or more accurately, I haven't been able to beat them yet.

You're using 2 Ultra Kaze's per S5? What are your temps like?

Have you tested pull-only with the open case? That was one thing I wanted to test but didn't get a chance to. I wonder if it would cool better than push.

[amelen]

I use this:  two pieces per S5
http://www.noctua.at/main.php?show=productview&products_id=78&lng=en


Size      120x120x25 mm
Connector      4-pin PWM
Bearing      SSO2
Blade Geometry      Heptaperf™
Frame Technology      FocusedFlow™
Rotational Speed (+/- 10%)      2000 RPM
Min. Rotational Speed (PWM, +/-20%)      450 RPM
Airflow      121,8 m³/h
Acoustical Noise      29,7 dB(A)
Static Pressure      3,94 mm H2O
Max. Input Power      1,2 W
Max. Input Current      0,1 A
Voltage      12 V
MTBF      > 150.000 h
Scope of Delivery      
Fan
4 Fan Screws
Warranty      6 years
Ingress Protection      IP52
Material      Fibre-glass reinforced polyamide
Motor type      Three-phase

But if you want more cool so I recommend this already but the noise level is insufferable:
http://www.noctua.at/main.php?show=productview&products_id=80&lng=en&set=1


Size      120x120x25 mm
Connector      4-pin PWM
Bearing      SSO2
Blade Geometry      Heptaperf™
Frame Technology      FocusedFlow™
Rotational Speed (+/- 10%)      3000 RPM
Min. Rotational Speed (PWM, +/-20%)      750 RPM
Airflow      186,7 m³/h
Acoustical Noise      43,5 dB(A)
Static Pressure      7,63 mm H2O
Max. Input Power      3,6 W
Max. Input Current      0,3 A
Voltage      12 V
MTBF      > 150.000 h
Scope of Delivery      
Fan
4 Fan Screws
Warranty      6 years
Ingress Protection      IP52
Material      Fibre-glass reinforced polyamide
Motor type      Three-phase

These fans are expensive but indestructible.
Corsair example, I broke when I accidentally tapped a finger in the propeller.
when I have tapped into Noctua, much it hurts and fans held as of stone.

What temp's are you seeing with those fans?

55 degrees Celsius in the room where it is 22 degrees or less.
but even if they increased the room temperature at 30 is really not more than 65 on S5.

55C at stock speeds, right?

[amelen]

The delta fan is holding out pretty well - but temps in the back aren't as low as they were with the stock fan. I wonder if there is any benefit from throwing on a cheap low-voltage fan on the back to pull? Would I be able to save 2-3C by adding a pull like this in the back?

http://www.newegg.com/Product/Product.aspx?Item=N82E16835200073

Once I'm done testing, I'll be selling the Delta's and stock fans, so if anyone is interested please pm me (have 2 of each).

[MarkAz]

You're using 2 Ultra Kaze's per S5? What are your temps like?

Have you tested pull-only with the open case? That was one thing I wanted to test but didn't get a chance to. I wonder if it would cool better than push.

I have several S5's running in the same conditions, so I don't generally bother with the base temps (and it's actually relatively warm by default where I'm testing), just the difference between units right next to eachother - here's some of the general numbers :

Stock fan:                                   ~54c
Kaze:                                         ~66c
Kaze + spacer:                            ~63c
2x Kaze Push/Pull + spacers:         ~57c
Kaze (Push) + spacer + enclosure: ~65c
Kaze (Pull) + spacer + enclosure:  ~61c

The focus when I was doing the tests was to experiment with ducting - originally I tried 4" flex ducting, but it's too restrictive, it boosts these numbers by about 10c.  I've ordered a 6" flex duct to experiment with. but since it's bigger than the outlet I suspect it should only affect things based on how straight it's kept.   Either way, the two big takeaways for me were that the spacers are a no-brainer - in my case I cut them out of 5.6mm acrylic, and they typically dropped temps by 2c (I know this says 3c, but I tried it across a bunch of different fans and this is Kaze specific).  The second was a surprise for me, and that was how well Pull worked in conjunction with an enclosure - I'm probably going to be redoing most of my miners in this configuration.

Let me know if you have any other questions...

[amelen]

Great info, thanks! I wonder if pull would also improve results without am enclosure. Or in a partial enclosure scenario like a metal plate below the units.

[MarkAz]

My 3rd number there is just two fans in push/pull - the spacers are just small fan cutouts that make it so there is some sealed space between the fan and the case, for better airflow.  I tried the push/pull with the case, but the results were marginal so I didn't even bother recording it.

I also tried pull only without a case, but that under performed push only - both don't cool the opposite side nearly as well, but push was less bad.  

As far as partial enclosure, I first tried a sealing up the top and bottom of the heat sink - not quite as fancy as the 3D printed on that someone else did (which looks awesome).  The affect on temps was like 1c lower, whereas when you enclose the sides it makes a much larger difference.  The way I made my top, it rests on the top of the PCB's, so it effectively seals off the heat sink channel as well.

[amelen]

My 3rd number there is just two fans in push/pull - the spacers are just small fan cutouts that make it so there is some sealed space between the fan and the case, for better airflow.  I tried the push/pull with the case, but the results were marginal so I didn't even bother recording it.

I also tried pull only without a case, but that under performed push only - both don't cool the opposite side nearly as well, but push was less bad.  

As far as partial enclosure, I first tried a sealing up the top and bottom of the heat sink - not quite as fancy as the 3D printed on that someone else did (which looks awesome).  The affect on temps was like 1c lower, whereas when you enclose the sides it makes a much larger difference.  The way I made my top, it rests on the top of the PCB's, so it effectively seals off the heat sink channel as well.

Got it. I got a cheap fan for the pull, have a Delta for the push. Going to do further testing later in the week. I wish the Delta was a little quieter, but it is what it is. Another interesting note - the stock fan had a label stating it was rated to 2AMP. Most of the after market fans use a lot less power (other than the Delta's), so swapping fans might further increase the power efficiency of these units.

[Evan]

Yes, it could, I would recommend the most efficient "mod" is going to be sealing that case and forcing air.

And No Cardboard isn't going to catch on fire unless you've got spark or flame, Cardboard is an AMAZING insulator... if you google around you can find a Cardboard CPU case...

[chek2fire]

In my opinion the key to have a cooler s5 with other fans is how thick they are. If you have notice the stock fan is very thick. This has the result to pull out more hot air and very fast. I dont think anyone will see low temp with any other thin fan.

[Evan]

In my opinion the key to have a cooler s5 with other fans is how thick they are. If you have notice the stock fan is very thick. This has the result to pull out more hot air and very fast. I dont think anyone will see low temp with any other thin fan.

Did you read about a page back when we talking about Static Pressure?

[amelen]

New fans came - did various testing. In all scenario's, with a good push fan (160cfm min), adding a pull fan only lowers temps 1-2C.

[alh]

In case you guys aren't looking:

https://bitcointalk.org/index.php?topic=1086882.0

Lot's of good info on S5 temperatures, hot spots, and enclosing an S5.

[Evan]

New fans came - did various testing. In all scenario's, with a good push fan (160cfm min), adding a pull fan only lowers temps 1-2C.

Open case? closed case? Push/Push? Pull/Pull??? Push/Pull?