Topic: ANTMINER U3 Discussion and Support Thread - page 19. (Read 149480 times)

You should download the latest cgminer 4.9.2 instead of using bitmain's old fork with the bmsc junk. https://github.com/ckolivas/cgminer

It was updated to "work better" with the U3, it is much improved over bitmain's old copy.

See the readme.asic for new u3 commands instead of the bmsc.

You should not need two passwords.   I  would bring up command window and go to folder then run, cgminer.exe --ndevs  , I believe that should show all miners hooked up so easy way to see the 2 U3's are both connected right.

If you see both with ndevs then next remove all the bmsc codes and run something like:

See if that starts running them both, then add back variables and see what happens.    I have been spoiled by my modified R1 from crazyguy where it's point/click.

Hi All, 

I've been reading through this long thread and haven't been able to find a clear answer.

I have a U3 working fine on cgminer using Windows 10.  I would like to have it work with two U3's at once.  Here is my current batch file, could someone please tell me what I need to modify to get it to work with two U3's?


Would really appreciate any help provide.

Thanks!

It is pretty amazing firmware I know I might mention it a lot in this thread.  It is because of how easy it made my life compared to when I was unplugging 3 U3's multiple times a day.

I was gone for 6 day's or so and it was still working.  The R1 seemed a tad slow after the time running all that time but a reboot fixed that.  Did not touch the U3 wires

The thing with the U3 is, what works for you will almost certainly not work for everyone else.  If you found something that works, that's great, and people should at least try others' suggestions, but no one "must" do anything you did because chances are it won't make a difference to them.

R1 as shipped controls a U3, no PC needed.  Then with Crazyguy's firmware as detailed above it restarts the U3 if it Zombies out. 

Rich

Ok, i have the point now ... U3 is build like the S1, you MUST cool the bottom of the device (blue alloy cover) because only 30% of the heat is exaust from the top dissipator with fan.

2nd point, this device is a strange ... chip.
You MUST kill CGminer every 4 hours to stay at 60GH/s in a normal environment (22°C).

any supply problem (cut from main line for exemple) ... and the chip automaticly decrease from 60GH/s to 42GH/s.

250MHz at 830mV (1 000 HW error with 600 000 accepted shares).

ok, Ill simplify because there is some confusion.  And part of it was my mistake.... For that I am totally sorry.  Reading labels through jumbles of wires can be annoying.

I am running the U3 on the same circuit as my pc that draws a steady 650W; and a 110W PC power supply powering a Gridseed Blade; and the U3; all on one 15A 110V circuit.   I have my Antminer S3 on a separate circuit.

The specifics on my power supply that I am running the U3 on:

110V 2.5A
12V 7.5A

So... technically.  It does now make sense why it is working fine.   And also explains my 5A power supply causing very strange outputs/readings.

I am 100% certain the issue is with the power circuitry on the U3.   More filtering and ripple control is my first guess... because its the most logical.

The fact that I used regular dialectic grease on the heatsync shouldn't be much of a factor IMHO.

Mine zombied out over the night.


So whats the deal with the R1?  are you guys using it as a miner controller?  Is it that you reprogram the firmware to recognize and mine off a U3 along with the router's built in ASIC?

I think he looked at the 2.5A max amps for the input voltage.

Something that I never really paid attention to until after the power noise came up and looking again to see the 100-240 50/60Hz 2.5A MAX for the input. That brings up a few questions.

Aren't most wall sockets 110V @ 15A (or 20A)?

So is that extra 12.5A on the 110V causing all the trouble? And how would someone go about regulating 110V @ 15A or 20A, down to 110V @ 2.5A if the extra amps are the problem??

Exactly that what confuses me. Even if his 2.5A power supply has 100% efficiency, it would be quite impossible to power up Antminer U3. According to BM1382_Datasheet_v3.0.pdf, one BM1382 chip typically requires 10.40A at VDD of 0.72V or 0.62A at 12V. So 4 BM1382 chips on Antminer U3 requires more than 2.5A at 12V to run at 0.725V VDD as according to AntMiner-U3 user guide.pdf, the minimum VDD that we can set is 0.725V.

Mine also came with a 7.5A PSU (although it seems it's actually lower). And the specs suggest not lower than 6A.

This is the second time you claimed that you are using 2.5A power supply, so I understand that you didn't mistype it on the first one. However, I couldn't believe that you have your Antminer U3 running using that low ampere of power supply. As far as I know, the older batch 2 of Antminer U3 shipped with 6A of power supply and the newer one shipped with 7.5A. Do you have a picture of that 2.5A power supply?

Honestly there already is a smart robot to do this... it's Crazy Guy's modified R1.  I can run 3 off of it and it fixes the zombies. - https://bitcointalksearch.org/topic/asicpuppynet-terminus-moonlander-2-2pacs-in-stock-btcltcethbch-612390

I also am running with the cheap power bricks that came with them.  So nothing fancy there.  Just a modified R1 and my daily resets of me doing it went away which as a lotto device is really nice not to have to do.

If you use the normal common settings you should have no worry about melting power supplies.  Its when you overclock you run into problems.  I have had my 3 U3's running for 4 months now and my only worry is the power supply is cold, which means it's silently zombied and isn't hashing.

I would be very curious to see a scope of the power consumption circuit.  I am still using the default bricks that came with the unit as nothing I've heard suggested they are more stable with a better supply, and a 25A PSU isn't cheap.  I thought of going the route of the computer supply and using the PCIe power connections but again, I haven't seen anyone here confirm that its more stable.

Earlier someone posted in this thread about the usb and power circuits being shared and I can validate this somewhat as I can run the U3's without the 12v connection but they don't mine, I assume the usb port is not sufficient to power it up properly and when the unit silently zombies I suspect the usb port is getting flaky power.  Resetting it and the port starts communicating again. 

I've tried software based resets but this does not seem to work.  I've tried shutting off the miners but as long as they stay connected via USB, they won't recover until that usb cable is pulled and plugged back in.

Perhaps a smart robot who can watch these all day is the solution

One thing I would suggest is to keep an eye on the power brick that they came with. There has been a few post's about them melting!

with my settings on the page back, with volts at 830;  and the old cgminer (yes, its what works, and I dont want to go through the rigamarole to compile in windows;  yes lazy)...


I am seeing A:54000 R:0 HW:5

Thats an ok HW rate to me (about 1 hour uptime).   It normally wants a replug once every other day.  sometimes goes a week without touching it.

This is on a 2.5A power supply that it shipped with.

The average hashrate is 62.24GH right now.

I have the diff set at 1K because it mines to a solo account.

The math as far as power consumption and settings doesnt make sense....  It defies logic.  it should be having trouble doing its job;  that's for sure.  I wonder what my oscilloscope will show me once I finally bring it home.

aside from thinking the harmonics of the power draw keeping the device charged with just enough juice to stay stable.... its a strange thought.

 I do have a little bit of active cooling on it, but presently, its minimal.  I also replaced the thermal grease with dialectic grease last time I took it apart to clean it.

Woah. Did it happen because of the miner or the connectors/wires were bad? AFAIK they should be rated like 10 A or so. Just repaired an old 550W PSU and made a Molex-5mm adapter to connect the U3. Maybe I should've connected it straight to the 24-pin or PCI-E connector?

Yes, they advertise the maximum speed assuming everything works perfectly and you overclock it, which is a bit misleading if you ask me.  Mine ran at 800v/250f for a while and saw about 60Gh, but it wasn't stable and actually melted two of my molex power plugs.

The default setting is what you get if you launch it without specifying a volt or freq in your batch file.

Seems legit. This means that Bitmain actually advertises it as a 64GHs miner without telling that it's an overclocked mode.

Were that default settings taken from the hardware specs or found experimentally?

stock setting are injected by cgminer = 225MHz 0,775v