[GUIDE] Undervolt antminer s1 [1.19W/GH at the wall]

[Lumanet]

I do use a HP DPS-800GB which gives me a nice 69AMP Rail. I not sure if I would even mess with an 80+ Gold Power Supply those aren't build for 24/7 operations and will sooner or later fail on you. If I would remove my network Switch and maybe the low quality fan's which the Antminer has I might be able to reduce the costs a bit but I am already very happy with what I archived with those old S1's before I upgrade to some high end gear.

How many month do your S1's did ran before you downgraded them ? Mine ran almost 2+ years straight 24/7.

I'm running the majority of them on Emerson DS450 server PSUs, and have a couple plugged into a cheap 850W HEC power supply I got for $40.
http://www.tigerdirect.ca/applications/SearchTools/item-details.asp?EdpNo=7621664
I got the first one mid January, so I guess I ran them almost 9 months before I modded them at the start of September.

Don't get me wrong, I'm not trying to be antagonistic here. The reason I asked about the power supplies was that using the kind of super cheap PSU of the kind that comes free in a $30 case is about the only reason I could see getting 2.75J/GH at stock settings. Pretty much everyone with a decent PSU is measuring closer to 2J/GH.

I would double check your measurement equipment, I have a feeling something is malfunctioning or hooked up incorrectly. Your setup is likely drawing much less power than you're reporting here.

Not sure I do run this one infront of the hole setup so Switch, Outlet , PSU, Antminer all are measured otherwise you don't get a clear picture. One thing most people forget it the Cooling of the room. I am in NEW MEXICO so Desert and have my setup in the Garage. WHich itself is heated up to almost 90 degree with the Antminers now which is nice and warm :-)

Your switch shouldn't be drawing 200W, and unless that 90 is in C that shouldn't be a problem either. Even pre-modding my S1s had no issue staying in spec even in ~50C ambient.

OK tell me than what resistor did you use?

[MrTeal]

I do use a HP DPS-800GB which gives me a nice 69AMP Rail. I not sure if I would even mess with an 80+ Gold Power Supply those aren't build for 24/7 operations and will sooner or later fail on you. If I would remove my network Switch and maybe the low quality fan's which the Antminer has I might be able to reduce the costs a bit but I am already very happy with what I archived with those old S1's before I upgrade to some high end gear.

How many month do your S1's did ran before you downgraded them ? Mine ran almost 2+ years straight 24/7.

I'm running the majority of them on Emerson DS450 server PSUs, and have a couple plugged into a cheap 850W HEC power supply I got for $40.
http://www.tigerdirect.ca/applications/SearchTools/item-details.asp?EdpNo=7621664
I got the first one mid January, so I guess I ran them almost 9 months before I modded them at the start of September.

Don't get me wrong, I'm not trying to be antagonistic here. The reason I asked about the power supplies was that using the kind of super cheap PSU of the kind that comes free in a $30 case is about the only reason I could see getting 2.75J/GH at stock settings. Pretty much everyone with a decent PSU is measuring closer to 2J/GH.

I would double check your measurement equipment, I have a feeling something is malfunctioning or hooked up incorrectly. Your setup is likely drawing much less power than you're reporting here.

Not sure I do run this one infront of the hole setup so Switch, Outlet , PSU, Antminer all are measured otherwise you don't get a clear picture. One thing most people forget it the Cooling of the room. I am in NEW MEXICO so Desert and have my setup in the Garage. WHich itself is heated up to almost 90 degree with the Antminers now which is nice and warm :-)

Your switch shouldn't be drawing 200W, and unless that 90 is in C that shouldn't be a problem either. Even pre-modding my S1s had no issue staying in spec even in ~50C ambient.

OK tell me than what resistor did you use?

8k75 in parallel with the existing resistor.

[Lumanet]

OK so if I do your way I come down to 2.97K Ohm. Which is almost 3K Ohm. Strange I replaced it with a 4.12K Ohm and I do get ~ 2.8 K Ohm.

Thank you for your info by the way.

I do will further research and maybe it's the Voltage of the Power Supply which need's to be adjusted ? I will buy a 10K Ohm Multiturn today if possible and a 6K8 Ohm resistor as well. It maybe the Power Supply I use, since it's a Server Power Supply kicks out a little to much Power :-) Once I have this in place I update the channel on my findings.

FYI who want's and need to calculate resistors in parallel here is a helpful link :

http://www.sengpielaudio.com/calculator-paralresist.htm

[MrTeal]

OK so if I do your way I come down to 2.97K Ohm. Which is almost 3K Ohm. Strange I replaced it with a 4.12K Ohm and I do get ~ 2.8 K Ohm.

Keep in mind that when you measure across the resistor you're not really measuring just the resistor, you're measuring the whole circuit between those points.
There shouldn't be much of a difference between the two methods, I just find it's a lot quicker to tack another resistor on than to pull the old one off.

[bitrev]

I followed this tutorial and it works fine.

Thanks mate.

[Lumanet]

Quick update for everone :

Antminer S1 isn't Antminer S1 anymore. Batch from this year act differently now !!! So please check your Antminer before you modify it.
If you do have a 2014 Antminer S1 you might not get as good of a result. The older batches actually do work a lot better. Take a look onto your motherboard and you can see if it's newer or older. An other indicator is the yellow resistors in between the ASIC chips. If they are smaller you most likely have a newer model.

I noticed running my Antminer's 24/7 that the newer model stalls and starts beeping when you clock it to 287 so i downed it to 275 and guess what... it also produce more GH's too.

updates will be posted.

[Beaflag VonRathburg]

I'm at 22 undervolted S1s and I'm having some mixed results with variance as well. I haven't quite figured out if it is just a single or combination of factors; binning variance, manufacture date, capacitors (1200 or 1500), temperature, etc...

What I am sure of is that I have a few units that are very unhappy at 2.8-3.0 ohms and a frequency of 275. I'm currently letting everything run for a few more hours to get more informative results. I'm planning on stepping those units down slightly, but there's a lot of mixed posts about frequency value, chip frequency, and timeout. I've been using 0A81 275 and 45 for months, but with some of the newer ones not operating correctly I'm starting to wonder. I searched and found a massive frequency table:

Code:

chip_freq timeout freq_value(s)
  37.50 373 0583, 0b87
  39.06 358 0c07
  40.62 344 0603, 0c87
  42.19 331 0d07
  43.75 320 0683, 0d87
  45.31 308 0e07
  46.88 298 0703, 0e87
  48.44 289 0f07
  50.00 280 0783, 0f87
  51.56 271 1007
  53.12 263 0803, 1087
  54.69 256 1107
  56.25 248 0883, 1187
  57.81 242 1207
  59.38 235 0903, 1287
  60.94 229 1307
  62.50 224 0983, 1387, 4983, 5387
  64.06 218 1407, 5407
  65.62 213 0a03, 1487, 4a03, 5487
  67.19 208 1507, 5507
  68.75 203 0a83, 1587, 4a83, 5587
  70.31 199 1607, 5607
  71.88 194 0b03, 1687, 4b03, 5687
  73.44 190 1707, 5707
  75.00 186 0582, 0b83, 0b86, 1787, 4b83, 5787
  76.56 182 5807
  78.12 179 0c06, 4c03, 5887
  79.69 175 5907
  81.25 172 0602, 0c86, 4c83, 5987
  82.81 169 5a07
  84.38 165 0d06, 4d03, 5a87
  85.94 162 5b07
  87.50 160 0682, 0d86, 4d83, 5b87
  89.06 157 5c07
  90.62 154 0e06, 4e03, 5c87
  92.19 151 5d07
  93.75 149 0702, 0e86, 4e83, 5d87
  95.31 146 5e07
  96.88 144 0f06, 4f03, 5e87
  98.44 142 5f07
 100.00 140 0782, 0f86, 4f83, 5f87
 101.56 137 6007
 103.12 135 1006, 5003, 6087
 104.69 133 6107
 106.25 131 0802, 1086, 5083, 6187
 107.81 129 6207
 109.38 128 1106, 5103, 6287
 110.94 126 6307
 112.50 124 0882, 1186, 5183, 6387
 114.06 122 6407
 115.62 121 1206, 5203, 6487
 117.19 119 6507
 118.75 117 0902, 1286, 5283, 6587
 120.31 116 6607
 121.88 114 1306, 5303, 6687
 123.44 113 6707
 125.00 112 0982, 1386, 4982, 5383, 5386, 6787
 128.12 109 1406, 5406
 131.25 106 0a02, 1486, 4a02, 5486
 134.38 104 1506, 5506
 137.50 101 0a82, 1586, 4a82, 5586
 140.62  99 1606, 5606
 143.75  97 0b02, 1686, 4b02, 5686
 146.88  95 1706, 5706
 150.00  93 0581, 0b82, 0b85, 1786, 4b82, 5786
 153.12  91 5806
 156.25  89 0c05, 4c02, 5886
 159.38  87 5906
 162.50  86 0601, 0c85, 4c82, 5986
 165.62  84 5a06
 168.75  82 0d05, 4d02, 5a86
 171.88  81 5b06
 175.00  80 0681, 0d85, 4d82, 5b86
 178.12  78 5c06
 181.25  77 0e05, 4e02, 5c86
 184.38  75 5d06
 187.50  74 0701, 0e85, 4e82, 5d86
 190.62  73 5e06
 193.75  72 0f05, 4f02, 5e86
 196.88  71 5f06
 200.00  70 0781, 0f85, 4f82, 5f86
 203.12  68 6006
 206.25  67 1005, 5002, 6086
 209.38  66 6106
 212.50  65 0801, 1085, 5082, 6186
 215.62  64 6206
 218.75  64 1105, 5102, 6286
 221.88  63 6306
 225.00  62 0881, 1185, 5182, 6386
 228.12  61 6406
 231.25  60 1205, 5202, 6486
 234.38  59 6506
 237.50  58 0901, 1285, 5282, 6586
 240.62  58 6606
 243.75  57 1305, 5302, 6686
 246.88  56 6706
 250.00  56 0981, 1385, 4981, 5382, 5385, 6786
 256.25  54 1405, 5405
 262.50  53 0a01, 1485, 4a01, 5485
 268.75  52 1505, 5505
 275.00  50 0a81, 1585, 4a81, 5585
 281.25  49 1605, 5605
 287.50  48 0b01, 1685, 4b01, 5685
 293.75  47 1705, 5705
 300.00  46 0580, 0b81, 0b84, 1785, 4b81, 5785
 306.25  45 5805
 312.50  44 0c04, 4c01, 5885
 318.75  43 5905
 325.00  43 0600, 0c84, 4c81, 5985
 331.25  42 5a05
 337.50  41 0d04, 4d01, 5a85
 343.75  40 5b05
 350.00  40 0680, 0d84, 4d81, 5b85
 356.25  39 5c05
 362.50  38 0e04, 4e01, 5c85
 368.75  37 5d05
 375.00  37 0700, 0e84, 4e81, 5d85
 381.25  36 5e05
 387.50  36 0f04, 4f01, 5e85
 393.75  35 5f05
 400.00  35 0780, 0f84, 4f81, 5f85
 406.25  34 6005
 412.50  33 1004, 5001, 6085
 418.75  33 6105
 425.00  32 0800, 1084, 5081, 6185
 431.25  32 6205
 437.50  32 1104, 5101, 6285
 443.75  31 6305
 450.00  31 0880, 1184, 5181, 6385

If you look at the values equal to or less than 275 you'll notice the timeout values are considerably different than the OPs, 45. The default asic-freq vi table also has frequency 250 at a timeout value of 55 vs this table at 56.

TLDR: What timeout values are people using for frequencies of 275 and below?

[Beaflag VonRathburg]

Update:

Freq Value 1505
Freq 268
Timeout 52

I just switch 9 miners over to 268 and it the values I posted from the table above work. The next step on the table is 262.5, but I have a few that don't really need to be stepped down that far. Anyone happen to have or are able to determine a frequency value and timeout number for a frequency of 265?

[aurel57]

Just wondering if anybody has this worked out for the S3 yet?

[voffka05]

Just wondering if anybody has this worked out for the S3 yet?

https://bitcointalk.org/index.php?topic=771979.0

[pallol]

Update:

Freq Value 1505
Freq 268
Timeout 52

I just switch 9 miners over to 268 and it the values I posted from the table above work. The next step on the table is 262.5, but I have a few that don't really need to be stepped down that far. Anyone happen to have or are able to determine a frequency value and timeout number for a frequency of 265?

what is your HW rate like at this frequency? I had mine sitting at 275 with a HW around 4%. I could not reduce this even though I walked the timeout from 40 through to 55, so now mine is sitting at the default 250, with an almost 0% HW (one machine has been 0% for the last three days) I want to get as close to 275 as possible and I know some of the defaults just do not work (300 for example give me a very very low hash rate and high HW)

[warrensgun]

I think I've been using either of these settings

                                           
       option 'freq_value'    '0A81'  #275M
       option 'chip_freq'     '275'   
       option 'timeout'       '45'
                                               
       option 'freq_value'    '0a01' #267M
       option 'chip_freq'     '267'
       option 'timeout'        '53'
                                                                   
           

[Beaflag VonRathburg]

Update:

Freq Value 1505
Freq 268
Timeout 52

I just switch 9 miners over to 268 and it the values I posted from the table above work. The next step on the table is 262.5, but I have a few that don't really need to be stepped down that far. Anyone happen to have or are able to determine a frequency value and timeout number for a frequency of 265?

what is your HW rate like at this frequency? I had mine sitting at 275 with a HW around 4%. I could not reduce this even though I walked the timeout from 40 through to 55, so now mine is sitting at the default 250, with an almost 0% HW (one machine has been 0% for the last three days) I want to get as close to 275 as possible and I know some of the defaults just do not work (300 for example give me a very very low hash rate and high HW)

Recheck the resistance. It is a giant pain in the ass, but you need to adjust it every once in a while. As the units heat up the graphite bakes in and increases the resistance and lowers the voltage. If you keep them in a warm environment or stacked against each other it accelerates this. I just had to readjust 16 of them as they were getting up in the 2-6% range.

Here's all of them after I readjusted them. I need to redo 24 + 27 as that was my first run through with those:

#-Frequency-HW Error Percentage

1 275 = 0.29526542916726889801986331068943
2 275 = 0.62115496524246610234090456970006
3 275 = 0.23817991100379235015136527593905
4 262 = 1.420715117786610214752061797818
5 268 = 0.17169630125425603119634490585804
6 275 = 1.20790671325427633955219624587
7 275 = 0.18331280114123078473516947311905
8 268 = 0.36165085608083697405127173722769
9 275 = 0.21186440677966101694915254237288
10 268 = 0.70918402741122569319816735687377
11 268 = 0.49968110951768275373912915248864
12 275 = 0.34170646345878661448545764874231
13 275 = 1.9673807729572612767789858972979
14 275 = 0.47481498730798894795862586065377
15 275 = 0.6791942830244073656741999662125
16 268 = 0.53198186854696549987408713170486
17 275 = 0.43863837101047957795230479100523
18 275 = 0.41021045428810598087537035300329
19 275 = 0.67922119374401532077880625598468
20 275 = 0.29032854982484574301775789789368
21 262 = 0.01070080052179141591973380484797
22 268 = 0.12752589397725269826129329918819
23 268 = 0.29771095000081261053350097734885
24 275 = 3.4735131299000746191284894664367
25 275 = 0.85411213665794186527069844331175
26 275 = 0.29528606269128995930662355517787
27 275 = 3.0170086204243341624534494288025
UP1 275 = 0.37350073902379235969789229544126
UP2 275 = 0.84665203041665056262135282470168
UP3 275 = 0.17782269147173546733228135310189

[pallol]

Recheck the resistance. It is a giant pain in the ass, but you need to adjust it every once in a while.

yep, that did it, back into a good range now - thx

[klondike_bar]

Recheck the resistance. It is a giant pain in the ass, but you need to adjust it every once in a while.

yep, that did it, back into a good range now - thx

I find it takes about 3 days to fully settle as the graphite oxidizes/laminates/whatever chemistry occurs. Generally the resistance will drift downwards  by a few percent

[spazzdla]

I think I've been using either of these settings

                                           
       option 'freq_value'    '0A81'  #275M
       option 'chip_freq'     '275'   
       option 'timeout'       '45'
                                               
       option 'freq_value'    '0a01' #267M
       option 'chip_freq'     '267'
       option 'timeout'        '53'
                                                                   
           

Can you set this from the web based application?

[terrapinflyer]

I think I may be missing something here I have not been able to reduce the power draw after completing the pencil mod.

My resistors have been modded to:

Code:

R3 2.96|3.0
R66 2.90|3.02
R38 2.87|2.87
R52 3.05|2.80

.86v

Code:

        option 'freq_value'     '0A81'  #275M
        option 'chip_freq'      '275'        
        option 'timeout'        '45' 

Hashing at 140gh~ with .5% error rate but still pulling 202w from the wall. 

[MrTeal]

I think I may be missing something here I have not been able to reduce the power draw after completing the pencil mod.

My resistors have been modded to:

Code:

R3 2.96|3.0
R66 2.90|3.02
R38 2.87|2.87
R52 3.05|2.80

.86v

Code:

        option 'freq_value'     '0A81'  #275M
        option 'chip_freq'      '275'        
        option 'timeout'        '45' 

Hashing at 140gh~ with .5% error rate but still pulling 202w from the wall. 

That is reduced power draw. You'd be pulling ~280W at the original voltage for 140GH/s.

[terrapinflyer]

I think I may be missing something here I have not been able to reduce the power draw after completing the pencil mod.

My resistors have been modded to:

Code:

R3 2.96|3.0
R66 2.90|3.02
R38 2.87|2.87
R52 3.05|2.80

.86v

Code:

        option 'freq_value'     '0A81'  #275M
        option 'chip_freq'      '275'        
        option 'timeout'        '45' 

Hashing at 140gh~ with .5% error rate but still pulling 202w from the wall.  

That is reduced power draw. You'd be pulling ~280W at the original voltage for 140GH/s.

Doh!  

What should I do to get it closer to 150w? Keeps lowering the resistance?

[MrTeal]

I think I may be missing something here I have not been able to reduce the power draw after completing the pencil mod.

My resistors have been modded to:

Code:

R3 2.96|3.0
R66 2.90|3.02
R38 2.87|2.87
R52 3.05|2.80

.86v

Code:

        option 'freq_value'     '0A81'  #275M
        option 'chip_freq'      '275'        
        option 'timeout'        '45' 

Hashing at 140gh~ with .5% error rate but still pulling 202w from the wall.  

That is reduced power draw. You'd be pulling ~280W at the original voltage for 140GH/s.

Doh!  

What should I do to get it closer to 150w? Keeps lowering the resistance?

I doubt you will get to 150W @ 140GH/s, but ~180W should be possible. Just keep lowering it.