ANOTHER FIRE HAZARD WARNING:
One of my 16-card October kits that came calibrated at 0.865 V started to melt/burn at the PSU. It is a Seasonic Platinum 860W connected to the M-board using the standard 2 PCI-e power connector setup. It had been running fine for over a month at 620W and then suddenly I started smelling smoke coming from part where the modular connectors plug into the PSU.
Fortunately, it happened while I was in the room so I managed to shut it down before anything serious happened. 2 PCI-e modular ports on the PSU were melted and no longer usable. However, I managed to repair the rig using the other unused PSU ports. I now have it running with 2 x PCI-e cables AND a jerry rigged CPU 12V cable going to the ring terminals.
I also brought the voltage for all the cards down to 0.83 V. The loss in GH/s was minimal (the setup runs at 540 GH/s now) but I sleep much better.
Note: This is not the first time I've had a high-end modular PSU burn out at the modular connectors.
Well, c'mon guys, what do you expect. These systems were not engineered properly with that much power draw in mind. 620 watts is far too much to draw from just two 6pin pcie connectors for extended period of time. It will run fine for a short while as I've experienced when I melted my PSU's module plugs too. I now have the rig tuned to 500 GH and a power draw of 475watts. I'm not going any further than that as I don't feel like melting every damn PSU I have. This is with a V2 m-board which does NOT have the rig connectors so I can't use that to spread the load a bit.
If you can, measure amperage on each of the PCIE 6pin cables and see how much is going and make sure you do not exceed the rated maximum for the wire gauge your psu uses.
I've said this before, but I will say it again; I am shocked that they did not engineer the m-board with 3 or 4 separate 6pin PCIE connectors.
