Cheap and simple repair of S7 hash board

[RadekG]

How much resistance is used for the adjustment range, I have 10k multiturn trimpots from S1 adjustment?

I am not sure if you will be able to reach correct voltage, but you can try and let us know.

[RadekG]

Very nice simple modification. The 135 chip Hash board has just been waiting for something like this. 

Strange that the PIC should fail, did you spend any time looking into what has failed in the original circuit?

I assume the 50K pot just replaces the original MCP4017 Digital Pot? Did you remove the connection to the Digital Pot?


Rich

I was wondering about this mod since our discussion in another thread. I had 4 dead boards and talking to our stupid UPS agent was so annoying that I missed RMA window, so I did this mod without risk.

Since 3 of my boards are not working at all (voltage about 8.7V) I did not remove MPC digi pot, so voltage can't go as low as original MPC pot, but boards start-up at 9.5V, so it is not necessary. The picture is taken from resurrected board which is hashing at about 0.22W/GH DC.

I have fourth board which has 9.5V before mod. It starts-up with modded voltage, but after cgminer starts it tries to push voltage back to 9.5V level. I guess that PIC listens when cgminer starts and after start-up period it can decrease voltage to make S7 more efficient. I don't think this FW function is imlplemented in any public batch.

[yslyung]

what are the indication of the not working or failed boards ?

-normally the board will show lots of errors
-shows 30 chips instead of 45 chips
-dashes on the chips
-low temp compared to working boards
-it don't work at all

but sometimes if lucky it "may" work by

-reducing the freq to about 500 then let it run for a while then the chips would come back but with slightly higher error than normal
-save & apply couple of times & wait, just let it run

appreciate if you or anyone could share more info or step by step guide with more pics.

thx for helping the community

[RadekG]

Hey, I checked that, and it has 10.4V there. But, D24 is burned and in short circuit. Does anyone know what part could I use to replace D24?. I think it's a schottky diode, but I don't have the exact specs.

Another thing: Is it worth the hassle, dusting the boards?.

Use Google. "SL diode smd". It is 1N5819HW, but you can use any 40V+ and 1A+ schottky diode. Dust on PCB is not dangerous, but it is flammable, so in case of fire it really doesn't help.

[RadekG]

what are the indication of the not working or failed boards ?

-normally the board will show lots of errors
-shows 30 chips instead of 45 chips
-dashes on the chips
-low temp compared to working boards
-it don't work at all

but sometimes if lucky it "may" work by

-reducing the freq to about 500 then let it run for a while then the chips would come back but with slightly higher error than normal
-save & apply couple of times & wait, just let it run

appreciate if you or anyone could share more info or step by step guide with more pics.

thx for helping the community

30 chips per chain is typical indicator of low voltage for chain - dead or malfunctioning PIC controller, exactly this thread. To confirm, you need voltmeter and check voltage at big inductor. Voltage should be about 10.5V. If less, you need RMA or this mod.

[yslyung]

30 chips per chain is typical indicator of low voltage for chain - dead or malfunctioning PIC controller, exactly this thread. To confirm, you need voltmeter and check voltage at big inductor. Voltage should be about 10.5V. If less, you need RMA or this mod.

i have 1 board that have this problem & it is a RMA'ed board !!! gessshhhhh....

some pictorial will be helpful. ordering a pot on monday.

[RichBC]

I hope it will help you. My opinion is that PIC malfunction is intentional from Bitmain to lower diff after RMA period.

Interesting thought on this being intentional? It would be good to see if they have code protected the PIC or if it can be read? Makes me want an S7.....

Rich

[RadekG]

Interesting thought on this being intentional? It would be good to see if they have code protected the PIC or if it can be read? Makes me want an S7.....

Rich

There is no reason for so high amount of not working PICs. I can imagine some high power device can fail because bad design, but PIC controller is not exposed to dangerous conditions.

1) PIC is allowed to work under higher temperatures than S7 working temperature is, so there is no reason for PIC to fail.

2) If the problem is firmware bug, after SO MANY repairs of one chip Bitmain should know where the bug is. It is not first or second batch.

3) If Bitmain knows, they should send repaired boards from RMA with updated FW.

I have no other explanation than it is their intention.

[cdjbolton]

Just wondering if either of these 2 would work, I assume you start at the lowest setting and work up?

As I think a new order S7 has one blade effected.

http://www.amazon.co.uk/TRIMMER-TURN-50K-3006P-1-503LF-BOURNS/dp/B01186ESAM/ref=sr_1_1?s=industrial&ie=UTF8&qid=1460404535&sr=8-1&keywords=Bourns+3006P-1-503LF+50k

http://www.rapidonline.com/electronic-components/bourns-3006p-1-503lf-50k-3006p-10-0-75-multiturn-cermet-pot-68-0140

[E]

No, I reached about 0.22 at the wall. Still much better than original 0.3+

That's pretty fantastic, especially since that's at a startup-friendly voltage.

Reading this inspired me to finally stick a logic analyzer on the MCP4017's I2C interface. I've only tried one hashboard so far, and the only traffic I saw was immediately after startup:

(About 0.5 seconds after power is applied)
Setup Write to 0101111 (This is the MCP4017's address)
ACK
Write 010h
ACK

SCL is brought out the header labled P1 (pin 5 if the square pad is pin 1), but I don't see SDA there.

[E]

BTW the title of this thread seriously undersells it -- make it "Cheap and simple undervolting of S7 hash board" and more people will jump in

[RadekG]

Just wondering if either of these 2 would work, I assume you start at the lowest setting and work up?

As I think a new order S7 has one blade effected.

http://www.amazon.co.uk/TRIMMER-TURN-50K-3006P-1-503LF-BOURNS/dp/B01186ESAM/ref=sr_1_1?s=industrial&ie=UTF8&qid=1460404535&sr=8-1&keywords=Bourns+3006P-1-503LF+50k

http://www.rapidonline.com/electronic-components/bourns-3006p-1-503lf-50k-3006p-10-0-75-multiturn-cermet-pot-68-0140

You can use any multiturn 50k pot. You can use also Suntan brand.

[RichBC]

That's pretty fantastic, especially since that's at a startup-friendly voltage.

Reading this inspired me to finally stick a logic analyzer on the MCP4017's I2C interface. I've only tried one hashboard so far, and the only traffic I saw was immediately after startup:

(About 0.5 seconds after power is applied)
Setup Write to 0101111 (This is the MCP4017's address)
ACK
Write 010h
ACK

SCL is brought out the header labled P1 (pin 5 if the square pad is pin 1), but I don't see SDA there.

So looks like it is as simple as power up and write a value to the Pot?

Could be 50K /127 * 16 = 6.3K

Have not checked if that value makes sense?

If that is all it does it makes you wonder why they did not just use a Non Volatile pot? Which makes me think that perhaps under some other circumstances something different is written?

Rich

[jmumich]

So looks like it is as simple as power up and write a value to the Pot?

Could be 50K /127 * 16 = 6.3K

Have not checked if that value makes sense?

If that is all it does it makes you wonder why they did not just use a Non Volatile pot? Which makes me think that perhaps under some other circumstances something different is written?

Rich

Is it possible that Bitmain is able to mine closer to the chip's spec using these boards without any physical modification?

[RichBC]

SCL is brought out the header labled P1 (pin 5 if the square pad is pin 1), but I don't see SDA there.

I suspect that the main function of P1 is to provide the ICSP programming connections for the PIC?

Rich

[RadekG]

I suspect that the main function of P1 is to provide the ICSP programming connections for the PIC?

Rich

I think Bitmain has 19" rackmount enclosures for S5+/S7 boards fitting 9 boards. We can see unpopulated power connectors using screws. We can also suspect other unpopulated connectors may be used for industry mining.

Still unclear, WHY it fails the same way and why Bitmain did not solved it yet. Just repaired another two boards with failed PIC.

[RichBC]

I think Bitmain has 19" rackmount enclosures for S5+/S7 boards fitting 9 boards. We can see unpopulated power connectors using screws. We can also suspect other unpopulated connectors may be used for industry mining.

Still unclear, WHY it fails the same way and why Bitmain did not solved it yet. Just repaired another two boards with failed PIC.

Yes the failure rate of a relatively simple piece of unstressed circuitry is very odd indeed. Do we know if it is the PIC or the Digi Pot that is failing?

Has anyone tried to read the contents of the PIC?

Rich

[yslyung]

More pics pls

[alh]

This PIC and it's actions make you wonder if perhaps a different firmware in the PIC couldn't improve the S7 efficiency. Consider something like:

1) At power up for the PIC, send out the required value to safely and reliably start the chain (e.g. 10h)

2) Delay a minute (whatever it takes to get board warmed up and settled in)

3) Send out a different value that reduces the voltage some to up the efficiency.

It wouldn't really matter how long the board ran at the increased "get it started" voltage, since what really matters is the runtime voltage for the numerous hours after startup that matter.

just a crazy thought.

[sidehack]

Not ...that crazy.