Quote from: lightfoot on October 03, 2014, 04:37:56 PM

Quote from: wrb2010 on October 03, 2014, 03:43:27 PM

I have at least two that are leaking. The biggest problem is now they are shutting off the PSU once connected. Any ideas.

Hm. Let's see if I can help. Questions:

Do they mine? Are the copper thingies on them getting hot to the touch? How about the back of the chips, there are two cut-outs in that bottom heat plate. Is it hot to your (DRY) finger's touch? Do you have them in that metal harness thing? What kind of power supply, and are the pumps plugged into the board (there are two, one on each block)

I'm wondering if when they leak they are creating cavities in the water flow, which would cause air pockets and bubbles to form. When you're cooling with water, an air bubble will *not* conduct heat. Heat goes up, resistance goes down, power supply goes crowbar.

What's the temperature in the room they are in?

My little Ebay Monarch here is mining at 700gh, but when I power it off then on it runs at 500 or so gh. Letting it sit for an hour to cool completely brings it back to 690-699 or so, which is odd but nice. I also have pretty good hearing, and when it is going to hash slower I can hear a high pitched "whine" from the FETs or the inductors. Just had to power it off this morning because it crashed,literally fell off the power supply and window sill. Note: Slight vibrations can do that. Oops. :-)

I'm running it on a good quality Corsair 500. The corsair is warm, and more importantly the two plugs going into the Monarch are warm. Given that the Corsair uses very good quality cables, and I have had crap-ola power supplies with much thinner cables, I would say that a less-than-good power supply could fail despite it's rating. Thin wires would heat up, drop voltage, increase current till the supply goes boom (did this with a 500 watt cheap-ass supply and a 32gh jally. Oops.)

However it does dump a fair amount of heat out of the radiator, way more than my 30gh turbo jallies ever did so the water cooling is more important.

If it's anything like the 65nm gear, the cores could be failing selftest when it's started hot and get disabled. Try running the cooling separately and turning off the card, let the cooling run a minute, and then turn the card back on to see if it pops right back up to 700GH/s.