Antminer S9 March 4th, 2019 firmware experience?

mettalmag

legendary

Activity: 1084

Merit: 1003

≡v≡

Quote from: mikeywith on April 10, 2019, 03:05:15 PM

can you try this method > https://bitcointalksearch.org/topic/m.50519867

i had a batch that did not allow me to change the firmware, it also refused ssh connection , it's not exactly the same issue, but give it a shot.

4 out of 35 machines are at 0 hashrate
I have no physical access right now but I will try that method
and also thinking to try third party OS maybe

AlecMe

full member

Activity: 392

Merit: 159

Hi all,

Anyone knows what could be the problem that can cause this:

"Chain[5] Chip[62] junction temperature=136"

Everything else at the end of the log seem to look ok:

"Done check_asic_reg
Chain[5] Chip[244] pcb temperature=62
Chain[5] Chip[62] junction temperature=136
Special fix Chain[5] Chip[62] middle Temp = 77
Done read temp on Chain[5]
Chain[6] Chip[244] pcb temperature=65
Chain[6] Chip[62] junction temperature=71
Special fix Chain[6] Chip[62] middle Temp = 80
Done read temp on Chain[6]
Chain[7] Chip[244] pcb temperature=62
Chain[7] Chip[62] junction temperature=66
Special fix Chain[7] Chip[62] middle Temp = 77
Done read temp on Chain[7]
Max pcb temp : 65
Set fixed fan speed=80
FAN PWM: 80
read_temp_func Done!
CRC error counter=0"

Artemis3

legendary

Activity: 1988

Merit: 1561

CLEAN non GPL infringing code made in Rust lang

Quote from: maxmad on July 29, 2019, 02:30:03 PM

To bypass the error while trying to flash to factory, i used the SD recovery method.
This time the flash was successful to NAND and S9 boot to factory OS.

Then i installed the 201711171757 and all three boards were working again!
Seems that 201711171757 is really a saver.

I tested to flash to latest 2018 firmware 201812051512 with success and all boards properly worked. Including of temp sensors!

I suggest you stay there then, that's asicboost enabled while still not locking you out of your miner.

maxmad

jr. member

Activity: 55

Merit: 3

This is a good thought.

However, if something will block the airflow, ie dust or even a fan failure, this will be catastrophic.
The strange part is that bOS detects the sensors:

Code:

Mon Jul 29 15:17:37 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:37.036] AUTOMATIC fan control, target temperature 75 degrees (hot is 90 deg, dangerous is 95 deg)
Mon Jul 29 15:17:37 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:37.348] chain 5: found sensor TMP451 at chip_addr=f4, i2c_addr=98
Mon Jul 29 15:17:38 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:38.853] chain 6: found sensor TMP451 at chip_addr=f4, i2c_addr=98
Mon Jul 29 15:17:40 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:40.358] chain 7: found sensor TMP451 at chip_addr=f4, i2c_addr=98
Mon Jul 29 15:17:46 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:46.153] set_pic_voltage: chain 5: setting voltage 8.70V (PIC=125)
Mon Jul 29 15:17:47 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:47.311] set_pic_voltage: chain 6: setting voltage 8.90V (PIC=91)
Mon Jul 29 15:17:48 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:48.469] set_pic_voltage: chain 7: setting voltage 8.90V (PIC=91)
Mon Jul 29 15:17:48 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:48.588] Probing for an alive pool
Mon Jul 29 15:17:49 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:49.188] Pool 0 difficulty changed to 8192
Mon Jul 29 15:17:49 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:49.195] Pool 1 difficulty changed to 8192
Mon Jul 29 15:17:49 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:49.535] Rejected untracked stratum share from pool 0
Mon Jul 29 15:17:49 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:49.589] Network diff set to 9.01T
Mon Jul 29 15:17:53 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:53.758] sensors: 5/0 temperature read failed, using previous
Mon Jul 29 15:17:53 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:53.758] sensors: 5/0 temperature (0.0,0.0)
Mon Jul 29 15:17:54 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:54.424] sensors: 6/0 temperature read failed, using previous
Mon Jul 29 15:17:54 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:54.425] sensors: 6/0 temperature (0.0,0.0)
Mon Jul 29 15:17:54 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:54.590] API running in IP access mode on port 4028 (5)
Mon Jul 29 15:17:54 2019 daemon.info cgminer_monitor.lua[1531]: state sick because low hashrate
Mon Jul 29 15:17:54 2019 daemon.info cgminer_monitor.lua[1531]: /sys/class/leds/Red LED mode blink-slow
Mon Jul 29 15:17:54 2019 daemon.info cgminer_monitor.lua[1531]: /sys/class/leds/Green LED mode off
Mon Jul 29 15:17:55 2019 daemon.err cgminer[6644]: [2019-07-29 15:17:55.090] sensors: 7/0 temperature read failed, using previous

To bypass the error while trying to flash to factory, i used the SD recovery method.
This time the flash was successful to NAND and S9 boot to factory OS.

Then i installed the 201711171757 and all three boards were working again!
Seems that 201711171757 is really a saver.

I tested to flash to latest 2018 firmware 201812051512 with success and all boards properly worked. Including of temp sensors!

Artemis3

legendary

Activity: 1988

Merit: 1561

CLEAN non GPL infringing code made in Rust lang

Quote from: maxmad on July 25, 2019, 08:37:57 PM

OK Braiin OS installed to NAND. ( Using 16GB card, with 8GB there was an error "firmware.tar.gz" is missing )
Seems that the missing hashboard appeared and hashing.

However in Braiin OS there is no temperature measurement!

...

Braiin OS seems very powerful firmware but missing temperatures sensors is very dangerous.

So i tried to install 201711171757 bitmain firmware from official page with no luck

Well you can also manually control the fan speeds. If you know it never got to dangerous temperatures with factory, it won't with Braiins OS unless you overclock it. It seems to be unable to detect the sensors but that's not so strange as you may think. It would be fine if you could get temperatures from one of the boards, then put that in the middle (which is the worst position).

You can try factory from 2017 and 2018 if you think it makes a difference. But yes, factory loves to disable a board when it can't read its sensors, something bOS can override as you did see. Sensors failing but hash chips actually ok is not rare.

If you are somehow still scared with bOS go the other route: Downclock. Something resembling 8.1v 500mhz does wonders. You could try 8.5v 600mhz or so, just make sure actual and ideal clock speed match or are very close, then keep lowering that voltage and when the values differ too much go back to the previous value. Downclocking is never dangerous, chips run cooler. And don't mind the suggested values, lower those voltages always until you find the sweet spot.

I mean a downclocked board is better than no board, but its your choice...

maxmad

jr. member

Activity: 55

Merit: 3

OK Braiin OS installed to NAND. ( Using 16GB card, with 8GB there was an error "firmware.tar.gz" is missing )
Seems that the missing hashboard appeared and hashing.

However in Braiin OS there is no temperature measurement!

and after a while 2 hashing boards stop hashing

Code:

Thu Jul 25 19:59:11 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:11.130] Accepted 06cba9a4 Diff 9.64K/4327 BC5 0
Thu Jul 25 19:59:11 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:11.522] sensors: 7/0 temperature read failed, using previous
Thu Jul 25 19:59:11 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:11.522] sensors: 7/0 temperature (0.0,0.0)
Thu Jul 25 19:59:11 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:11.522] sensors: no working sensors
Thu Jul 25 19:59:17 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:17.432] Version submitting share mask 0x00c00000 with work version 0x20c00000 and pool version 0x20000000
Thu Jul 25 19:59:17 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:17.517] Accepted 6dddf4ae Diff 153K/4327 BC5 0
Thu Jul 25 19:59:18 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:18.203] Version submitting share mask 0x00800000 with work version 0x20800000 and pool version 0x20000000
Thu Jul 25 19:59:18 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:18.288] Accepted 0241bc1f Diff 29K/4327 BC5 0
Thu Jul 25 19:59:20 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:20.351] sensors: 5/0 temperature read failed, using previous
Thu Jul 25 19:59:20 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:20.351] sensors: 5/0 temperature (0.0,0.0)
Thu Jul 25 19:59:21 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:21.020] sensors: 6/0 temperature read failed, using previous
Thu Jul 25 19:59:21 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:21.020] sensors: 6/0 temperature (0.0,0.0)
Thu Jul 25 19:59:21 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:21.127] Version submitting share mask 0x00800000 with work version 0x20800000 and pool version 0x20000000
Thu Jul 25 19:59:21 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:21.310] Accepted 0d27c29e Diff 4.98K/4327 BC5 0
Thu Jul 25 19:59:24 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:24.932] sensors: 7/0 temperature read failed, using previous
Thu Jul 25 19:59:24 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:24.932] sensors: 7/0 temperature (0.0,0.0)
Thu Jul 25 19:59:24 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:24.933] sensors: no working sensors
Thu Jul 25 19:59:33 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:33.471] Version submitting share mask 0x00000000 with work version 0x20000000 and pool version 0x20000000
Thu Jul 25 19:59:33 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:33.557] Accepted 020c7705 Diff 32K/4327 BC5 0
Thu Jul 25 19:59:33 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:33.765] sensors: 5/0 temperature read failed, using previous
Thu Jul 25 19:59:33 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:33.765] sensors: 5/0 temperature (0.0,0.0)
Thu Jul 25 19:59:34 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:34.432] sensors: 6/0 temperature read failed, using previous
Thu Jul 25 19:59:34 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:34.432] sensors: 6/0 temperature (0.0,0.0)
Thu Jul 25 19:59:38 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:38.344] sensors: 7/0 temperature read failed, using previous
Thu Jul 25 19:59:38 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:38.344] sensors: 7/0 temperature (0.0,0.0)
Thu Jul 25 19:59:38 2019 daemon.err cgminer[6079]: [2019-07-25 19:59:38.345] sensors: no working sensors

Braiin OS seems very powerful firmware but missing temperatures sensors is very dangerous.

So i tried to install 201711171757 bitmain firmware from official page with no luck

Artemis3

legendary

Activity: 1988

Merit: 1561

CLEAN non GPL infringing code made in Rust lang

Quote from: ccgllc on June 07, 2019, 12:16:39 PM

Not sure why. Suggest using a 16GB or small memory card.

Because SD, SDHC and SDXC each support different max capacities. If the port is SDHC, putting an SDXC card in there isn't going to work. A sure bet is always keep a small SD card (2gb max). But if someone mentioned 16gb it might mean it is SDHC. I never tried with such big capacity cards (wasteful). There is small chance you got a fake one, which are usually printed with large capacities (but are much smaller in reality). Best thing about low capacities is that nobody bothers faking them...

To OP: Try booting BraiinsOS, (same jp4 jumper) if it boots, try installing it into the nand using its web ui. Move back jumper and see what happens... If it still boots Bitmain's firmware, then the nand my be damaged. Unlike Bitmain's, you don't have to install Braiins OS into the nand to use it, just leave it with the smallest (cheapest) SD card you can find and let it always boot from sd. I don't think a new controller is worth buying when any cheapo 1gb microsd card fixes it.

BitMaxz

legendary

Activity: 3234

Merit: 2943

Block halving is coming.

Can you try to flash it to Braiin OS first through SD card after successfully flash it check the miner in the browser if the interface change to Braiins OS.

If successful then you can flash it back to bitmain image.

Then check it again if nothing change and the OS still BraiinsOS it means that your image from bitmain you download for SD card is corrupted you need to download the firmware again without using any 3rd party downloader browser should be enough to download the image/firmware for SD card.

Or maybe it's not flashing properly because the terminal has full of dirt/corroded. So check the SD card slot and maybe the terminal is corroded/full of rust that is why it's not flashing properly. If you have Laquer flo thinner you can clean them and resolder the terminal may solve your issue.

maxmad

jr. member

Activity: 55

Merit: 3

Made 4 flashes through SD card using 2 different SD cards.

Still the board has the same 201905311059 latest firmware.

No way to downgrade.

i want to use an older firmware to test if i can alive a dead hashing board

https://bitcointalksearch.org/topic/a-quick-fix-for-antminer-s9-missingnot-showing-hash-board-no-skills-needed-d-5034849

Jumper flipped, both led are flashing after some seconds but the board has still this latest firmware installed!

ccgllc

copper member

Activity: 658

Merit: 101

Math doesn't care what you believe.

Yes... the latest (anything tagged 2019) firmware is the reason one needs to go through the pain of reflashing via the SDHC port.

tw0rak

hero member

Activity: 600

Merit: 501

Im probably doing sth wrong with the SD card - as if there is none in the slot, the board act the same.

wait a second...

EDIT: Yes - friday evening - thats the problem. I was using file S2S4S5S7-img.img - and now Ive used files for T9+ and... voila it works. Thanks for your patience!

Maybe another one:

Is it possible to do with the very latest bitmain firmware for S9? As I made it now at 3 miners w/o problem but this one Im not able to downgrade.

ccgllc

copper member

Activity: 658

Merit: 101

Math doesn't care what you believe.

Not sure why. Suggest using a 16GB or small memory card.

We followed the following steps which worked on over 100 controllers (from repetitive memory):

1) Pull controller from chasis
2) Flip jumper 4 (the one on the outside of the jumper block)
3) Attach power cable from standalone PSU
4) Power up PSU
5) Wait for a few minutes
6) Power down PSU
7) Flip jumper 4 back to normal
Cool

Disconnect power
9) Reinstall controller

tw0rak

hero member

Activity: 600

Merit: 501

Hi guyz,
Im asking for help - Im absolutely NOT new in this but I just cant flash S9 board from the SD.
Everything tried and checked multiple times - jumper is in the right possition, both leds - red and green are shining SOLID and never start to blink.
Doublechecked if there is that IMG uploaded with Diskimager, tripplechecked the jumper is right...
After all this (never starts blinking) I have to disconnect it and there is still the same old (well better to say new) firmware.
Any help?

EDIT: I have many S9s and just tried the same way at another one with very same results...I do not absolutely know how to continue...

ccgllc

copper member

Activity: 658

Merit: 101

Math doesn't care what you believe.

Quote from: JohnDon87 on April 22, 2019, 01:17:34 AM

I got it. Although the instructions weren't thoroughly clear on the bitmain page where to switch the jumer. It took alittle working on it but I finally did it. Thank you!

Too say the least. Took me several back-and-forth e-mails with Bitmain tech support to figure it out. But for others, its Jumper 4 - a 3 pin jumper. Simply change it to cover the other two pins. Then switch it back afterwards.

Note: we have been using this as an excuse to clean our machines. Once the hard cards are out, there is room to slide the control card out (if you disconnect the front fan). You can boot the sdhc card without any any cards or fans connected. We simply place a block of wood on top of the case, place the control card on that, plug into a space power supply, and clean the hash cards while the sdhc card is doing its thing. Then reassemble.

JohnDon87

newbie

Activity: 17

Merit: 0

Quote from: ccgllc on April 21, 2019, 11:53:57 PM

It doesn't "boot" from the sdhc, it resets itself. After a minute or so, both the red and green like should be blinking together. Once that happens, power down, put the jumper back, reassemble, and it should boot back to the original factory firmware.

I got it. Although the instructions weren't thoroughly clear on the bitmain page where to switch the jumer. It took alittle working on it but I finally did it. Thank you!

ccgllc

copper member

Activity: 658

Merit: 101

Math doesn't care what you believe.

Quote from: JohnDon87 on April 21, 2019, 09:44:10 PM

Hey ccgllc, I just mistakenly installed the new 2019 firmware on my antminer S9. I am also pretty sure I blew out a hash board in doing so as well. I installed the T9+ toolkit and moved the 4 jumper to the right position. Miner still isn't booting off the SD card. I am only running 2 hashboards at enhanced lower power now. I am running 110V power from the wall on a 1850W PSU. How can I troubleshoot the hashboard and how did you boot the S9 miner from the SD card?

It doesn't "boot" from the sdhc, it resets itself. After a minute or so, both the red and green like should be blinking together. Once that happens, power down, put the jumper back, reassemble, and it should boot back to the original factory firmware.

JohnDon87

newbie

Activity: 17

Merit: 0

Quote from: ccgllc on April 17, 2019, 09:30:28 PM

Which is not true. You can downgrade via the sdhc card as outlined at the Bitmain site. We have down well over 20 downgrades without a hitch, more almost every day.

https://support.bitmain.com/hc/en-us/articles/360019493654-S9-series-S9-S9i-S9j-S9-Hydro-Control-Board-Program-Recovery

Hey ccgllc, I just mistakenly installed the new 2019 firmware on my antminer S9. I am also pretty sure I blew out a hash board in doing so as well. I installed the T9+ toolkit and moved the 4 jumper to the right position. Miner still isn't booting off the SD card. I am only running 2 hashboards at enhanced lower power now. I am running 110V power from the wall on a 1850W PSU. How can I troubleshoot the hashboard and how did you boot the S9 miner from the SD card?

mikeywith

legendary

Activity: 2170

Merit: 6279

be constructive or S.T.F.U

Quote from: philipma1957 on April 18, 2019, 10:22:42 AM

[...]

ya mathematically it does seem like a good "plan"- if i go with 810w instead of 1250w i will do a lot of room to fill in, but that means new wiring,racks etc, right now i have no near future plans for investing more in mining after that last batch of S9s i got.

also the next step would be moving to much more efficient gears, T15/S15 is what i am thinking, this all depends on how much will my S9s worth at that time as well as the price for those T15/S15.

Quote from: Artemis3 on April 18, 2019, 02:57:05 PM

AND you can go the other way, overclocking, if you don't care about spending more per hash. 16T or so is possible using the 1600w PSUs or a bit more with the 1800w ones.

Not a big fan of overclocking.

Quote from: Artemis3 on April 18, 2019, 02:57:05 PM

I suggest you officially post the bounty offer (Service demand?) in the Services area to get things moving..

does not seem like a bad plan, will do that , not that i really need it, but willing to share it with the community.

what would be a good title for the topic in the services section? something that can attract potential programmers who may be capable of breaking this firmware security.

Artemis3

legendary

Activity: 1988

Merit: 1561

CLEAN non GPL infringing code made in Rust lang

Quote from: mikeywith on April 18, 2019, 09:14:35 AM

@Artemis3

Non of the above factors matters to me as i have totally free power for all my gears, the reason why i never went with BraiinsOS in the first place was the extra time needed to flash all the gears, i also don't know much exactly each s9s consumes with this 2019 firmware, but by looking at the voltage regulators that are installed for some of them its something between 1250w and 1300w more or less.

So technically i have no problem with the 2019 firmware (yet), but as far as the "bounty" goes i would be willing to make a good donation to the first person who break this firmware, tho i am not a big fan when it comes to bounties in such cases, as most of us here provide all kind of information and tools for free, but still if you or anyone else for that matter wants a bounty for it - i will be the first to reward them.

Hmm too bad you didn't ask in the Braiins OS thread. It is possible to install it to all of them using a python script of theirs. AND you can go the other way, overclocking, if you don't care about spending more per hash. 16T or so is possible using the 1600w PSUs or a bit more with the 1800w ones.

I suggest you officially post the bounty offer (Service demand?) in the Services area to get things moving...

ccgllc

copper member

Activity: 658

Merit: 101

Math doesn't care what you believe.

Quote from: Artemis3 on April 18, 2019, 08:09:53 AM

At some point you should be seriously scheduling this. I got an independent report from a miner that reached 78t/w on S9s using BraiinsOS. You have a serious profit incentive to do so... And the longer you take, the more you lose.

I've come close to that number three times using the firmware-that-shall-not-be-named (or my post will get deleted, which it likely will anyhow):

550MHz at 8.0V yielded 11,486GH/sec using 905 Watts at the wall for 78.8 J/TH
475MHz at 8.2V yielded 10,300GH/sec using 800 Watts at the wall for 77.7 J/Th
475Mhz at 8.1V yielded 10,100GH/sec using 775 Watts at the wall for 76.7 J/Th - I was shooting for 75J/TH but never made it.

Note that only a few of my S9s can pull 550Mhz at 8.0Vs off, most have at least one hash card that simply needs more voltage. 550MHz at 8.2 volts is generally stable on most 14TH machines, with an rare but occasional card needing 8.3V to avoid auto-tuning significant numbers of chips to a lower frequency. 8.3V is pretty commonly needed on my 13.5TH machines, and 8.3-8.5V on my 13TH machines (later batches, not the original early batches).

Topic: Antminer S9 March 4th, 2019 firmware experience? (Read 1173 times)