[Review] Avalon 6 Miner - Winter Mining - Notlist3d - Also FAQ and Help

[philipma1957]

I spun the A6 around and gave it plenty of room to breath, plus some cool evening air from outside.


It ran really well for about 15 minutes, fan did go up and down a bit from 6-8000 rpm, and just after I took the screen cap below, the fan died down and I watch the hash drop down to nothing.  Wattage at the wall went from 1050+ to ~60.  The light on the front of the unit was flashing yellow.



I refreshed the gui and it still read the same.  I restarted cgminer and it still didn't change (yellow light still flashing), so I rebooted it via the gui.  It's been running now for

7m 15s  Temps 35 65 74
3.4TH

17m 8s  Temps 35 66 75
3.5TH

34m 34s  Temp 35 67 76
3.57TH

1h 19m  Temps 35 67 75
3.5TH

Will update, but seems to be stabilized now.

At 65 dB it is quieter than my S5.

The Sanyo ace should drop the fan speeds down and you will be quieter.

What did you set the fan speed at 40?  The freq at 500 with fan at 40-50 works well.

[1714309484]

1714309484

[1714309484]

1714309484

[1714309484]

1714309484

[1714309484]

1714309484

[bctmke]

Hah I messed up my cron job.  It constantly restarting cgminer today (the 18th).  I had the right day of the month - but I should've just told it to do it once   Was wondering why my hashrate stayed low and numbers kept resetting!

[Mikestang]

The Sanyo ace should drop the fan speeds down and you will be quieter.

What did you set the fan speed at 40?  The freq at 500 with fan at 40-50 works well.

The fan rpms dropped as the evening got cooler, they hung around 5000rpm all night.  The unit ran for over 9 hours and averaged ~3.4+ TH/s for the run.  I shut it off for the day because it will be too warm in my house when I get home if I left it running.

I am impressed with how quiet the stock fan is compared to the stock fan on the S5, so if it gets even quieter from here that will be a bonus.  Also on the comparing to S5 front, I like that for only 300 watts more I get more than double the hash.  I look forward to receiving the fan and checking it with my dB meter.

Yes, I set the fan to 40 (lucky guess I guess!), but I do not know how/where to read the freq - isn't it dynamic?  Do you change the fan setting with the Sanyo ace?

[philipma1957]

The Sanyo ace should drop the fan speeds down and you will be quieter.

What did you set the fan speed at 40?  The freq at 500 with fan at 40-50 works well.

The fan rpms dropped as the evening got cooler, they hung around 5000rpm all night.  The unit ran for over 9 hours and averaged ~3.4+ TH/s for the run.  I shut it off for the day because it will be too warm in my house when I get home if I left it running.

I am impressed with how quiet the stock fan is compared to the stock fan on the S5, so if it gets even quieter from here that will be a bonus.  I look forward to receiving the fan and checking it with my dB meter.

Yes, I set the fan to 40 (lucky guess I guess!), but I do not know how/where to read the freq - isn't it dynamic?  Do you change the fan setting with the Sanyo ace?

the 40 is good for the sanyo or the stock.

basically they move the same air  on 40 but the stock will spin at 5500 rpm to 7500 rpm

the sanyo will spin at 3800 rpm to 5000 rpm.

the setting is very dynamic  the freq setting is right above the fan setting..... I do 500 and 40

frankly the software has a mind of its own  .. it checks volts and temp readings as set a fan speed and a hash speed based what it reads.

Ie it over rides the freq 500 fan 40  but not by much.  but it does overide and I have not yet ever been able to take over full control of freq and fan .

your settings are okay biggest issues with this gear are:

do not over volt  12.2 max gui reading which means 12.4 max leaving the psu.

let the fan blow air without restrictions. big heat sink 1 fan  you need to allow fan to move the air.

lastly  keep room cool.




I do wonder what will happen to these in 95f summer heat.  Well I will find it out this summer.

[tlhIlwI]

do not over volt  12.2 max gui reading which means 12.4 max leaving the psu.

Are you getting a 12.2 reading from 12.4 at the PSU on batch 2?  I'm curious since I get 12.2V from a 12.15V PSU reading on my last one.  I've been seeing around 3780 on Kano.is and in the GHSav field which seems more consistent with 12.15V than with 12.2V (which should be in the 3800 to 3900 range).  The miner rate at the bottom of the GUI shows 3850 though, even though I'm only seeing a 3780 average.  I've often wondered if I'm being too conservative with the voltage adjust and the internal sensor is just on the high side, but I've been afraid to bump it up any higher (to an actual 12.2V).

I'm eager to see what the one arriving later today gives me at 12.2V in the GUI.

Edit:  When I think about it, I'm only running 8 inch 16 gauge leads from the PSU to the miner instead of a more typical 2ft. of 18ga., so maybe the discrepancy is simply in having less cable loss.

[Mikestang]

I didn't recall the freq. setting in the gui, I'll check it out tonight.  Although I was very happy with how it ran without me changing anything except the fan setting.

let the fan blow air without restrictions. big heat sink 1 fan  you need to allow fan to move the air.

lastly  keep room cool.

Yes, I gave it plenty of room to breath, that was my problem with the initial setup, now it exhausts into a large open area.  I guess my dryer vent exhaust maxes out at the S5, anything more needs to move more air than the vent can handle.  The volume of air coming out of the A6 is impressive, I can stand 8' away and still feel a strong, warm breeze.  I left with window open last night in addition to the dryer vent providing cool air for the unit, that's how I'll continue to run it until summer starts to set in here.

I won't have to worry how it will do in the summer heat because I can't run my miners in the summer, my electricity bill is high enough with just my AC, and miners make is worse x2 because they 1) use more power than my AC and they 2) add more heat to my house for my AC to deal with, so I just turn them off when it gets too hot - or will be having them hosted this year.  Ahh, life in a desert...

[philipma1957]

do not over volt  12.2 max gui reading which means 12.4 max leaving the psu.

Are you getting a 12.2 reading from 12.4 at the PSU on batch 2?  I'm curious since I get 12.2V from a 12.15V PSU reading on my last one.  I've been seeing around 3780 on Kano.is and in the GHSav field which seems more consistent with 12.15V than with 12.2V (which should be in the 3800 to 3900 range).  The miner rate at the bottom of the GUI shows 3850 though, even though I'm only seeing a 3780 average.  I've often wondered if I'm being too conservative with the voltage adjust and the internal sensor is just on the high side, but I've been afraid to bump it up any higher (to an actual 12.2V).

I'm eager to see what the one arriving later today gives me at 12.2V in the GUI.

Edit:  When I think about it, I'm only running 8 inch 16 gauge leads from the PSU to the miner instead of a more typical 2ft. of 18ga., so maybe the discrepancy is simply in having less cable loss.

I agree with the bold statement.

[tlhIlwI]

do not over volt  12.2 max gui reading which means 12.4 max leaving the psu.

Are you getting a 12.2 reading from 12.4 at the PSU on batch 2?  I'm curious since I get 12.2V from a 12.15V PSU reading on my last one.  I've been seeing around 3780 on Kano.is and in the GHSav field which seems more consistent with 12.15V than with 12.2V (which should be in the 3800 to 3900 range).  The miner rate at the bottom of the GUI shows 3850 though, even though I'm only seeing a 3780 average.  I've often wondered if I'm being too conservative with the voltage adjust and the internal sensor is just on the high side, but I've been afraid to bump it up any higher (to an actual 12.2V).

I'm eager to see what the one arriving later today gives me at 12.2V in the GUI.

Edit:  When I think about it, I'm only running 8 inch 16 gauge leads from the PSU to the miner instead of a more typical 2ft. of 18ga., so maybe the discrepancy is simply in having less cable loss.

I agree with the bold statement.

The new miner from the group buy arrived and is (finally) set up.  This one wouldn't get past a flashing yellow even with everything set up and configured perfectly.  The miner wasn't detected in the UI.  I switched USB-SPI adapters, cables, etc. with no joy.  Finally I swapped out the SD card with a spare I keep on hand and it immediately came to life.  It looks like I had a corrupted SD card!

The new miner reads 12.1V in the UI with the PSU set at 12.220V.  It is reading 3900GH, so I'm inclined to believe my multimeter more than the UI on this one... I'm not going to push it all the way to 12.2V in the UI. 

This does put a hole in the cable length hypothesis though.  It seems the voltage sensing can be off by a bit on these miners.

[philipma1957]

do not over volt  12.2 max gui reading which means 12.4 max leaving the psu.

Are you getting a 12.2 reading from 12.4 at the PSU on batch 2?  I'm curious since I get 12.2V from a 12.15V PSU reading on my last one.  I've been seeing around 3780 on Kano.is and in the GHSav field which seems more consistent with 12.15V than with 12.2V (which should be in the 3800 to 3900 range).  The miner rate at the bottom of the GUI shows 3850 though, even though I'm only seeing a 3780 average.  I've often wondered if I'm being too conservative with the voltage adjust and the internal sensor is just on the high side, but I've been afraid to bump it up any higher (to an actual 12.2V).

I'm eager to see what the one arriving later today gives me at 12.2V in the GUI.

Edit:  When I think about it, I'm only running 8 inch 16 gauge leads from the PSU to the miner instead of a more typical 2ft. of 18ga., so maybe the discrepancy is simply in having less cable loss.

I agree with the bold statement.

The new miner from the group buy arrived and is (finally) set up.  This one wouldn't get past a flashing yellow even with everything set up and configured perfectly.  The miner wasn't detected in the UI.  I switched USB-SPI adapters, cables, etc. with no joy.  Finally I swapped out the SD card with a spare I keep on hand and it immediately came to life.  It looks like I had a corrupted SD card!

The new miner reads 12.1V in the UI with the PSU set at 12.220V.  It is reading 3900GH, so I'm inclined to believe my multimeter more than the UI on this one... I'm not going to push it all the way to 12.2V in the UI.  

This does put a hole in the cable length hypothesis though.  It seems the voltage sensing can be off by a bit on these miners.

Very smart to think of the sdcard.

I just stumbled on a rasp pi in a parts box and I used the link on first page of this thread to make a new sdcard.
So I now have four rasp pi's and six Avalon 6's
.
Forces me to  run some of them as daisy chain but I have some back up.

I will be shifting some of the avalons from my garage to the solar array some time next week.

I will photo the new set up there.

[tlhIlwI]

Is it safe to disable and turn off odhcpd on the Avalon6 Raspberry Pi System/Startup tab?  

I *think* this is a DHCP server, not the DHCP client.  I ask because I happened to notice PCs on the LAN showing up as DHCP leases in the Status/Overview.  I don't want the Avalon dishing out DHCP leases (that is the job of my firewall/router).  Only one DHCP server should be on the LAN-- I'm not sure why it is trying to act as a DHCP server in the first place.

[philipma1957]

Is it safe to disable and turn off odhcpd on the Avalon6 Raspberry Pi System/Startup tab?  

I *think* this is a DHCP server, not the DHCP client.  I ask because I happened to notice PCs on the LAN showing up as DHCP leases in the Status/Overview.  I don't want the Avalon dishing out DHCP leases (that is the job of my firewall/router).  Only one DHCP server should be on the LAN-- I'm not sure why it is trying to act as a DHCP server in the first place.

Simple answer no.

Correct answer make a back up sdcard check that it works .

Once you have two working sdcards try to do just what you asked.

Worst case is you lose the ability to connect to the rasp pi.

If so pull that sdcard stick in the working sdcard.  And re image the card you killed.

[Mikestang]

I didn't recall the freq. setting in the gui, I'll check it out tonight.  Although I was very happy with how it ran without me changing anything except the fan setting.

Frequency was/is set at 500.

New fan is installed an spinning at 3780 rpm.  Ambient temp. tonight is same as last night, but unit is running cooler after an hour it's 34 62 71, but hash is also down tonight vs last night at 3.29TH so that could be the temp. difference.

It is much quieter, barely 61 dB.  I can't hear it when I watch tv or when I'm using my tower computer.

[tlhIlwI]

Is it safe to disable and turn off odhcpd on the Avalon6 Raspberry Pi System/Startup tab?  

I *think* this is a DHCP server, not the DHCP client.  I ask because I happened to notice PCs on the LAN showing up as DHCP leases in the Status/Overview.  I don't want the Avalon dishing out DHCP leases (that is the job of my firewall/router).  Only one DHCP server should be on the LAN-- I'm not sure why it is trying to act as a DHCP server in the first place.

Simple answer no.

Correct answer make a back up sdcard check that it works .

Once you have two working sdcards try to do just what you asked.

Worst case is you lose the ability to connect to the rasp pi.

If so pull that sdcard stick in the working sdcard.  And re image the card you killed.

Yeah... that was about what I was thinking as well.  Here is a report back:

odhcpd has been disabled on both of my A6s for 18 hours now.  I even rebooted one just to make sure it wasn't required for startup.  So far I have seen no problems whatsoever with keeping it disabled.  I really don't know why it is there in the first place (other than it came along with OpenWRT and nobody disabled it when adapting OpenWRT to be a miner firmware).

I've narrowed my issues down some.  To further clarify the issue with running odhcpd-- it appears the biggest problems were related to interfering with IPv6 addresses, not IPv4, on the LAN and some LAN specific DNS issues.  My ISP does support IPv6 and I have it fully set up and in use here.  Where I first noticed an issue was intermittently with updates on one of my Linux systems and the occasional glitch reaching Google on the first try on the same (both of which tend to prefer IPv6 before IPv4).  I was also noticing intermittent problems with resolving names from my local DNS server (I have one set up on my firewall to resolve local hosts from a fake .lan TLD to their assigned DHCP addresses)  Those issues began when I added my first A6, but the connection between my issues and the A6 didn't occur to me until I saw the DHCP lease listed in the web interface.  Turning off odhcpd on the A6 cleared up all of my IPv6 and local DNS problems.

I still am worried that odhcpd was turned on for some reason other than mere oversight of whoever built the firmware, but after 18 hours I haven't seen any problems with leaving it off.  I'll keep testing it, but for now if anyone sees any IPv6 and/or DNS issues on their LAN and sees their PCs showing up as DHCP leases in the A6/rPi, you might want to look at backing up your SD card and trying to turn off odhcpd.  There should only be *one* DHCP server per LAN subnet!

[Mikestang]

Ran all night and all day like a champ, it's so much quieter with the new fan I can barely hear it in the house, very nice.

Code:

{"hashrate1m": "3.76T", "hashrate5m": "3.39T", "hashrate1hr": "3.33T", "hashrate1d": "1.86T", "hashrate7d": "458G", "lastupdate": 1458419183}

[alh]

My observations on the A6 are fairly new, but now that I have a correctly manufactured 4-pin cable connection, it's easier to experiment a bit and report some accurate results.

My first observation is that the "Chip Maximum Frequency" is pretty much ignored and a useless value. While I wish to repeat some experiments, I am confident that that after 2 hours of running, there is virtually no difference of the actual hashrate, or the fan speed that is present. I've tried values of 300, 400 and 500, and in all three cases after two or more hours, the miner is at roughly 3500 GH, and the fan at about 6700 RPM. That of course is way above the 35 value I set for a minimum fan speed. If you only look at the first 15 minutes of running, there is a difference in 300, 400, and 500 for chip frequency. After a couple of hours though, those differences are pretty much gone. I credit this to the the "Smart Speed" firmware. I think this may well be related to Phil's comment "the software has a mind of its own".

I am betting right now that the "Smart Speed" functionality is contained within the MM firmware, and NOT the Raspberry Pi image. On my used miner, that is version 601512-57532250 as I read the Advance cgminer status display. As I read the canaan website, that is one of the MM firmware versions that was released after the initial A6 hardware release in November.

I'd be interested to hear about other folks experience with their ability to "underclock" the A6 in terms of it's actual speed and fan RPM after a few hours. I am probably a bit spoiled by my SP20 which had really good range and control of the actual hashrate, and therefore heat and fan speed.

[Mikestang]

That of course is way above the 35 value I set for a minimum fan speed.

It is permissible, by definition, to exceed a minimum value.  Your fans will not go below 35%.

[notlist3d]

My observations on the A6 are fairly new, but now that I have a correctly manufactured 4-pin cable connection, it's easier to experiment a bit and report some accurate results.

My first observation is that the "Chip Maximum Frequency" is pretty much ignored and a useless value. While I wish to repeat some experiments, I am confident that that after 2 hours of running, there is virtually no difference of the actual hashrate, or the fan speed that is present. I've tried values of 300, 400 and 500, and in all three cases after two or more hours, the miner is at roughly 3500 GH, and the fan at about 6700 RPM. That of course is way above the 35 value I set for a minimum fan speed. If you only look at the first 15 minutes of running, there is a difference in 300, 400, and 500 for chip frequency. After a couple of hours though, those differences are pretty much gone. I credit this to the the "Smart Speed" firmware. I think this may well be related to Phil's comment "the software has a mind of its own".

I am betting right now that the "Smart Speed" functionality is contained within the MM firmware, and NOT the Raspberry Pi image. On my used miner, that is version 601512-57532250 as I read the Advance cgminer status display. As I read the canaan website, that is one of the MM firmware versions that was released after the initial A6 hardware release in November.

I'd be interested to hear about other folks experience with their ability to "underclock" the A6 in terms of it's actual speed and fan RPM after a few hours. I am probably a bit spoiled by my SP20 which had really good range and control of the actual hashrate, and therefore heat and fan speed.

I don't think anyone has been able to turn off "Smart Speed".   So it's a good miner just not going to be one you will under and overclock.   I eventually just let mine sit on default and am happy.  Fan speed is not a issue as I have some big fan's pushing hot air away from miners.

The ones that were getting control of speed was through voltage.... I choose not to do it to mine.   You have to be very careful with playing with voltage or you end up with a brick.  

[borisg26]

Hi

  I have some, not working  Avalon 6 blades , with expired warranties, anyone knows if Avalon in China or USA offers service repairs for these blades.

Thanks

[notlist3d]

Hi

  I have some, not working  Avalon 6 blades , with expired warranties, anyone knows if Avalon in China or USA offers service repairs for these blades.

Thanks

I would guess they would sell you another blade, but don't know on repairs.  Have not heard of repairs, just paying for new blades.   But contact BlockC and see what they say - http://www.blockc.co/#contact

Out of curiosity did you over volt them?  Or what caused the problems.

[tlhIlwI]

Is it safe to disable and turn off odhcpd on the Avalon6 Raspberry Pi System/Startup tab?  

I *think* this is a DHCP server, not the DHCP client.  I ask because I happened to notice PCs on the LAN showing up as DHCP leases in the Status/Overview.  I don't want the Avalon dishing out DHCP leases (that is the job of my firewall/router).  Only one DHCP server should be on the LAN-- I'm not sure why it is trying to act as a DHCP server in the first place.

Simple answer no.

Correct answer make a back up sdcard check that it works .

Once you have two working sdcards try to do just what you asked.

Worst case is you lose the ability to connect to the rasp pi.

If so pull that sdcard stick in the working sdcard.  And re image the card you killed.

Update: odhcpd has been disabled for 4 days now with no problems.  It appears safe to disable it to fix any issues caused by multiple DHCP servers on the same LAN subnet.  It probably just came along for the ride and wasn't caught when OpenWRT was adapted into a mining firmware.