S17 Pro Back from Repairs or Disrepair.

[NoKeyboard-BTW]

Well got the S17 Pro back and was able to put eyes on the kernel myself.  It went in for repairs limping and came out even worse, so I was hoping there might be a magic (sorcery will do too) recommendation to save this on my own with tools and hardware online. Boards are non-existent so any help is appreciated.  Board 3 is off for now because it arrived with a loose heat sink. Board 2 is hashing well. Board 1 is the current issue.

1st Diagnosis - Tested all 3 hashboards, All ASICs report correctly with all temp sensors failing.

2nd Diagnosis and Final Repair. (Below)

Hashboard 1 (Chain 0)-
Temp sensors 2 & 4 replaced.
Reflowed Chips
Boost circuit failed board is unrepairable.

Kernel Log:

Code:

[2021/02/21 03:37:27] INFO: Power ON
[2021/02/21 03:37:29] INFO: Starting FPGA queue
[2021/02/21 03:37:29] INFO: Initializing hash boards
[2021/02/21 03:37:29] INFO: chain[1] - Initializing
[2021/02/21 03:37:29] INFO: chain[0] - Initializing
[2021/02/21 03:37:33] WARN: chain[0] - Failed to reset pic (attempt = 1), resp: 0x01 0x01
[2021/02/21 03:37:35] WARN: chain[0] - Failed to reset pic (attempt = 2), resp: 0x9a 0x9a
[2021/02/21 03:37:37] WARN: chain[0] - Failed to reset pic (attempt = 3), resp: 0x9a 0x9a
[2021/02/21 03:37:38] WARN: chain[0] - Failed to start pic app (attempt = 1), resp: 0x01 0x01
[2021/02/21 03:37:39] WARN: chain[0] - Failed to start pic app (attempt = 2), resp: 0x01 0x01
[2021/02/21 03:37:39] WARN: chain[0] - Failed to start pic app (attempt = 3), resp: 0x01 0x01
[2021/02/21 03:37:41] WARN: chain[0] - Failed to reset pic (attempt = 1), resp: 0x01 0x01
[2021/02/21 03:37:42] WARN: chain[0] - Failed to reset pic (attempt = 2), resp: 0x01 0x01
[2021/02/21 03:37:44] INFO: chain[1] - 48 chips detected
[2021/02/21 03:37:44] WARN: chain[0] - Failed to reset pic (attempt = 3), resp: 0x01 0x01
[2021/02/21 03:37:44] WARN: chain[0] - Failed to start pic app (attempt = 1), resp: 0x01 0x01
[2021/02/21 03:37:45] WARN: chain[0] - Failed to start pic app (attempt = 2), resp: 0x01 0x01
[2021/02/21 03:37:45] WARN: chain[0] - Failed to start pic app (attempt = 3), resp: 0x01 0x01
[2021/02/21 03:37:47] WARN: chain[0] - Failed to reset pic (attempt = 1), resp: 0x01 0x01
[2021/02/21 03:37:49] WARN: chain[0] - Failed to reset pic (attempt = 2), resp: 0x01 0x01
[2021/02/21 03:37:50] WARN: chain[0] - Failed to reset pic (attempt = 3), resp: 0x01 0x01
[2021/02/21 03:37:51] WARN: chain[0] - Failed to start pic app (attempt = 1), resp: 0x01 0x01
[2021/02/21 03:37:51] WARN: chain[0] - Failed to start pic app (attempt = 2), resp: 0x01 0x01
[2021/02/21 03:37:52] WARN: chain[0] - Failed to start pic app (attempt = 3), resp: 0x01 0x01
[2021/02/21 03:37:52] ERROR: driver-btm-chain.c:465 chain[0] - Failed to init pic controller
[2021/02/21 03:37:52] INFO: chain[0] - Shutting down the chain
[2021/02/21 03:37:52] ERROR: driver-btm-base.c:356 chain[0] - Initialization failed
[2021/02/21 03:37:54] INFO: Start-up temperature is 28 C (min -15 C)
[2021/02/21 03:37:54] INFO: Switching to manual fan control (100 %)
[2021/02/21 03:37:54] INFO: Changing voltage from 21000 to 17400 mV gradually
[2021/02/21 03:38:34] INFO: Raising freq from 50 to 500 Mhz gradually
[2021/02/21 03:38:45] INFO: Switching to automatic fan control (75 C)
[2021/02/21 03:38:45] INFO: Start mining!

Thanks for taking a look!

NKBTW

[wndsnb]

The error messages from chain 0 are from not getting any response from the pic microprocessor on the hashboard. This is a pretty basic function, the board only needs power supplied from the control board and a functional pic to get a response. Without the pic, the control board can't turn on the power to the ASIC chips, so there is no chance of getting anything working. Could be that the pic got fried and needs to be replaced.

[BitMaxz]

This sometimes happens if the control board and hashboard have a bad connection.

Can you try to replace the cable or use the cable from hashboard 3 then test it again.

Source: I got from here https://bitcointalk.org/index.php?topic=5217949.msg55698844#msg55698844

Just read the 3rd response from him which is solved by replacing the cable.

[NoKeyboard-BTW]

BitMaxz, definitely the same errors. So I detached the board 1 cable from the hashboard and plugged the cable that usually runs hashboard 3 into hasboard 1. I didn't switch controller board cable positions from 3 to 1, just plugged 3 into hashboard 1. Same result.

Code:

[2021/02/24 20:10:01] INFO: Power ON
[2021/02/24 20:10:03] INFO: Starting FPGA queue
[2021/02/24 20:10:03] INFO: Initializing hash boards
[2021/02/24 20:10:03] INFO: chain[2] - Initializing
[2021/02/24 20:10:03] INFO: chain[1] - Initializing
[2021/02/24 20:10:06] WARN: chain[2] - Failed to reset pic (attempt = 1), resp: 0x00 0x00
[2021/02/24 20:10:08] WARN: chain[2] - Failed to reset pic (attempt = 2), resp: 0x01 0x01
[2021/02/24 20:10:09] WARN: chain[2] - Failed to reset pic (attempt = 3), resp: 0x9b 0x9b
[2021/02/24 20:10:10] WARN: chain[2] - Failed to start pic app (attempt = 1), resp: 0x9b 0x9b
[2021/02/24 20:10:11] WARN: chain[2] - Failed to start pic app (attempt = 2), resp: 0x9b 0x9b
[2021/02/24 20:10:11] WARN: chain[2] - Failed to start pic app (attempt = 3), resp: 0x9b 0x9b
[2021/02/24 20:10:13] WARN: chain[2] - Failed to reset pic (attempt = 1), resp: 0x9b 0x9b
[2021/02/24 20:10:14] WARN: chain[2] - Failed to reset pic (attempt = 2), resp: 0x01 0x01
[2021/02/24 20:10:16] WARN: chain[2] - Failed to reset pic (attempt = 3), resp: 0x01 0x01
[2021/02/24 20:10:16] WARN: chain[2] - Failed to start pic app (attempt = 1), resp: 0x01 0x01
[2021/02/24 20:10:17] WARN: chain[2] - Failed to start pic app (attempt = 2), resp: 0x01 0x01
[2021/02/24 20:10:17] WARN: chain[2] - Failed to start pic app (attempt = 3), resp: 0x01 0x01
[2021/02/24 20:10:18] INFO: chain[1] - 48 chips detected
[2021/02/24 20:10:19] WARN: chain[2] - Failed to reset pic (attempt = 1), resp: 0x01 0x01
[2021/02/24 20:10:20] WARN: chain[2] - Failed to reset pic (attempt = 2), resp: 0x01 0x01
[2021/02/24 20:10:22] WARN: chain[2] - Failed to reset pic (attempt = 3), resp: 0x01 0x01
[2021/02/24 20:10:23] WARN: chain[2] - Failed to start pic app (attempt = 1), resp: 0x01 0x01
[2021/02/24 20:10:23] WARN: chain[2] - Failed to start pic app (attempt = 2), resp: 0x01 0x01
[2021/02/24 20:10:24] WARN: chain[2] - Failed to start pic app (attempt = 3), resp: 0x01 0x01
[2021/02/24 20:10:24] ERROR: driver-btm-chain.c:465 chain[2] - Failed to init pic controller
[2021/02/24 20:10:24] INFO: chain[2] - Shutting down the chain
[2021/02/24 20:10:24] ERROR: driver-btm-base.c:356 chain[2] - Initialization failed
[2021/02/24 20:10:26] INFO: Start-up temperature is 25 C (min -15 C)
[2021/02/24 20:10:26] INFO: Switching to manual fan control (100 %)
[2021/02/24 20:10:26] INFO: Changing voltage from 21000 to 17400 mV gradually
[2021/02/24 20:11:06] INFO: Raising freq from 50 to 500 Mhz gradually
[2021/02/24 20:11:17] INFO: Switching to automatic fan control (75 C)
[2021/02/24 20:11:17] INFO: Start mining!

Thanks wndsnb, Super insightful as always. How hard would it be to self replace this part? Where would I get the part from? If I can replace the PIC microprocessor then in theory I could use the pickit to flash back on.

[wndsnb]

Don't know if I'd just plow ahead and start replacing things, the boost circuit they said failed has nothing directly to do with the pic so there's likely more wrong than just the pic. The one board I've repaired that had a blown pic also had every 1.8v regulator blown (12 of those on the board), and the boost circuit controller.

But the pic it isn't a very difficult part to replace. Just search for some video tutorials for surface mount rework.

This is the part: https://www.mouser.com/ProductDetail/579-33EP16GS202TE-SS/

Mouser also sells the pickit, but only the newest one. You can get the pickit 3 for cheaper elsewhere.

https://www.mouser.com/ProductDetail/579-PG164140/

This is the schematic of the connector where the 3.3V and i2c signals that the control board uses to talk to the pic.



Here's the schematic for the pic



I got both of those from the repair manuals you can get from zuesbtc here:

https://www.zeusbtc.com/News.asp?Sort=FILES+DOWNLOAD

You can run the guides through online translators to get a reasonably understandable translation. Some of those manuals also have info on using the pickit.

[NoKeyboard-BTW]

I agree, Just to assume that if I do successfully replace the PIC the booster circuit and possible regulators would be next and without digging into that as of now, it could be a non-starter and something that just can't be replaced at a hobbyist level. It sounds like you have had a similar experience with this particular repair on one of your boards so I'm super appreciative of your knowledge and help in putting this into perspective.

A few questions about potentially gearing up for this. It's been a while since I hot-air soldered. What is good out there now? Are there better alternatives, all in one unit, brand to stick to? I see myself doing this in the future so I'd like to prepared. Do you have any recommendations and what tips have worked best for you specifically for the PIC?

Off subject, I am replacing remounting the heat sink on board 3 and ordered a thermal epoxy from Atom Adhesives which are direct replacements for the now discontinued Arctic Silver:

AA-DUCT 902 Silver Epoxy Adhesive, Electrically Conductive, Room Temp Curing
AA-BOND 2153 Thermally Conductive, Electrically Insulating Compound, 2 Part, Thixtropic

I bought both after talking to them directly. Which one of these would you recommend using? Or do you solder yours back on?  

A lot to think about so thanks for the direction here.

[wndsnb]

The problem with the adhesives is that if just re-attaching those heatsinks doesn't work, it's gonna be a pain in the ass to get those heatsinks back off. So if all you're comfortable doing is re-attaching the heatsinks and if that doesn't work you're giving up, that's a fine route to take. But if you'd want to continue down the rabbit hole to try to repair it, I'd suggest re-attaching with solder.

https://bitcointalk.org/index.php?topic=5217949.msg56114781#msg56114781

I now mount all heatsinks the way I describe doing the S17+ in that post. It is just too easy to shift the chip when applying the heatsink if you heat the heatsink and place it on the chip rather than heating the heatsink up in place.

You'll need a test jig to have a reasonable chance at getting boards back up. The standard Bitmain type, like this, you can get at a bunch of different places. It is a S17 control board with special firmware that will exercise the hashboard so you can probe signals with a DMM or oscilloscope. They are pretty cumbersome and slow. I bought another tester from https://tester.asic.repair/en. It is more expensive and doesn't do as in-depth of a test as the normal Bitmain type, but it is 10X faster. With it you can connect the tester and run a test in less than 10 seconds. It takes a few minutes to do the same with the bitmain tester.


Here are some replacements for other parts:

Boost circuit controller https://www.digikey.com/en/products/detail/monolithic-power-systems-inc/MP1517DR-LF-Z/9433294

Boost circuit 1.8V regulator https://www.digikey.com/product-detail/en/monolithic-power-systems-inc/MP2019GN-Z/1589-1557-1-ND/9433323

1.8V regulator https://www.digikey.com/product-detail/en/on-semiconductor/NCP114ASN180T1G/NCP114ASN180T1GOSCT-ND/6560646

25mhz oscillator https://www.digikey.com/product-detail/en/diodes-incorporated/FK2500065Z/FK2500065ZDICT-ND/9952907

For ASICs, I've ordered from Zuesbtc and Aliexpress.

https://www.aliexpress.com/item/4000098353215.html?spm=a2g0s.9042311.0.0.11884c4dXGgZuU

https://www.zeusbtc.com/ASIC-Miner-Repair/Parts-Tools-Details.asp?ID=165

[NoKeyboard-BTW]

It looks like I'll be referencing this post for a while so thank you again for this information. Something I know you have experienced with periods of trial and error in your own experiments/research/documentation. After reading your feedback on the adhesive I just threw it in the fridge for emergencies....down the rabbit hole this goes. I actually took a visit to our electrical dept. today and they are using Pace units for desoldering and surface mount work. They said the PIC work could get tricky because each one of the PIC prongs will have to be carefully soldered. Ordered the PIC chip anyways. With some of their recommendations and a little research, it's between Weller WT's (for the tweezer compatibility) + the Hot Air WTHA1 or Pace ST115 with Sodr-X-Tractor. Looking at these numbers maybe take it to a professional with the right tools locally with hardware provided from your list. I will be picking up one of the test jigs soon for diagnosis purposes and got a pickit just in case by a miracle I come across a working 53 pro board.   

https://www.weller-tools.com/professional/USA/us/Professional/Product+lines/WT+Line/Soldering+stations+and+sets/WT2M

https://www.weller-tools.com/professional/USA/us/Professional/Product+lines/WT+Line/Soldering+stations+and+sets/WTHA1

or

https://paceworldwide.com/st115-digital-desoldering-station-sx-100-sodr-x-tractor

[wndsnb]

With some solder paste and a hot-air tool of some sort applying a package like that pic is pretty easy. You can be pretty messy with the solder paste, and when you heat it up to the melting point with a hot air tool the solder will flow to the pins, so no need to worry about individual leads of the package. You can go back after reflowing the solder to clean up any bridges that might be caused by excess solder. Recently I've been doing that by applying flux over the pins, applying heat with the heat gun, and then running the blade of a x-acto knife over the solder, this does a great job of sucking up the excess. But many of these techniques are all personal preference, and what works for me might not work for you. So the best is to find some old electronics to rip apart and practice with. Find similar packages and remove and replace them. better to practice with worthless junk than a hashboard that could be worth a lot if/when repaired.

Here's an example of removing/replacing using an iron https://www.youtube.com/watch?v=HGgat5IVfFE

I normally do this by putting down a bead of solder paste over the pads, place the part, and then heat the whole thing up with my heat gun to reflow it all at once.

More like this: https://www.youtube.com/watch?v=f_yFDpSTfao


And the weller/pace/hakko/... stuff gets pretty expensive. Some is worth the $, but some not. I keep an eye out for a decent deal on used items on ebay.

But I'd just start with some lower cost ones, they may not have as high of a build quality but as long as you don't get complete no-name brands most will do the job just as well as the $$$ versions. For a hot-air tool, you just need to be able to set airflow and temperature and need something around 1000W. For an iron, you need one that you can set temperature. Those are the only necessary requirements I think. I have never used de-soldering tools, not sure they are of much use for surface mount components.

So for a decent quality low cost iron, I use this: https://www.tequipment.net/HakkoFX888D-23BY.html?v=118031

And as I said earlier, I just have been using a heatgun with temperature and airflow settings , but a reasonably priced hot-air tool like this one I'm sure would work fine

https://www.tequipment.net/Quick/861DW/Desoldering-Equipment/Rework-Stations/?v=7450

Here's a pretty in-depth review of it: https://www.youtube.com/watch?v=N_yHkrVYrBE

In fact, I just ordered one of those after watching that review.

[NoKeyboard-BTW]

Invaluable information wndsnb. It's a great start and this thread has become a great resource.  I ordered all the parts (and a little bit more for back ups) you listed including chips for the 53 pro boards along with a stencil jig for the chips. Have some old electronic stuff to practice on following your tips once product research is complete over the weekend. After some other research into the Bitman test jig and the Asic Repair Center Jig, I'm not sure which one to purchase because it seems the Bitmain jig has all the more comprehensive information and slower but the ARC is faster and more user friendly where you don't have to do a lot of deep diving and interpretation, has a board tester option and EEPROM flash as options as well. Do you use lead or lead-free solder paste and have any best recommendations? Also, I need to look at a new DMM or oscilloscope - is one preferred over the other for this specific application and perhaps a specific one to look at?

[wndsnb]

I listed the exact solder and flux I use here: https://bitcointalk.org/index.php?topic=5217949.msg56114781#msg56114781.

For multimeters, I always go with Fluke, https://www.tequipment.net/Fluke117.html. You'll want to make sure the one you get can measure diode voltage drop, circled in green here:



For an Oscilloscope, a decent low-cost option is this: https://www.tequipment.net/Rigol/DS1102Z-E/Digital-Oscilloscopes/?v=7401 or this: https://www.tequipment.net/Rigol/DS2102E/Digital-Oscilloscopes/?b=y&v=7906 (this is the series I use).

For testers, it's hard to say what route would be best. Most issues can be found with the ARC tester. The BM tester will identify chips that are not hashing at full speed, but you can do that same thing in a miner with vnish firmware since it will show you the performance for each chip separately.  I guess I would probably lean towards the ARC tester. If you get one of those, you'll also need to get a separate lab PSU, it does not use a APW9 like the BM tester does.

[NoKeyboard-BTW]

Just took an unnecessary dive into oscillators but I think MM Fluke MM will do just fine and I'd like to keep it portable for multiuse. The portable digital oscillators on Amazon are temping for a cheap price. Micsigs are nice but just out of budget. For jigs its looking like the ARC with a good point about monitoring chip hash in Vnish. Do you recommend adding the board testing and EEPROM options with the ARC and what lab PSU are you using? I watched their video and looks like they just clipped power onto the board.

[wndsnb]

You need a 20V or greater, 10A lab power supply. I'm using one of these that I found on ebay: https://www.tequipment.net/InstekPSP2010.html?v=7421.

If you want to mix and match boards with stock firmware then you'll need some way to program the eeprom. From what I can tell, the ARC tester just programs in some default nominal values, so it is good if those values work for the hashboard you're working on. But if the hashboard doesn't run well at the frequency/voltage, there isn't anything you can do about it. I just asked them about it and told them it would be a nice feature to enable editing of those values. So who knows, maybe they'll implement it. Their support has been pretty good so far, I message them on Whatsapp and they have responded within 15 minutes every time I think.

Also, the Bitmain tester is just a S17+ control board with different firmware. They hook an lcd screen to it, but it really isn't necessary. I'm pretty sure you can take a stock s17+ control board and run the test fixture firmware on it. You'd just need to buy a USB to serial adapter, like this: https://www.sparkfun.com/products/15096, and hook it up to the uart signals on the control board.

Here's what my workbench currently looks like... a bit messy.

[NoKeyboard-BTW]

Thanks for the lead on the power and the glimpse of a mad scientist's workbench. I have yet to see a clean workbench, ever. As it should be. Great equipment setup. It's hard not to overlook, I have a Hakko myself - it's such a great design. Agreed, I think Vnish offers the path of least resistance when attempting to mix and match. Perhaps Marc from Asic.to could confirm this one day or Vnish himself because this is what I would assume a lot of x17 series miners are up against. Hopefully, ARC implements your suggestion and it's not "extra"

[wndsnb]

https://www.tequipment.net/Quick/861DW/Desoldering-Equipment/Rework-Stations/?v=7450

Here's a pretty in-depth review of it: https://www.youtube.com/watch?v=N_yHkrVYrBE

In fact, I just ordered one of those after watching that review.

Got my new hot-air tool and tried it out last night for the first time. It works great, bang for the buck factor is very high...

[NoKeyboard-BTW]

Awesome to hear it passes the test and what's great is it's about half the price of others. Did you buy any additional tips or have any go-to's for the job? Still waiting on all the board hardware to arrive. Shipping has been really affected lately and/or inventories. 

[wndsnb]

I got the 12.7mm nozzle, and I'm using that. It is just the right size to cover the ASIC. That's the other nice thing, the most nozzles are only around $10.

[NoKeyboard-BTW]

I was expecting at least close to $50 for tips. This is welcome on the wallet. Circling back in regards to the PIC chip, when preparing to take the PIC chip off of the board should I use the PIC IT to take that PIC information off first or could I just as well use the 1 working board's PIC information later when the chip is installed on this non-working board?

[wndsnb]

The pic firmware from the other boards in the same miner would be the same, so either way would work.

[NoKeyboard-BTW]

All the hardware besides ASIC chips from AliExpress arrived - the shipment was "canceled" by them. Maybe they didn't actually have them in stock so I just asked for a refund and didn't ask any questions - on the search to find a reliable source for BM1397AD's. On a whim, I took the boards to a large electronic repair place locally and they wouldn't touch them. wndsnb, I was circling back and taking a look at your setup. Are you using a stereoscope or digital? Looking around Ebay and have come across some Nikons and Zeiss Stemi 2000's. Taking a step back from there and thinking about budget, is there something that will decently get the job done for these smaller chips that you looked at?