Antminer S7 hash board versions

[jackacelives]

I picked up a couple S7's hoping to have a fun project to tinker with for winter. One of them works perfectly hashing at almost 5TH/s nice and cool in my cold garage. The other one will only hash when the middle hash board is unplugged. I have tried every combination of PSU, reset the antminer, updated the firmware, tried each board individually, tried running all combinations of 2 boards (two outer boards work perfect together, but either one with the middle board will struggle to hash @~10GH/s for maybe 1 min then crashes to 0 and sits). I have swapped I/O ribbon cables, I/O locations on the controller, still only happens to the middle board. The seller sent me screenshot of the miner just before he unplugged it to meet me and it was hashing @ 4.7-4.9TH/s with all boards showing o.

The middle board:
shows all O
shows 45 ASIC chips
red light on the board is on

If all 3 boards are plugged in, it will not hash, but after 10-15 minutes, the middle board will finally show all X, and several chips on the other boards will also start to show X. ONLY when the middle board is plugged in. I have ran the miner with the two outer boards plugged in for 2+ days nonstop hashing at ~3.1TH/s without a single hiccup.

Heres the only thing I can think of as to why its happening after opening the unit and looking at the boards yesterday:
left board - Version:has V1.31
middle board - Version:has V1.11
right board - Version:has V1.31
Controller board - Version:has V1.31

Is the reason it isnt working because it is a different version? if so, why was it working previously? what are the differences between different board versions other than the obvious 45 and 54(?) ASIC chip versions? Can I buy a replacement board that isn't a V1.31 and it work? I am having a very hard time finding replacement boards for the S7, and then having to narrow it down to a V1.31 is going to be worse.

Any help is greatly appreciated! I come from GPU mining on a home scale so I am an enthusiast and look forward to continuing to learn and contribute!

[lightfoot]

Funny you should mention this, I have a pair of boards with a 1.71 version (45 chips) that do not seem to play well with a 45 chip 1.31 version board. Weirdest part is the "V" numbers on the boards:

1.31--works ok in the box "V3"
1.81--does not work "V2"
1.81--does not work "V4"

All read 45/700, so they are 45 chip units. I wonder if they can't work together, however the odd part is that the controller firmware is the S7-F1 4.73 code.....

Interesting.

[jackacelives]

Indeed. It is very hard to find information on board versions and compatibility. I am concerned that sending the middle board to Colorado to the bitmain repair center will be useless as the board shows up fine, receives power, but seems to just not want to work with different version boards... If I get time today between husband duties, I will pull the rear cover on the full hashing S7 and look at its board versions for what they are as well. Maybe we can start to piece this together!

On a separate note, if you want to part with that V1.31 board, I actually bought the second machine for my father for christmas and want to give it to him in full functioning shape. I'll PM you

[lightfoot]

Well it's one of my control boards for testing, so I'll see if I can get another board working first as a stable control. Issue might be with firmware, as there are two unique firmware versions from Antland for the S7/700/45 boards. I'll try that this week in my spare time.

Meantime check the voltage from the inductor to ground on the PCIe plug for the errant boards. A working board will read about 10.5volts to ground from either pin on the inductor, if yours reads zero it's not a bad fix. If you pull the board out and bench test it without a controller it should read around 7 volts.

[jackacelives]

I measured voltage across the inductor hoping that was the issue (like here: https://bitcointalk.org/index.php?topic=1420909.0). However I believe 8V is ok when testing on a bench? I basically just plugged in the boards 3 PCIe 12v connections and measured, didn't power on the machine. Tested at the PCIe plug and across the inductor to ground (see below):
https://i.imgur.com/Mpa0L7n.jpg
https://i.imgur.com/kuRbjnO.jpg

Is this voltage OK in testing or does it need to be 9.3V like the potentiometer fix says the board needs to power on? Again, all boards show "o" on the status page but refuses to hash and brings the other boards down with it.

Lightfoot, I cannot PM more than twice a day apparently... hope you are looking here too  

[lightfoot]

Yep. On the bench that's about right; if you take a voltage reading with the boards in and plugged into the controller it should be about 10.5 as 8.mumble is the default on the power controller chip when it's at idle. So that's working.

I'm leaning towards the wrong firmware: You have the later firmware on the board which works with the same boards I have in my unit (1.31) with the slightly different earlier firmware. When I get a chance I'll reflash mine with the later firmware and see if it runs my later 1.81 boards, I'll guess you need the other firmware to run 1.11 boards and 1.31 boards work because I'm using it.

Do the logical dance....

CC

[jackacelives]

Is there an archive for previous firmware versions? Bitmain only shows the two from May of 2016 so I'm unsure of where to find legit versions... And the stigma against firmware updates to antminers in general!

Related, does the backup function in the settings also back up the firmware on the machine or just user settings? I did create a backup before I flashed firmware trying to get it running, wondering if that could work.