also, currently trying out 6206 (215.625MHz) on a U1, and expect to implement it on my S2 after a few more minutes of confirmation. 212.5MHz gave me 0.75% hw errors while 218.75MHz was about 2.02%.
hopefully 215.625MHz will be around 1-1.2% error rate and thus about the highest stable speed for an antminer S2 (batch 2, external DPS-2000 PSU)
Have you set the U1 to the same voltage as on the S2s?
no i haven't - my main use of the U1 is to make sure the code isnt wrong and causing too high, low, or impossible a frequency and thus hardware failure.
6206 worked fine on the U1 (1.730GH average) and is now implemented on the S2 problem free (1098GH, 1.136% hw errors) after 1.5 hours of operation. Presumably there's a ~1180W power draw to go with it, but my 208V PDU is not metered. Temps range from 48-55 degrees
still waiting on someone to identify which resistor on the S2 would allow pencil-modding to raise the voltage. The chips and arrangements are quite similar to the S1, but not identical enough that i can firmly state 'it is resister Rxx'. I would love to be able to slightly bump up the voltage on some of the hashing boards to see the effects on hashrate, but I assume if temps get much higher than 55C (board 1) the error rate might really start to jump up.
I still think that these units could handle 1280GH/1500W if the resistors were just set just a bit differently for a few extra millivolts to reduce the amperage and allow a little further tuning into the 225-250MHz range (frequency 0981 would give 250MHz, or about 1280GH)