Topic: [GUIDE] Undervolt antminer s1 [1.19W/GH at the wall] (Read 69360 times)

Does anyone know where I can find a frequency list for these from 200-300? I am currently making stacks of these as well although I am not pushing them down as far as you are. I have a technician replace the surface mount resistors for me.

I run them all solar so for me it is not about getting them low as possible it is just about getting them low enough. I have two units/four boards running right now at 290GH for about 310W @ the wall. I built a RPi stratum proxy that I am using to reconfigure the mining cluster when needed.

Thanks for the help.

Yeah it seems like 0.70V is the absolute limit for these chips.

same here. <0.72 some chips start to get a bit iffy, and <0.68 i found that some chips dont initiate or produce high errors

Just wondering what the lowest voltage was possible to achieve with this chip? I got an S2 that I am undervolting and can't get it stable under 0.70V

Hey guys I am back :-)

so where are we on all this ?

Also I finally got my PSU adapter done :-) for the Delta DPS-800-BG A and guess what I even had 2 Antminer S3 overclocked to 890 Watt and 1000GHS on it running just fine for over 24h no problem.

http://www.lumanetsolarenergy.com/index.php?route=product/category&path=81_85

Back to the other topic :-)

SO I do have like 6 Antminer S1's and I done the swap of the resistors but now I like to even get rid of the control board and use USB adaptors instead to have than with a raspberry PI all 12 or more in a 19" rack stacked up and still outperforming the S5 chips or at least getting close to it.

So how far are we on it???

Olaf from LSE LLC (Lumanet)

I ment that I stack 6 hashboard with spacers and put a fan on eatch end if/when I use original controllers one pair of boards don't have fan connection.


Silly me it was easy test, I took the 4-pin original connector off and miner worked with one extra 3-pin fan.

you need a fan if you stack the boards, or youll see everything reading in the 65-75C range. I put all my boards on 1" spacers and cool it with a pair of silent ~70cfm cougar fans. Its barely audible and brings the temps down to 55-65C even while tucked in a cubby. (in a better open location it might be 50-60C). This is without heatsinks

I have undervolted my S1s with 10k trimmers to 0.85V and 0.75V.

Spring and summer is coming and I want to mod the rest of S1 space heaters to 0.75V but I can't find a table for settings of MHZ near 200.

200Mhz didn't work at 0.75V I had to use 225 and got over 1% errors, I did find some S2 settings but they have totally different timing setting (smaller values).

I want stack 6 plates with original controllers and 2 fans, do boards without fan work?

I had the "x" problem, too. Maybe I still have.  

Here is what I did and tried:
1. Undervolt to ~0.85V with pencil mod, everything fine on various MHz-Settings, no x´s.

2. Undervolt to ~0.78V with pencil, on 2 of my ants the x´s show up. Usually the first 8 in first row, and sometimes even more, both rows. Sometimes I was able to get rid of them by reboot, sometimes they disappered alone. After a lot of trying clocks/timeouts from 162MHz up to 240MHz I was about to give it up. Nothing worked properly. Some days the x´s vanished, then show up again randomly. The pool hashrate also showed that the affected miners ran okay for several hours, then hashrate suddenly dropped from ~120 to 100 or even 80 GH/s and stabilized there. After all, I thought it was the pencil, because I was not able to get exact the same Voltages to all ASIC-sections of the antminer.

3. I removed the pencil, and soldered 4.7k resistors in parallel to our R3 and friends. Now I got a stable 0,776V (Load) - 0,779V (Idle) at each section of the hashboards. The Antminer is pulling now ~10,5Amps @ 11,97Volts DC and is hashing at ~120GH/s with 237MHz. So far so good. (Before it pulled ~13Amps with the 0,78V pencil mod) But when I moved my rig into the garage in november, the damn x´s got back.
Random drops from 120 to 100, on the other ant from 120 to 90 GH/s.
Temperatures in garage were around freezing, and I got it to barely 10 Celsius with the rig (4 Antminers total, and some Litecoin ASICs) running 24/7. Good to work some hours a day on my cars. COOLING INFO: even at 10 Celsius the fans ran so slow that you were able to count the blades on them. ASIC showed around 20 Celsius. Nice silence.

4. Since december the rig is back in my house, in my living room as a little helper to cut heating costs. Suddenly the x´s disappeared after 2 days of (intermittent) running! (I need to shut the rig down when nobody is at home, to avoid risk of fire.)

Ok, so what could have caused the x´s? I believe it is only temperature and voltage that can cause it yet. I think the capacitors on the boards are aged, and some of them fail in cold conditions. Especially with voltmod, maybe the voltage gets very low under sudden load for a short time, causing the ASICs to crash... dont know.  
And I do not know how well or bad the pre owners treated the miners, got all my 4 ant´s at the bay.
After a little research, I also think my voltage is a little low, maybe too low to power all the ants properly. Got 2 HP DPS-800GB with backplane that power my rig. The litecoin-section surprisingly NEVER crashed with this setup, at the same 12V rail.

Now I run 1 of the suspicious ants in the garage on a 12,34V PSU (instead of 11,97V from my HP proliant powermonster), to prove if voltage does matter. Actually running 4 hrs without x´s.
Maybe I got it....

Next step is to get closer to 0,75V... a 3,9k Ohm in parallel gets me 2,1k and a resulting 0,76V under load. I will report if it works fine and how much the miner is pulling then in DC and AC.

At .75v 193mhz i get all x on first two rows of both chips at 193mhz
If i set a high timeout i get better ghs but xs still come just not as fast. Around 30-75ghs

How are you optimizing units some read .74 to .75 should they be more stable maybe remove some lead...

Mine is a jan or feb 2014 antminer s1 on the stamp on pcb are these different

I really want an extreme undervolt it was so stable overclocked. Should I go up to .8v

Also off topic i notice my fan on corsair 500 bronze doesnt spin anymore on the s1 since undervolting.  Is it because watts are so low or should I try another psu.  I believe I had same results on my gold psu with x's but fan spun on that psu.

A little off subject, but what program are you using to get the "Rig Stats" displayed?

Why not do away with the fans, fun at full voltage, and just submerge the stack in mineral oil ?

living on the edge :-p

edit:

if they do come off, voltage to the chip drops to 0.5

I would recommend against installing the pots as CHOASiTEC did. That will work fine as long as nothing goes wrong, but you just have the leads of the pot connected to the pad. Since there's no resistor in parallel if you accidentally pull the pot off, you'll destroy the chips. It's safer to leave the resistor on and put the pot in parallel.

why not just add a second fan to the stack, either side-by-side or as a push-pull configuration? that would probably do the trick. TBH i have found higher temps dont seem to be a big issue - I disconnected the fans on some of my undervolted/underclocked S1 units that were still in the original frame/heatsinks, and they peaked at about 58-65 C after maybe 20min, and seemed to hash just fine at that temperature. With the fans connected they run below 40C

Once my standoffs arrive ill probably aim to get 48-54 C on the boards when stacked similar to your setup

10K multiturn trimpot, two pins laid across the top feedback resistor? That'd be a snazzy setup.

Could you post a photo of how you have the added resistors installed? I'm running 36x S1 at 140gh / 175 watts now, but this is the next logical step since I'm out of electricity.

this is what im using: