Seem to mostly depend on the PSU quality, on EVGA G2's i mostly get 393, one take 400 before a loss, for 1.29 and 1.32TH.

Np. If you get server PSU with adjustable volt, i suspect you might be able to get significant overclock upgrade, at efficiency loss.

A chain just popped after a soft reset at 28 ASIC and missing a bit of hashrate. Any idea of what to try and such? Beside the obvious replugging and power cycling?

I dont have any X, though. The o's are just missing. Its the two last oo's in the chain, if that matters.

I dont have any -. The o's are just missing. Its the same i get if i change voltage on a S4 without a power cycle.

I do have 2 failed board but i doubt i should do this kind of work. (Extracting and replacing the chip) i mean i've had trouble resoldering an uart cable.
And for 2 ASIC its not really worth shipping it out to say you or sidehack for repair. I'm mostly worried on if this can cause further deterioration to the board/controller/psu if i continue to run the board. What do you think?

I dont know what SC is, OC could be... overcurrent? Anyhow, it is hashing normally-ish (75gh/s loss), the red led seem dead. I never really paid attention before but right now its right blade's led does not seem to be on.

Whats the diagnostic, doctor?

Short Circuit, Open Circuit, referring to the chips being in a string from a power point of view. So if you have an OC chip none of them will work, if SC then the rest of the chain is powered at a slightly higher voltage so it might work, which yours is. The fact that the red led is not on means there is a fault in the IO (In Out  circuitry in one of the final pair of chips.

Bottom line is that probably there is a fault in one of the final pair of chips, but that it is not preventing the rest of the chain from working. May not be a SC core, could just be a fault in the IO signal side of the chip.

However although I have a fair understanding of the circuit treat the above with a pinch of salt as so far I have not been able to repair my board with the last 4 chips not working....

Rich

There is not much to go wrong on each of the stages. A big decoupling capacitor , the level shifting components between the stages, and the 1.8V Low Drop Out regulator. A few measurements are needed to confirm that chip failure is the most likely problem, then just the job of robbing a replacement chip from another board.

As I said I have not yet successfully repaired mine and the next job on the list is to make an ASIC test jig so that I can check the chip I remove before using it as a replacement.

If I ever get this done I will write it up along with a trouble shooting guide. However at this point I feel that my knowledge is just in theory rather than practice.

I have found removing and replacing the chips difficult with the IR rework station I have. Because of the heatsinking effect from the copper areas on the board they need a lot of heat, and I am worried I am damaging the chips. I suspect I need a more expensive device than the one I am using

Rich

I place them on a S1 cool rib. Ambiant temp is 23c


https://bt.irlbtc.com/view/1095605.msg11674556#msg11674556