Also, after some testing here, it seems that these "mods" are somewhat bad. As far as I know, we get paid per share. I'm hashing away at ~530Kh/s (Running at 1213Mhz with 0 HW), but after 1 hour my machine that is hashing at 400Kh/s (900Mhz with 0 HW) is making just a few shares bellow the OC one (Like less than 5). I'm not even sure if I'm going to mod my other miners (I could be wrong). Anyone else who modded their units are running into this problem?
Thanks!
Yes. Seen that.
I only have 7 of these, but, just about every single quirk someone has reported in this thread I have encountered on one or more of these. These things are workhorses in some ways but can be really finicky.
At this point I am almost regretting removing VMOD2 (2 jumpers) on the 4 I had done it to to go to VMOD3 (replace r52 with higher resistance). Even with only 46k it seems the power usage is so much more for little benefit over VMOD2. The one I put 49k on only now gets to 1163.
I have definitely seen one of them seemingly hash for a long time and only report a few or no shares actually accepted. Lowering the speed helped. Also seen ones that seem to do better at lower speeds than higher, even though they never report a HW error. Of course you fall into the trap of variance when watching them over any short period, but some of these were 12 hour runs(still short) on the same pi, same pool and same cgminer instance.
I like to run per chip stats, it seems to usually be only 1-3 chips on each pod that throws HW errors, guess it depends on if the other chips produce more valid work or not by going higher whether it is worth it.