Topic: Official Avalon Technical Support Thread - page 6. (Read 48885 times)

Have you checked to make sure all the cables are firmly connected?

Gotta love your comments

All this is 220V so the PSUs are more efficient.

600 watts was lying around, I pulled power and data off one blade and it powers 2, and yes, didn't work for even 1 for more than an hour.
The list of known working PSUs started with 650 for 3 blades, so why wouldn't I try 600 with 2?

1000 watts didn't work with 3 for more than a day, well, it didn't work with 2 blades nor 1 blade either for more than a few minutes.  The PSU didn't die, the Avalon did.

800 watts worked fine with 3 for me for more than a week.

So, if 800 is underpowered, why did the 1000 watt PSU die after a day and the 800 work for a week?

You are totaly stupid. First you buy a 600 Watt PSU then a 1000W and then a 800 Watt? Are you fucking serious? You did recognize that the 600W is waaaay underpowered. WHY the hell do you buy a 800 watt psu?

800 watt is waaaay underpowered!!! It cant power all chips. Its not strong enaugh. It gives you only 700 watt on the 12volt line and thats not made for 24/7

If you want a psu and use it 24/7  then dont give him more then 80% load

I received my Avalon batch 3 w/ 3 blades last week, and here's some of the problems I encountered.

The first set of problems seems to be with power supplies.

The first I tried was an OCZ 600W, obviously underpowered, and mining would stop as soon as I started.

So I bought a second one, a Seasonic Platinum 1000W.
This can power on, but it proved to eventually intermittently die every other time.
When it dies, the WRT703 is still alive, but all the fans would suddenly stop and it would no longer mine.  It always needs a hard power off/power on.
Finally I got it running for a day, but in the middle it died again, and after that it never did manage to power on the blades again, only the WRT 703.
At this point I changed everything, upgraded the firmware, plugged each blade one by one, nothing seems to work.
When I turn on the power it keeps powering the WRT 703 but the fans go on high, and then one second later immediately shuts off - and mining will never be successful.

I changed this to a corsair 800 GS and now it is mining for a week continuously, I even overclocked to 350Mhz.


However, now I have another problem, my miner never goes above the advertised 66Gh/s spec.
Even at 350Mhz it maxes out at 58Gh.

I looked in this forum and saw this is how to find out if my avalon blades are working:

# cgminer-api stats  # shows:

[STATS0] =>
(
   [STATS] => 0
   [ID] => AVA0
   [Elapsed] => 4711
   [Calls] => 0
   [Wait] => 0.000000
   [Max] => 0.000000
   [Min] => 99999999.000000
   [baud] => 115200
   [miner_count] => 24
   [asic_count] => 10
   [timeout] => 36
   [frequency] => 350
   [fan1] => 2160
   [fan2] => 2280
   [fan3] => 1680
   [temp1] => 19
   [temp2] => 44
   [temp3] => 45
   [temp_max] => 46
   [no_matching_work] => 186
   [match_work_count1] => 3833
   [match_work_count2] => 3724
   [match_work_count3] => 3837
   [match_work_count4] => 3855
   [match_work_count5] => 3861
   [match_work_count6] => 754    <-- slower
   [match_work_count7] => 419    <-- slower
   [match_work_count8] => 0       <-- no response?
   [match_work_count9] => 3797
   [match_work_count10] => 3907
   [match_work_count11] => 3733
   [match_work_count12] => 3768
   [match_work_count13] => 3850
   [match_work_count14] => 3823
   [match_work_count15] => 0      <-- no response?
   [match_work_count16] => 0      <-- no response?
   [match_work_count17] => 3724
   [match_work_count18] => 3847
   [match_work_count19] => 3856
   [match_work_count20] => 3798
   [match_work_count21] => 3931
   [match_work_count22] => 3791
   [match_work_count23] => 0      <-- no response?
   [match_work_count24] => 0      <-- no response?
   [USB Pipe] => 0
   [USB Delay] => r0 0.000000 w0 0.000000
)


It seems I received a unit missing 28% of my hash power
Does anyone have any ideas on what I can do with this?


In summary, here's a list of things I had problems with
- Didn't know how to check to see if blades mine - look at the cgminer stats using "cgminer-api stats" and looking at the [AVA0] section
- power supply connection -
  I suspect the power supply may have shut down by itself if it detects some sort of loop, maybe, is it possible that the high grade PSU shuts itself down when detecting a defective board that drew imbalanced voltages between the avalon blades?   When I changed to a Corsair 800GS, it is a much lower grade power supply than my Seasonic platinum 1000w, and the lower grade power supply works - maybe it doesn't do these checks and powers away anyway, enabling my avalon to mine until now?

Any comments appreciated,

Thanks!

Avalons are DIY. There is no Customer Support. There is no Warranty Service. You are on your own.

I consider buying one Avalon from batch 2. Can anybody tell me is there some warranty for Avalons? I'm just wondering what would happen if it brokes in a month or two.

Hi, I'm doing my own version of avalon and I find that error to start the cgminer, someone would know tell me that can occur?

I have solved the problem:
https://bitcointalksearch.org/topic/m.2901631

My batch#3 Avalon with (4 modules) has just arrived.
After few hours of mining I noticed that some of the match_work_counts are zero:


It looks like some of the chips are not working.
I rebooted the unit many times but the same match_work_counts are always zero.
The unit running stable but only with 50-57 Gh/s.
What can I do?

Most intriguing part is that HW error rate in my case is far lower than hardcoded 2% but it still don't want to return to 350 MHz  

Manual cgminer restart fixes this in a seconds.

So anybody know how to add 'kill cgminer' into cron schedule on this OpenWrt OS?

In the past month, ambient for my Avalon has ranged between 45°F and 90°F.

What I would like is a program on the TP-Link 703n that will send the statistics to a mySQL database somewhere.

I'd be happy with just simple restart as it cools down in a minute and then auto-tunes itself according to the new temperature conditions.

and better -> restart with another configuration because of other ambient temperature...

Can you suggest how one can make cgminer restart scheduled with cron? I'd like to add restart in the evening, after sunset.

Mine does this.  I assumed it would before I ever powered it up.  Hotter transistors leak more.

Guys,

I see strange behavior of --avalon-auto option on my Avalon B2 unit. It finely clocks down modules at the day's hot time to 318 MHz but then at night it fails to climb up Freq to the original value (350 in my case). Manual terminating of the cgminer process followed by restarting it by built-in watchdog process /usr/bin/cgminer-monitor fixes this issue immediately. FW 20130703.

Anybody sees the same?

@ckolivas,
Do you suspect any bugs in the algorithm?

335MHz and stable

the batch 1 avalons that I have - the ones I have upgraded to 4 modules myself and upgraded the PSU's I have always just ran the cables out the back and kept the PSU outside the case. Its practical, allows for more psu choices and also improves airflow.


Also I am curious if anyone has had any luck overclocking batch 1 avalons?

-JJ

Simply put the psu outside