No, I'm thinking that I'll absolutely need to lower the wattage/GH/s as the summer heats up and these get more inefficient.  I've seen the thread where resistors are soldered in parallel within the voltage regulator but not prepared to do that.

Firmware version Thu Aug 7 Kernel 3.10.12

I was hoping to move up to a voltage option along with frequency.

I have my miners on inexpensive, 4 shelf, wheeled carts from Walmart's.  Supplies up top and miner below.  S3's could slide in but C1's I had to cut out a bar to increase access.    Shelves are too close to remove a cover in place.  For the S3's the mobility is good as I can just spin the cart if I reverse a window fan but of course the C1's don't have much latitude with the radiators mounted in another window.

May I ask which firmware you're using?  I tried one and got a failure to load the GUI situation forcing me back.  Is this usual or something to do with my setup?

Can't get the cover off without breaking down its installation.  There's only a subnet accessible in the back room.  Resetting puts it back on the .1.0 network.  I can leave it in place with circ water cooling and reset while hooked to a router not connected to the internet.  Big discussion in S5 forum about how loss of internet gets overheating.  At least the C1's cooling continues after internet break.  But it's hashing perfectly and I have the GUI.

I'd like to know if this failure is a built-in safeguard of some kind.  Putting a short script in /sbin/, which is on the path unlike /usr/sbin/ then saving with copy -rf /etc/shadow to /config/ then rebooting and losing access via ssh.  Tried root/mypassword, root/admin, admin/admin and none get me in.  If there were a cron entry like the S3's I might try putting in passwd root and running it once but no cron on C1's.  ..and now that I think of it - it wouldn't work as it comes back with the enter new password request

Well lost password access via ssh on 2 C1's just now.  I knew that when new to sign in via ssh one would need to sign in as root and use admin as the password.  So, when my regular password wouldn't work I tried that but that's no go either.

What happened was I put a script in /sbin called cg.sh to restart /etc/init.d/cgminer.sh  but of course it would be lost with the next boot so I ran the 'cp -rf /etc/shadow /config/' then on the next boot I tried to login to run the script but can't get in.

I have tried changing the password in the GUI.  I can still get into the GUI no problem just can't ssh in.

Perplexing.  The JB Weld quite cured I put the after market pump back in the loop miner loop after running an hour on a closed loop without a leak.  Once in the miners' loop I got a leak at the out port.  Figuring it might be the hose I replaced the hose at that port. It leaks again.  I suspect this new leak at the seam between the ports or one side of a port tube.  I noticed the halves weren't mated perfectly and lightly filed down the ridge along the port tube.  After removing the pump from the circuit, with a short tube on one port and my thumb on another I submerged it in a sink and applied air pressure expecting to see bubbles - no bubbles.  Perhaps the out port needs the downward pressure of a tube filled with water to leak.  ...Yep, applying downward pressure on the out port while submerged and applying air pressure shows a leak along its top interface with the box.

The stock pump seems to have thicker walls than the after market Syscooling pump, the one that broke.

After watching immersion cooling in Novec I know I'd love to do the same except it's so expensive.  Naturally my thoughts go to immersion in R134a but at room temperature there'd be 89.9psi so something like 1" Lexan without thru holes would be needed with a steel top for thru connections of hoses and conductors and who can afford the 1" Lexan while considering ROI.  Could you imagine subjecting a thin-walled after market Syscooling pump to hundreds of psi?  Or a steel tank with LED lighting and a videocam.

It seems clear Bitmain lost confidence in Syscooling.  The fault as I see it may be entirely in Syscooling's bonding process although the wall thickness is likely an element.  Removing the tank leaving only a small section to break off would mean only that small section was fully bonded to the bottom.

Pump repair was a failure.  Will try to remove the tank and re-glue.

Put the second radiator after the stock stock and before the pump feeding C1#1.  C1#1 is now the cooler of the two C1's.  Previously C2 was cooler, cooled with the condenser.

Took the pump, gripped the tank snuggly, pulled, the tank came free of the bottom.

-------------------------------------------------------------------------------

I thought the top simply separated from the bottom but it turns out it wasn't quite that.  The edge of the clear plastic seemed jagged and in a back corner a considerable amount of the clear plastic remained glued to the black bottom, broken off about an eighth inch high.  Perhaps the top wasn't glued but welded to the bottom with a chemical that liquified the edge for a bit and allowed them to bond.  I went with JB Weld instead of 5 minute epoxy, putting a bead along both surfaces and mating.  Tomorrow I'll run a tube from the in port to the out port and run to check for leaks.  Leaks near the electronics are disconcerting.

My S5 order shows paid, unshipped, valid, order date 3/18 but then again so does my second C1 order which is happily hashing away in my backroom.

My stock pump is okay but I chose to position it between the radiator cool port and the miner reasoning lower temperature means less wear.  

The second pump, different model manufactured by Syscooling, only developed the leak when I tapped on it with a screwdriver to bring bubbles to the surface.  After laying down a line of glue along the back seam and letting it dry some hours, it wasn't completely dry except on its surface.  I pressed on the dried surface and saw a couple of dots of wet glue pass to the inside where the leaks had developed.

My experience today and generally since I added the second C1, is that good coolant circulation and cooling make a difference.  The C1 #2 getting cooled by a car AC condenser with 5 fans is much better behaved than the 1st C1 with a stock radiator and pump.  My second pump should be usable by tomorrow and will go back in the circuit - would have been sooner if I could find my Devcon 5 minute epoxy.  The radiator from China I bought on Ebay arrived in Atlanta earlier this evening.  If it gets to my local post office by tomorrow, great.  I'll add it in series with the stock radiator on C1 #1 which should already be able to tolerate the 243.75M/0675 again once the second pump is put back.

So perhaps your system could improve with better circulation or by using distilled water instead of coolant.  Coolants, like auto antifreeze, only move half the heat that pure water can.

To get more efficiency a week or so ago I knocked the C1's down to 243.75M/0675

No, no, I did recall, but I was wrong, didn't jot the numbers down before changing, but thought the original settings on the C1's had been 250M/0750 but I was mistaken.  Ran at that for a little while then searched and found 250M/075 to be the default on these C1's.

As I said, to get more efficiency a week or so ago I knocked the C1's down to 243.75M/0675 so the two would be drawing about 9.5amps each on a 20 amp circuit tho they had been running 10a each for a while and the breaker wasn't warm to the touch.  Hot summer days I needed to leave the breaker box door open so a living room breaker didn't trip with the AC on - later added two 15 amp lines and that's no longer an issue.  So one can easily tell if a breaker is prone to trip as it will be warm.

For some reason, perhaps because of the humidity, after switching back to a single pump driving water thru two C1's, the original C1 was coming up x's.  I jumped to the conclusion it was due to having added a slightly restrictive PVC barb but taking that out didn't correct the x's.  The x's were corrected when that miner was put back to its original settings of 250M/0725.  Now the two miners draw 10a and 9.5a, breaker cool.

When my S5 arrives, a circuit I had put in for my living room treadmill will be disconnected and a new temporary 15 amp line will be run to the back room.

The second C1 is still running at 243.75/0675 and gets 986.66GH/s(avg).
 

Spoke too soon.  Back up, and x's are appearing in the one C1 again.    Wonder if it's related to the f=243.75M/0675V I'm running for efficiency.
--------------------

Okay, up 55s at normal power and frequency and no x's............1m34secs and still okay...............2min11sec 1,019 GH/s(avg) and no x's........3min15sec and all okay

Odd.  It's happier at full power.

And running with a pump/fans power supply divorced from the miners, it's easy to shut down the miners power and forget to shut down the pump which leads to a bit of a mess once a hose is separated.

I should note that it's raining here and as many have observed high humidity doesn't afford as much cooling.

Okay, I originally ran the two C1's from the one stock pump no problem.  Then I added the second pump.  Ran great until I tapped the new pump to get an air bubble to the top and started a leak in the seam of its tank. 

Now back to running the two C1's from the one pump but am getting x's from the original C1.  The difference?  I had to add a piece of tubing when I added the second pump.   The metal barbs have a wide port and don't restrict flow.  This time I spliced the inflow and outflow of the pump taken out with a straight thru PVC with barbs.  Problem is the PVC straight through with barbs has a restrictive opening about equal to what one would find on the air hose disconnects that I previously avoided.

So, just restarted and after 5 minutes got an x again and more will appear.

The second C1 which sees water cooled in the automobile AC condenser has always gotten cooler water than the first C1 which gets its water from the stock radiator.  Even with the slower water flow due to the restrictive PVC, C2 shows no x's.  Up 7 minutes and six x's on the first C1.

I'll now shut down and take the PVC out putting the hose to the condenser instead of the PVC.  If after I purge air again the original C1 runs with no x's then those x's were due entirely to the more narrow port of the PVC barb.  PVC barbs by design are thicker walled than the metal barbs.

The poorly working S3 I got 'new' from Pines Computer in Florida via Amazon does that.  The hashrate falls but a hot reboot doesn't help as much as running /etc/init.d/cgminer stop && sleep 2 && /etc/init.d/cgminer start but sometimes I do a shutdown and restart and x's or -'s will show - then I shut down for a full minute or more then restart and restarts all good.

I'll venture a guess...not all the air was flushed; it seemed full; remaining air expanded as it heated; coolant came out the port (if you had left it open to vent); coolant started boiling in the heatsink.

One of the advantages of running the pump(s) from a divorced 12v supply is that one can more easily run the pumps with the miner(s) shut down.  Running the pumps with the miners shut down is great for filling the system (e.g. suction and return into a gallon of distilled water) or just for purging.

!?! Just tapped this pump tank to bring a bubble to the top and the glue between the tank and bottom started leaking.  So, the two C1's down while I remove the pump and glue.  Meanwhile the C1's will go back up with 1 pump.