[GUIDE] Undervolt antminer s1 [1.19W/GH at the wall] - page 16.

elaramus

full member

Activity: 219

Merit: 100

Quote from: Beaflag VonRathburg on May 18, 2014, 11:15:10 PM

Finished my other 3 antminers. I'm guessing the results on the first one were a bit of an oddity, but I'll try and work on that one a little later. More results:
Ant4

Code:

Ant4
R3  Resistance 3.92 1.011v top / 1.004v bottom
R66 Resistance 3.85 1.007v top / 1.000v bottom
R38 Resistance 3.89 1.006v top / 0.999v bottom
R52 Resistance 3.85 0.982v top / 0.976v bottom

R3  Resistance 3.91 1.000v top / 0.992v bottom
R66 Resistance 3.84 0.991v top / 0.984v bottom
R38 Resistance 3.98 1.020v top / 1.012v bottom
R52 Resistance 3.85 0.990v top / 0.984v bottom

Frequency = 350
Power Consumption = 293 watts
Roughly =  171.05 GH
15 Minute Error Rate = 3.4754098360655737704918032786885

-

Frequency = 325
Power Consumption = 273 watts
Roughly = 168.46 GH
45 Minute Error Rate = 0.21840130246594925147917245761029

I'd still like to see a 200-300 Frequency table with some different numbers in there. My first antminer runs with a super low HW error rate at 250, but horrible at 275, and I think I could bump it up a little bit.

Beaflag VonRathburg, excellent results and information. Instead of using the pencil-mod, I've chosen to add a 1/8w film resistor in parallel with the ones listed above. For instance adding a 10K ohm resistor creates a total resistance of 3.08k which drives the chips at .89v. But I too am seeing sporadic results with the hardware errors. I've ordered a number of other value film resistors from Mouser to play with this week (12K,14K,15K,16K,17K,18K,20K and 22K). I think the results you've posted for Antminer #4 are promising and that the chips enjoy being fed at least 1v.

The point of all this blather is I was wondering if you could run ANT 4 specifically at 275 and 250MHZ and post the results?

Code:


#option 'freq_value'	'0a81'	#275
        #option 'chip_freq'	'275'
        #option 'timeout'	'51'
        
        #option 'freq_value'    '0a01' #267M
        #option 'chip_freq'     '267'
        #option 'timeout'	'53'
        
        #option 'freq_value'    '0981'  #250M
        #option 'chip_freq'      '250'        
        #option 'timeout'         '56'         
                                             
        #option 'freq_value'    '0901' #237M
        #option 'freq_value'    '237'
        #option 'timeout'	 '59'
        
        
        #option 'freq_value'    '0881'  #225M
        #option 'chip_freq'     '225'
        #option 'timeout'       '63'

Brassguy

full member

Activity: 139

Merit: 100

I'm curious Beaflag... which of your ants have the 1200uf capacitors and which have the 1500uf caps?

Beaflag VonRathburg

sr. member

Activity: 472

Merit: 250

Finished my other 3 antminers. I'm guessing the results on the first one were a bit of an oddity, but I'll try and work on that one a little later. More results:

Ant2

Code:

Ant2
R3  Resistance 2.911 kOhms .876v top / .875v bottom
R66 Resistance 2.949 kOhms .879v top / .877v bottom
R38 Resistance 2.930 kOhms .863v top / .857v bottom
R52 Resistance 2.966 kOhms .870v top / .866v bottom

R3  Resistance 2.915 .858v top / .853v bottom
R66 Resistance 2.909 .856v top / .850v bottom
R38 Resistance 2.908 .858v top / .853v bottom
R52 Resistance 2.928 .867v top / .865v bottom

Frequency = 275
Power Consumption = 175 watts
Roughly = 139.90 GH
30 Minute Error Rate = 0.70297018742181532364304613034285

Ant3

Code:

Ant3 (Repaired Capacitor)
R3  Resistance 2.933 .869v top / .869v bottom
R66 Resistance 2.897 .860v top / .854v bottom
R38 Resistance 2.875 .853v top / .847v bottom
R52 Resistance 2.938 .871v top / .870v bottom

R3  Resistance 2.976 .861v top / .856v bottom
R66 Resistance 2.856 .851v top / .846v bottom
R38 Resistance 2.960 .872v top / .869v bottom
R52 Resistance 2.909 .871v top / .869v bottom

Frequency = 275
Power Consumption = 174 watts
Roughly =  140.12 GH
30 Minute Error Rate = 0.75778838057816446814482177939938
5.75 Hour Error Rare = 1.0067589922956940965165193552961

Ant4

Code:

Ant4
R3  Resistance 3.92 1.011v top / 1.004v bottom
R66 Resistance 3.85 1.007v top / 1.000v bottom
R38 Resistance 3.89 1.006v top / 0.999v bottom
R52 Resistance 3.85 0.982v top / 0.976v bottom

R3  Resistance 3.91 1.000v top / 0.992v bottom
R66 Resistance 3.84 0.991v top / 0.984v bottom
R38 Resistance 3.98 1.020v top / 1.012v bottom
R52 Resistance 3.85 0.990v top / 0.984v bottom

Frequency = 350
Power Consumption = 293 watts
Roughly =  171.05 GH
15 Minute Error Rate = 3.4754098360655737704918032786885

-

Frequency = 325
Power Consumption = 273 watts
Roughly = 168.46 GH
45 Minute Error Rate = 0.21840130246594925147917245761029

I'd still like to see a 200-300 Frequency table with some different numbers in there. My first antminer runs with a super low HW error rate at 250, but horrible at 275, and I think I could bump it up a little bit.

Beaflag VonRathburg

sr. member

Activity: 472

Merit: 250

I've been working on this right now as I'm completely out of electricity in my house right now. It's either turn things off or undervolt these guys so the option was kind of easy. Here are the results I've achieved so far with one out of four antminers. I'll keep this updated as I get the others setup as well.

Code:

Completely Stock:

Frequency = 350
Power Consumption = 364 watts
Roughly = 178.60 GH
30 Minute Error Rate = 0.22854691951839710793174228730737

---------------------------------------------------------

Modified:

Network Side (Measured closest to the fan moving towards the back)
Resistance 2.850 kOhms .864v top / .864v bottom
Resistance 2.903 kOhms .869v top / .867v bottom
Resistance 2.904 kOhms .871v top / .869v bottom
Resistance 2.941 kOhms .872v top / .870v bottom

Non Network Side (Measured closest to the fan moving towards the back)
Resistance 2.915 kOhms .868v top / .867v bottom
Resistance 2.933 kOhms .876v top / .875v bottom
Resistance 2.915 kOhms .869v top / .868v bottom
Resistance 2.894 kOhms .872v top / .870v bottom

Frequency = 275
Power Consumption = 184 Watts
Roughly = 115.21 GH with erros
30 Minute Error Rate = 17.142677798003891783315548302053

-

Frequency = 250
Power Consumption = 162 watts
Roughly = 128.60 GH
30 Minute Error Rate = 0.11227745005435654328028372333411

---------------------------------------------------------

As can be seen I was shooting for the OP's results out of this, but 275 gave massive amounts of HW errors. I was looking at the frequency formula and was having issues working it out. Could someone come up with a frequency table between 250 and 300 in small steps?

Something along the lines of this, but for undervolted results would be excellent:

5e82 = 387.50 MHz
5f02 = 393.75 MHz
5f82 = 400.00 MHz
6002 = 406.25 MHz
6082 = 412.50 MHz

lucazane

legendary

Activity: 1198

Merit: 1000

Quote from: tzortz on May 14, 2014, 08:21:29 AM

Thanks!

HW errors???
Frequency?

HW = 1.35% ,1.12% & 0.22%
Freq :

option 'freq_value' '0881'
option 'chip_freq' '225'
option 'timeout' '45'

tzortz

hero member

Activity: 728

Merit: 500

Thanks!

HW errors???
Frequency?

lucazane

legendary

Activity: 1198

Merit: 1000

Quote from: tzortz on May 14, 2014, 04:02:57 AM

Anyone tried results bellow 1.19W/ghs undevolting?

Yep

345 watts for 3 antminers (113GH + 112GH +106GH)

= 1.04 W/GH

chip @~0.78

shamacala

newbie

Activity: 32

Merit: 0

Thanks think thats what the problem may be, was doing it on top as i had read this somewhere else.

tzortz

hero member

Activity: 728

Merit: 500

Anyone tried results bellow 1.19W/ghs undevolting?

Raptor2213

full member

Activity: 210

Merit: 100

Quote from: shamacala on May 14, 2014, 03:18:32 AM

Thanks, watched the video, is it a case of over the top of the resistor or around the side, also how many pencil strokes on all resistors.

On the edges, just like the video shows.
As many as it takes to get the right resistance (use a meter).

shamacala

newbie

Activity: 32

Merit: 0

Thanks, watched the video, is it a case of over the top of the resistor or around the side, also how many pencil strokes on all resistors.

Raptor2213

full member

Activity: 210

Merit: 100

Watch this video and make certain that you are applying the pencil mod similar to how it is depicted in the video.
http://youtu.be/LNXSHpPACGs

I've modded several of them so far with no issues. Likely you're modding the wrong resistor, or applying the pencil mod wrong. Remember, you need to mod 4 resistors per blade (total of Cool

.

ernie-

sr. member

Activity: 252

Merit: 250

You have to rub the pencil across the resistor several times, make sure the pencil is sharp, and blow away any excess before you measure the resistance. Don't press real hard and dislodge the resistor.

bjalbert

full member

Activity: 166

Merit: 100

Quote from: shamacala on May 13, 2014, 03:50:45 PM

Hi Guys was wondering if someone could tell me where im going wrong? tried to undervolt all r3 resistors tonight but made no difference to ohms or volts, perhaps could i be using the wrong type of pecil (4hb i was using)

thanks

4HB should work fine. What exactly did you do and how many times did you trace it?

shamacala

newbie

Activity: 32

Merit: 0

Hi Guys was wondering if someone could tell me where im going wrong? tried to undervolt all r3 resistors tonight but made no difference to ohms or volts, perhaps could i be using the wrong type of pecil (4hb i was using)

thanks

klondike_bar

legendary

Activity: 2114

Merit: 1005

ASIC Wannabe

i might be undervoltng sooner than later - the results people are seeing is impressive. 75% speeds for 60% wattage

just holding out until for a few more weeks while the difficulty is low (11% jump = party)

povlhp

newbie

Activity: 35

Merit: 0

I have 1.5 Antminers.
One that always performced fine, and one that always performed badly, with the 2nd board shipped back after dying, and now supposedly lost in mail. The bad performing board is april 2014 issue.

Got all 3 board modded to < 3kOhm, and Now I am running around 248 Watts for 210 GH/s (GHash.io average, 1 day). The single board is doing 65 GH/s, the fullminer is doing 145 GH/s. Both do have some HW errors, < 1%.

So I am at < 1.2W/GHS. So even the relatively high power costs here allows me to make a profit for a few more weeks/months.

drukoz

member

Activity: 67

Merit: 10

the 2 @ 140 gh/s are both at 3.0- 2.9 Kohm @ 275 freq
the other 2 @ 153-160 gh/s are both at 3.6-3.7 kohm @ 300 freq

they're both using 750 watt PSU 1 @ 160ghs and 1 at 140ghs each psu

they take 2.5 amps - 2.7 amps each PSU got 2 of them around 6 amps for 600 gh/s

i was thinking about buying more ants and a better psu i dont like the PSUs that i have anyways

s1gs3gv

legendary

Activity: 1316

Merit: 1014

ex uno plures

Is R3 an 0805 sized component ?

ie: 0805 (2012 metric): 2.0 mm × 1.25 mm (0.079 in × 0.049 in)

also see: http://en.wikipedia.org/wiki/Surface-mount_technology#Packages

hashfun

member

Activity: 117

Merit: 10

Quote from: helipotte on May 11, 2014, 11:13:52 AM

Quote from: hashfun on May 11, 2014, 10:29:20 AM

Quote from: helipotte on May 10, 2014, 10:05:34 PM

All of my Ants are operating at 300Mhz and using about 200-220W at the wall. I like only needing one CX750M for two ants!

is this without hardware modification?
what Avg speed u got @ 300Mhz

This is achieved by "pencil modding" one of the feedback resistors and indicated in the beginning of this thread. I modded them from 4.7K to about 3.3K. This produced much

lower power usage with acceptable errors (<2%). They average about 153Gh. Temps are MUCH more manageable now too.

Thank You For share your states..

but i dnt know how to mod so i test some without hardware modification and result are here

300Mhz = 1.2 AMP (270W) 0HW Error = 153 Gh
325Mhz = 1.3 AMP (292.5W) 0HW Error = 163 Gh
350Mhz = 1.4 AMP (315W) 0HW Error = 178 Gh
400Mhz = 1.7 AMP (382.5W) 0HW Error = 203 Gh

BR

Topic: [GUIDE] Undervolt antminer s1 [1.19W/GH at the wall] - page 16. (Read 69365 times)