S5 board gone bad, options

[soy]

Ah, great.  UPS shows a delivery scheduled for tomorrow, Monday.  I was wondering what would happen if the hashing board were mailed without tracking to my home.  Having a fence, UPS sometimes doesn't deliver, returning the package to the office instead of just leaving it inside the gate.  Without a tracking number I couldn't request email updates.  But, having my name and email address on file, UPS looked it up and sent me a notification!

[talks_cheep]

I have an S5 with bad boards, how do I get in touch with Sherri?

[BTC_ISTANBUL]

I do not think that the S5 replacement and repair is still available.By the way her name is Liu Sherry.

[soy]

Board arrived today.

I know.  The name order confuses me too.  Had a friend, chess player, Hshaun Kuo Yu, called him Kuo.

Sherry Liu <xiaorui.liu@bitmaintech.com>

Busy, and it was warm today so the board is still in its shipping box.  Who knows, perhaps when a farm S5 goes half bad they retire it then send out the good side as a replacement for a fee (only when a board is available) but officially not still repairing the boards.

[soy]

Opened the package and it's the same board I had sent for repair.  Will test it later.

[soy]

Found time today.  Repaired board in, fired up, it runs fine but my first good board is coming up short, only 18 o's showing on that board no x's, suspect a power supply issue.

[soy]

Putting the repaired board on the side with the 4 wire ribbon cable and no power pcie connectors to the older board I get chain 1 showing 32 o's and am getting 582.48 GH/s.  Shutting it down, adding power to the long time good board, I fire it up after 2 minutes cold and the long time good board is showing as chain 2 with 20 o's, no x's.  After 6 minutes the 20 o's of the older dependable board has changed to all x's which agrees with an earlier test where it was the only board, showing 20 o's no x's but wouldn't show any hashing.  So, my old board has gone bad.  When I had it apart I brushed it off and hit it with compressed air.  Will have to give it a good cleaning and see if it improves.  But the board I had sent for repair, clearly marked with a sticker, is working well.

[RichBC]

Putting the repaired board on the side with the 4 wire ribbon cable and no power pcie connectors to the older board I get chain 1 showing 32 o's and am getting 582.48 GH/s.  Shutting it down, adding power to the long time good board, I fire it up after 2 minutes cold and the long time good board is showing as chain 2 with 20 o's, no x's.  After 6 minutes the 20 o's of the older dependable board has changed to all x's which agrees with an earlier test where it was the only board, showing 20 o's no x's but wouldn't show any hashing.  So, my old board has gone bad.  When I had it apart I brushed it off and hit it with compressed air.  Will have to give it a good cleaning and see if it improves.  But the board I had sent for repair, clearly marked with a sticker, is working well.

Not sure from the above but have you tried the "Long time good board" on it's own, Chain 1 side with the 4 pin power connector?

Rich

[soy]

So a repaired board good and a good board gone bad, no gain on the play and a cost of ~$75.

[soy]

Not sure from the above but have you tried the "Long time good board" on it's own, Chain 1 side with the 4 pin power connector?

Rich

Yes

[soy]

Interesting.  Over time the 20 x's increased to 26 x's.  Anyone seen anything like this before?
---------------
Shutting down, swapping ribbon cables, waited, started it up, 19 x's.
---------------
Shut it down, removed first hashing board, cleaned thermal paste and reapplied as well as a quick clean of the topside with brake fluid.  Back in as secondary, now showing 22 of 30 ASICs but after a minute they're all x's.  So, no gain.
---------------
Just to be sure I wasn't mistaken, I shut it down, put the fan on the other side, ripped off the sticker, flipped the controller board to the other side, and only connected the old board that had been working yesterday with nothing to repaired board.  It's showing a chain of 16 ASICs, all o's but it's only been a little over 3 minutes since firing it up so I expect to see x's soon as there's no hashing going on.  5 minutes, 10 getworks, zero hashing.  Will ssh in and /etc/init.d/cgminer.sh restart....resstart shows 'Have bitmain-asic 7'.   Now it's 17 ASICs on the GUI, all o's, but no hashing.

[soy]

So, postage and repair costs, with the other hashing board going bad, has eliminated any savings I would have made this winter using electric (miners) instead of propane to heat my home.

[soy]

Took the old board out, inspected, D102 appears damaged, got knocked somewhere.  Will address tomorrow earlier in the day.

[soy]

Couldn't just watch TV and wait until tomorrow.  Removed the broken diode and replaced it with a small signal diode, 1N4148 probably, unmarked.  Fired it up just now and have both chains, all 60 ASICs, but chain 1 is running at 49° and chain 2 is running at 61°.  That chain 2 temp seems hot.  I have a second fan on the back driven at 12v.  Okay, just shut it down as the Dell server supply has a voltage too low and am going to other supplies.
-----------
Okay back up and temps are 51/57 but will see in a while.

[soy]

At 32 minutes, 1,129GH/S(avg), HW 5, 52°/58°, fan 3960.  I think it's fine.
94 minutes, 1,145GH/s(avg), HW 5, 51°/58°, 3960.

[RichBC]

Took the old board out, inspected, D102 appears damaged, got knocked somewhere.  Will address tomorrow earlier in the day.

Interesting on D102. It is part of the level shifter that couples the reset signal between the chips. If faulty then chips after 14 would not be receiving the reset signal.


Rich