Topic: Bitmain AntMiner U1 Tips & Tricks - page 15. (Read 106767 times)

Okay, am I a complete idiot?

cgminer: opens instantly. I still have to plug them in one at a time, but it works.

bfgminer: Nothing. No devices found. It opens, I hit "M+", then try "all", nothing, "auto", nothing. The Ants are recognized by the computer, but bfgminer simply doesn't see them. Clock speeds don't seem to matter, it still won't find them. Windows 8.1 x64 machine.

I'm using:


What in the hell am I missing?

Thanks, anyone.

Q1 - Cheers, also want to know about adding pools to run as a backup incase one goes down (currently cant connect to BTC guild, this is my current setup, what would i change

cgminer --bmsc-options 115200:20 --bmsc-freq 0981 -o eu-stratum.btcguild.com:3333 -u myusername -p mypassword

Q2 - When CGminer shows up your Miners ie AMU 0, AMU 1, is there anyway having your miners working on different pools

You'll need to solder different resistors on t give the chip the voltage it needs to hit 4GHs.

on the OC guide on the PDF Guide says the frequency hex is 5381, does that mean it would go to 4ghs, or is there soldering involved to get that done?

Hi again,

ok, in Lab I have a regulatet PSU but that should be last resort to fiddle around in the circuit. I'll try to find out what U5 is. With 6 pins it seems to be an adjustable VR...but messing around on this SMD board...without proper tools...this will have to wait till tuesday.

Thanks,

           miniasic.

You can try to provide PLL voltage externally....
Dont you have regulated power supply?=)

Or maybe you can create 1,1 V from your 5V power supply by  building  voltage divider;) (If  so, then use high values to avoid  high current loss)

Hi,

I'd use 0A01 for 2.1GHs.

To chromsoma: Interesting, your PLL-Voltage is out of Specs and about 10% higher than mine. Maybe that's my problem...I'd guess
the to unknown ICs are VR for 5V -> 3.3V and one VR for PLL Voltage. Unfortunately I couldn't read the exact type and for the S1 boards bitmaint seems to have used other types.

Greetings,

            miniasic.

If i wanted to OC to 2.1gh what would i put the value at?

hi, i measured the voltages on violet dot on my Hardware hacked devices (even though i dont think replacing resistors will change PLL voltage)

The voltage is about 1.15-1.19 V

Hi again,

meanwhile I found some time to improvise a powered USB connector.

http://i.imgur.com/NhqTrX2.jpg

The plug is now powered by an external 5V 2.5A supply. USB signals D+, D- and Ground are connected straight through, power is supplied by external PSU and is routet over the blue jumper so I can measure current in case I want to know it. What I now have is:

USB Voltage 5.03V (while hashing)
USB-Serial converter 3.28V
PLL/Com: 0.998 V
ASIC: 0.83V

What I'd like to know is the voltage close to U5 (violet point) on the picture in below URL

https://bitcointalksearch.org/topic/m.4681379

I think this is the PLL Voltage as the ASIC has different Voltages for Clock and Operation.
Operations Voltage seems ok but I'm not sure about PLL Voltage. Maybe someone can measure his Ant
for PLL voltage. Due to specs Core is 0.75 - 1.1V, IO is 3.0 - 3.6V and PLL 0.9 - 1.1. So my values should be ok, main USB-Voltage is ok, current of 2.5A should suffice even at much higher clock rates but my ant is still a better block erupter 

Greetings,

               miniasic.

My config gives me: 7 min data, I am sure I will not get any HW errors @ X0981.

color 06
bfgminer -S antminer:all --set-device antminer:clock=x0981 --no-submit-stale

 bfgminer version 3.10.0 - Started: [2014-01-24 01:19:09] - [  0 days 00:07:13]
 [M]anage devices [P]ool management ettings [D]isplay options                     [H]elp [Q]uit
 Connected to us1.eclipsemc.com diff 26 with stratum as user screaming_trees_2
 Block: ...258f0087 #282194  Diff:1.79G (12.81Ph/s)  Started: [01:19:09]
 ST:2  F:1  NB:1  AS:0  BW:[ 20/  0kB/s]  E:0.05  I:25.63uBTC/hr  BS:10.2k
 1            |  2.00/ 1.89/ 1.99Gh/s | A:17 R:0+0(none) HW:0/none
---------------------------------------------------------------------------------------------------
 AMU 0:       |  2.06/ 1.90/ 1.98Gh/s | A:17 R:0+0(none) HW:0/none
---------------------------------------------------------------------------------------------------

 [2014-01-24 01:24:44] Stratum from pool 2 requested work update
 [2014-01-24 01:24:44] Pool 2 is hiding block contents from us
 [2014-01-24 01:24:45] Pool 1 is hiding block contents from us
 [2014-01-24 01:24:46] Pool 0 is hiding block contents from us
 [2014-01-24 01:24:55] Pool 2 now providing block contents to us
 [2014-01-24 01:25:05] Accepted 038c8834 AMU 0  pool 2 Diff 72/26
 [2014-01-24 01:25:39] Stratum from pool 2 requested work update
 [2014-01-24 01:25:39] Pool 2 is hiding block contents from us
 [2014-01-24 01:25:51] Pool 2 now providing block contents to us

The capacitor is soldered between GND and +5V

I`ve tweaked & match my four Ant`s

Ant1 -> 393MHz@1V stable
Ant2 -> 393MHz@1V stable
Ant3 -> 393MHz@1,05V stable
Ant4 -> 393MHz@1,1V stable with HW/F <0,1% <- lazy ant...

After ~5h ...



Infrastructure...

These help on the back of the chips. Also where did you solder the cap? It's hard to make out in the image.

http://www.frozencpu.com/products/14952/vid-178/ModMyToys_Solid_Copper_Passive_Micro-Heatsink_-_8mm_x_8mm_x_8mm_Pack_of_4_-_Nickel_Plated-MMT-HS-8X8-CU-NK.html

Hi,   a few more news from the front:)

 I measured the input voltage  right from the  usb port on the Antminer (Black and Red dots according to Walking Glitch's picture).
First i noticed that there is  massive voltage drop of about 0,3 V. SO my miners getting about 4.9-4.8 V instead of 5,1 or so.
That was my problem, i am using passive cooling, and    need  USB extension cables. I used cheap  and long ones (3 Dollar, 2.5m). Then i measured the resistance of the cable: 100-150 Ohm.  Chinese crap:)
Now i bought  quality ones, and   getting  5,1 V before Antminer  starts mining.

 Second thing I noticed, the miner gives massive ripple to the  usb ports (usb Hub or in my case PCI-USB 3.0 Card).
If the miner 1 suddenly  increase current consumption you will notice voltage drop on  other   miners if they are sharing same power supply.
 For example if on miner1 the  input voltage  peaks to 5,2 (and so the hashrate),  the miner2 will experience voltage drop  to 4,8 V etc.
That is why you can see this bouncing effect in you mining software.: some miner getting faster while others are getting slower  at the same rate.
To prevent this i added additional Tantalum capacitor of 470 µF on each miner (see Picture).  As the result, the miners stabilize MUCH faster, HW drops even more.

Next  i once again tried  Hardware hack:
10K+3,7K gave me stable 2,6 GHS. But saddly enough i could only run one Miner on my PC. If i switch another one,  they both stop mining... (not sure why, since my PSU can provide 25 Amp on 5V rail, perhaps  too high ripples/interference)

After using 2,2K instead of 3,7K I managed to get stable 2,2 -2,4 GHS with two devices at the same time. Since i only can cool two miners right now, i dont know if  running 4 Miner on this mode will be stable. Right now they are running at nominal 2GHS, while hacked ones perform  5-15% better.
-------------------------------------------------------------------------------------------------------------
One more thing: I noticed if  input current  falls under 4,7 V the miner stops mining and  switches off.
Guys, who cant run their miner  past few hundred MHS  should measure if there is any voltage drop on the input....



PS
Still no idea how to bring  antminer to advertised 4GHS....
Maybe some  separate  power supply, not just regular USB3.0 Port with maximum of 900 mA. For example connect GND and +5V from other source, and D+ and D- to USB ports.

Hmm, one of my units quit hashing last night, and I can't get windoze to detect it. -_-


EDIT: Seems my USB port is dead...

Ah...!!!
For now, I'm a little afraid to modify my Ants.
Some basic soldering skills, no SMT skills at all...

Thank you, Janezki and miniasic!

Ok, more of this:

300.00 MHz | 2400 MHs | 1785 | 017 ms
306.25 MHz | 2450 MHs | 1805 | 017 ms
312.50 MHz | 2500 MHs | 1885 | 017 ms
318.75 MHz | 2550 MHs | 1905 | 016 ms
325.00 MHz | 2600 MHs | 1985 | 016 ms
331.25 MHz | 2650 MHs | 1A05 | 016 ms
337.50 MHz | 2700 MHs | 1A85 | 015 ms
343.75 MHz | 2750 MHs | 1B05 | 015 ms
350.00 MHz | 2800 MHs | 1B85 | 015 ms
356.25 MHz | 2850 MHs | 1C05 | 015 ms
362.50 MHz | 2900 MHs | 1C85 | 014 ms
368.75 MHz | 2950 MHs | 1D05 | 014 ms
375.00 MHz | 3000 MHs | 1D85 | 014 ms


or even smaller steps:

300.00 MHz | 2400 MHs | 2F86 | 017 ms
303.12 MHz | 2425 MHs | 3006 | 017 ms
306.25 MHz | 2450 MHs | 3086 | 017 ms
309.38 MHz | 2475 MHs | 3106 | 017 ms
312.50 MHz | 2500 MHs | 3186 | 017 ms
315.62 MHz | 2525 MHs | 3206 | 017 ms
318.75 MHz | 2550 MHs | 3286 | 016 ms
321.88 MHz | 2575 MHs | 3306 | 016 ms
325.00 MHz | 2600 MHs | 3386 | 016 ms
328.12 MHz | 2625 MHs | 3406 | 016 ms
331.25 MHz | 2650 MHs | 3486 | 016 ms
334.38 MHz | 2675 MHs | 3506 | 016 ms
337.50 MHz | 2700 MHs | 3586 | 015 ms
340.62 MHz | 2725 MHs | 3606 | 015 ms
343.75 MHz | 2750 MHs | 3686 | 015 ms
346.88 MHz | 2775 MHs | 3706 | 015 ms
350.00 MHz | 2800 MHs | 3786 | 015 ms


The timeout value is for the original cgminer-bitmaint version where you had to provision a timeout-value (default 20 * 100 ms). The slower the device is, the longer it takes to calculate the hash. bfgminer uses 1.5sec timeout for all values. cgminer-bitmaint uses the deployed value. After value * 100ms it times out the calulation an no nonce was found. You can see this for bfg and cg if you run it in debug mode.

Greetings,

            miniasic.

Definetely not stock - I changed the resistors to 10kOhm & 3,9kOhm and stuck some heatsinks into it.
Only one Ant in use at the moment, but ordered second one just yesterday - plan to do the same mod...