I got 2 U3 running on Linux Ubuntu cgminer 4.9.0.
Not once have I managed to run them longer than 24 hours. Does anybody know why they keep going ZOMBIE?
Once they go ZOMBIE like this they are not detected by cgminer and sometimes show in lsusb but not always. The only way to fix this is to remove USB cable and Power, restart cgminer, then plug power back in and USB.
This is a known and notable problem with Antminer U3s. It's discussed at various locations around the Web.
The design couples the USB and miner +5V power rails, which is not good practice. USB interfaces should be both electrically and optically isolated. When a unit goes ZOMBIE, it generally requires unplugging and replugging to restart it, which fits with the above. Some commenters claim that cutting the power lines in the USB cables improves reliability.
However, the maintainer of cgminer comes right out and says that the design is inherently unstable. In light of my own experience I agree. I ran three Jalapenos on a dedicated Pogoplug linux system for a year with no problems. Two U3s have been on the same Pogoplug for two weeks now, and neither has run for 24 hours straight. They're going back to the seller tomorrow.
Long and short of it, if you get a stable unit it will serve you well. But most of them are unstable to at least some degree.
Using CGMiner 4.9.0 I have been trying for MONTHS to get all five of these to all work at the same time without ZOMBIE and ERRORS that make the whole process a waste of time. I did the Zadig thing, a million times, can do it in the sauna, while reading bad Hungarian poetry. All my devices are running the right drivers. I tried new power supplies, new cables, new software...NOTHING WORKED. I want all 5 U3 Miners running normally at once on the same machine! I could get them all to run in 2's or 3's, but with 4 or 5 they would hash at ridiculously low speeds, and crash to zombie a LOT.
I could run ANY THREE at a time, but 4 would start the errors and crashes all over again.
Well, I bought FIVE of these, and I am determined. So I start moving things to a different directory, to "start all over" and while doing this, I notice... I HAVE TWO EXPLORER WINDOWS OPEN AT ONCE. I think to myself. TRY OPENING TWO CgmINERS!!! OK, Now what? one is running 5 U3's terribly, and the other is whining about all the devices being busy, or not there, etc. so I Disable two of those first 5 with the USB Menu in CGMiner. THen I BLACKLIST THOSE TWO. I look at the other window and VOILA!! The two I blacklisted from one CGminer are now hashing like crazy in the other one! I have 5 U3 Miners running at once on the same machine! And if one goes ZOMBIE, the other miner grabs it and wakes it back up!
Here's the setup for the BAT file... --au3-freq 238 --au3-volt 775 I am so happy, I wanna give myself a kiss!!!
did you read the manual?
To prevent the interruption of USB power supply, please ensure the computer is not in standby or sleep mode. Otherwise, the cgminer wouldn’t find the miner even make it ZOMBIE.
source: Page 4 of https://www.bitmaintech.com/files/download/AntMiner-U3%20user%20guide.pdf
I think I might have cracked the issue with the U3 going zombie. I'm sure most miners have their computers set to never sleep, so perhaps the user guide is not specific enough. Under Windows 10, I've gone into device manager, listed devices by connections, found the USB to UART bridge, and then under the power management tab set windows to never set the device to sleep for every hub down to the root hub. So far running a treat.
Hey Friends,
I run the the Antminer U3 with Minera 0.6.1 and the latest official cgminer on my raspberry PI 1 B -> Stable arround 60 GH/s with these Extra settings:
--au3-volt 825 --au3-freq 237.5 = 60 GH/s
--au3-volt 825 --au3-freq 225.0 = 55 GH/s
---> It will start everytime with 4 HW errors, but then after some minutes it will go stable
Have the same problem, keeps getting ZOMBIE.
I solved the problem but power cycle my controller which is a custom beaglebone board, power cycle is made fully automatic by cron jobs.
screen and crontab issues poweroff or equivalent command to power cycle the controller board.
– c2h2
Nov 26 '15 at 4:55
I've recently obtained a U3. I've experienced the zombie issue at 250MHz, 830mV settings. Sometimes cgminer wakes it up again as AU3 1 instead of AU3 0, sometimes it does not. I also have a U2+, so I use the --usb bus:dev option to run them in separate cgminers. I experience a much higher HW rate error on the U3, even at stock settings. HW errors on the U2+ running at 250MHz are rare. So far it is running stable again at 237.5 MHz, 800mV. The default is 225MHz, 750mV, which is 3.33mv per MHz. It could be that running below that is causing the zombie issue. Might try 250MHz again with a higher mV once I am happy at this setting. I have my U3 sat on my computer top front exhaust fan, which is very cool as all the CPU/GPU heat is being blown out of the back.
I added a timer to my rig to powercycle it daily as reboots didn't work anymore. Then this started failing after 3-4 hours of processing. I've adjusted my setting to 237.5 MHz, 825mV and have been stable for a couple of days now, still power cycling it daily.
