THANK YOU!!!!!!       

2x45 for 0.17

So I've done a ton of experimenting and investigating on the S5 noise problem and found something very interesting.

The fan on the front of the S5 is NOT a 38mm fan. It's a 38mm form factor with more 25mm blades and a smaller static set of vanes on the inside to direct airflow. That means it has extra noise compared to an equivalent 38mm fan and does not generate the higher static air pressure that a 38mm fan would - presumably it was chosen for higher CFM compared to the 38mm version they used to include on the S3. So I ripped the fan off the back of the S3 and stuck it on the front of the S5. Noise levels are now significantly lower and the cooling is only marginally less effective.

None of these seemed to make any significant difference. The loudness and temperature seems almost entirely dependent on the one fan on the front.

.0118 btc  per day so the firsts ones came about 1 week ago  7 x .0118 = around .0826  x 277 = 22 usd not counting power.

I used 1th hash so go to 1.3th which is the max it can do 1.3 x 22 = 28.60  - the power which at best would be 8 bucks = 20 profit. looks like you are correct

Personally, I think it's a firmware issue - maybe if Bitmain would confirm......

This is different to the S3 fan, which is a generic "12025" with a 38mm deep rotor. The S4 and S5 fan utilises a 25mm deep rotor coupled with what I call a prespin. This type of design is typically required to prevent backflow at very high static pressures + high CFM combinations. Weird things would happen to the airflow patterns if you didn't use one.

tldr, its not a 'cheaper' or 'noisier' design, its a requirement to reach that performance threshold.

This is the first I heard of this and I have a batch 3 order so shipping any time now.  Is everyone experiencing this?  Does restarting cgminer fix it temporarily?  I have a batch 2 S3 that needs a ton of babysitting also because the hashrate drifts lower and lower the longer it runs, so what I finally did was create a cron job to restart cgminer.  I originally did a cron job to reboot the thing but one time it rebooted and got all hosed up and didn't mine at all for hours so I just restart cgminer now. 

Temperatures are key to getting these guys running stable. I had issues with xx's appearing around 64C.

Since keeping them cool and <~60C, they've been solid, overclocked to 375, for a few days now.

---
Had to reboot this one when I moved it to a circuit to share with an S3. Was running solid 3 days prior.



---
This one has been running for 4 days now solid.

From my experience so far with the S5, the dreaded X's don't always show up - the hash rate just halves due to one card "freezing" so to speak, so even mdudes monitor can't detect/reboot it. The only way of knowing is to constantly check the pool hash rate, which is a pain in the face