Avalon6 safety or how I burnt my avalon6

[philipma1957]

 Second try to write this last try was here.

https://bitcointalk.org/index.php?topic=1335879.msg13629552#msg13629552


Let me state I have had double cataract eye surgery and while I like to read It is harder to do then it used be when I was younger.
I am also colorblind.

So I stated I found wiki's writings to have confused me a bit

the correct number is on the left the number on the right is the one my eyes focused on
that was my error.  But if the 12.8 volt number was not there in larger print on a white background
Maybe I would have realized that 12.7 volts is over volting the avalon6  if it is in the gui reading.

I do not want to discuss if this is clear to every one in the entire world.  It was not clear to me and I messed up.




here are the typical settings I would use.



YEAH  my dual 2000 = 4000 watt server psu  set on high which is 12.75 volts gives me this for 2 avalon sixes




[/quote]

[RichBC]

If it's of any consolation.... Scanning though the documents I could easily see myself making the same mistake...

Also have to say that 12.2V is a relatively low maximum with many PSU's being set to at least that voltage as standard. Am quite surprised that 12.7V would actually cause damage? With this being a string design with 20 chips would mean a core voltage of 0.635V with 12.7V supply v 0.61V with 12.2V supply.

Rich

[VirosaGITS]

If it's of any consolation.... Scanning though the documents I could easily see myself making the same mistake...

Also have to say that 12.2V is a relatively low maximum with many PSU's being set to at least that voltage as standard. Am quite surprised that 12.7V would actually cause damage? With this being a string design with 20 chips would mean a core voltage of 0.635V with 12.7V supply v 0.61V with 12.2V supply.

Rich

Is it possible something carrying the current before it get to the chips burnt out? Maybe something is being used up to its rated safety already and raising the voltage by .7 bust the limit? I dont know.

Either way, thanks Phil for the information, i'll respect that 12.2V at the UI, but that seem kind of icky, since i already get 12.0 to 12.1V at the UI with standard ATX PSU.

[philipma1957]

If it's of any consolation.... Scanning though the documents I could easily see myself making the same mistake...

Also have to say that 12.2V is a relatively low maximum with many PSU's being set to at least that voltage as standard. Am quite surprised that 12.7V would actually cause damage? With this being a string design with 20 chips would mean a core voltage of 0.635V with 12.7V supply v 0.61V with 12.2V supply.

Rich

yeah  well it could be worse  when I killed the 1 board I had 3 avalon6s hooked up which could have killed 6 boards vs 1 board.

Right now I am waiting for a pm or an email from blockc  >  I want to open the ½ dead avalon to see exactly what I did.

But I know the smell of magic smoke   and that is what I smelt and right after that the 1 board went cool the hash rate dropped by 50%.

As I have stated In more then one spot  I should have really looked harder at the documents and I do not want anyone else to do the same.



And like you said the tolerance is really tight. 12.7 vs 12.2 is   under a  5%  over voltage    as 105% of 12.2  volts = 12.81 volts.


@ virosagits 


  well  all power comes from 1 psu

the dual     2000/4000 finsky/j4bberwock


So  maybe part of the  breakout board burnt?  I will check on this idea.

[RichBC]

If it's of any consolation.... Scanning though the documents I could easily see myself making the same mistake...

Also have to say that 12.2V is a relatively low maximum with many PSU's being set to at least that voltage as standard. Am quite surprised that 12.7V would actually cause damage? With this being a string design with 20 chips would mean a core voltage of 0.635V with 12.7V supply v 0.61V with 12.2V supply.

Rich

yeah  well it could be worse  when I killed the 1 board I had 3 avalon6s hooked up which could have killed 6 boards vs 1 board.

Right now I am waiting for a pm or an email from blockc  >  I want to open the ½ dead avalon to see exactly what I did.

But I know the smell of magic smoke   and that is what I smelt and right after that the 1 board went cool the hash rate dropped by 50%.

As I have stated In more then one spot  I should have really looked harder at the documents and I do not want anyone else to do the same.



And like you said the tolerance is really tight. 12.7 vs 12.2 is   under a  5%  over voltage    as 105% of 12.2  volts = 12.81 volts.


@ virosagits 


  well  all power comes from 1 psu

the dual     2000/4000 finsky/j4bberwock


So  maybe part of the  breakout board burnt?  I will check on this idea.

It is possible that the failure is nothing to do with the slightly elevated voltage, it really should have been fine at 12.7V. It's just as likely to be an infant mortality of a part as it is a component that has been overvolted or overpowered? Let's see what you find when you get the go ahead to open up? At least as there was the smell of smoke there will be something to see. Fingers crossed.

And have you have said, and my Mum would have said "It could have been a lot worse".....


Rich

[philipma1957]

If it's of any consolation.... Scanning though the documents I could easily see myself making the same mistake...

Also have to say that 12.2V is a relatively low maximum with many PSU's being set to at least that voltage as standard. Am quite surprised that 12.7V would actually cause damage? With this being a string design with 20 chips would mean a core voltage of 0.635V with 12.7V supply v 0.61V with 12.2V supply.

Rich

yeah  well it could be worse  when I killed the 1 board I had 3 avalon6s hooked up which could have killed 6 boards vs 1 board.

Right now I am waiting for a pm or an email from blockc  >  I want to open the ½ dead avalon to see exactly what I did.

But I know the smell of magic smoke   and that is what I smelt and right after that the 1 board went cool the hash rate dropped by 50%.

As I have stated In more then one spot  I should have really looked harder at the documents and I do not want anyone else to do the same.



And like you said the tolerance is really tight. 12.7 vs 12.2 is   under a  5%  over voltage    as 105% of 12.2  volts = 12.81 volts.


@ virosagits  


  well  all power comes from 1 psu

the dual     2000/4000 finsky/j4bberwock


So  maybe part of the  breakout board burnt?  I will check on this idea.

It is possible that the failure is nothing to do with the slightly elevated voltage, it really should have been fine at 12.7V. It's just as likely to be an infant mortality of a part as it is a component that has been overvolted or overpowered? Let's see what you find when you get the go ahead to open up? At least as there was the smell of smoke there will be something to see. Fingers crossed.

And have you have said, and my Mum would have said "It could have been a lot worse".....


Rich

I got tired of waiting for an answer opened it  >

 I popped a cap

[philipma1957]

Yan will be sending me an avalon6 replacement board.  Not free of charge which is fair since I burned it from my stupidity.

He did give me a discount for running the group buy. 


So to all avalon6 owners  do not use a psu with more then 12.5 volts  and when you check the gui make sure it is under 12.3 volts


I want to thank Yan for the replacement part in advance.

.

[VirosaGITS]

Yan will be sending me an avalon6 replacement board.  Not free of charge which is fair since I burned it from my stupidity.

He did give me a discount for running the group buy.  


So to all avalon6 owners  do not use a psu with more then 12.5 volts  and when you check the gui make sure it is under 12.3 volts


I want to thank Yan for the replacement part in advance.

.

Can you repair this thing? If you're lucky and only the cap is damaged it could be pretty straightforward to get working again?

[RichBC]

My gut feeling is that if only the cap had failed the board might still be working to some degree. More likely is that one or both the ASIC's on that node have gone Open Circuit which has resulted in an increased node voltage which has blown the cap? However well worth replacing the cap, would also check the 4.99 Ohm resistor that is across the Cap as that could have failed as well.


Rich

[philipma1957]

My gut feeling is that if only the cap had failed the board might still be working to some degree. More likely is that one or both the ASIC's on that node have gone Open Circuit which has resulted in an increased node voltage which has blown the cap? However well worth replacing the cap, would also check the 4.99 Ohm resistor that is across the Cap as that could have failed as well.


Rich

I have packed it for shipment to sidehack.  He can play with it.

I am getting a replacement board  at a slight discount from regular price.

[thesmokingman]

Hey Philipma1957-

I know you don't normally perform software updates unless it fixes something from your posts in other threads, but was wondering if you had upgraded your Ras Pi to the latest image? (openwrt-brcm2708-bcm2708-sdcard-vfat-ext4.img   25-Dec-2015 03:37   76M)

Only reason I ask is I'm using Jabber/Finksy BB also, and the volts on the GUI read 11.8~11.9 now no matter what PSU I use. I have 32 DPS-2000BB PSUs and the same ones that I received a readout of 12.1~12.2 on the GUI before the update, now show 11.8~11.9. So I'm wondering if the software is reading low on the volts now? Instead of you running at 12.6, you could of actually been running a lot higher...

Just a thought from my observations. The LCD screen on my PSUs right now reads 12.35, but GUI shows 11.8

[yun9999]

This is where Avalon support stands out.  Well done by Block C.  Although, 1 thing I don't think is right is that Avalon 6 advertise the Hash rate at 3.5 to 3.65TH everywhere but most will only get between 3.2 to 3.4TH unless they risk it via Over volting to get the 3.7-3.9TH and void warranty.  If this was the case with the S7, everyone would be bitching and asking for compensation.  Why is no one asking for compensation for the Avalon 6?  My brother in law is not impress with Avalon 6 and after seeing this post, he stops trying to get a PSU for the over volting so he's just basically have a weak 3.2TH system.  300GH missing is almost 10% difference!

[notlist3d]

This is where Avalon support stands out.  Well done by Block C.  Although, 1 thing I don't think is right is that Avalon 6 advertise the Hash rate at 3.5 to 3.65TH everywhere but most will only get between 3.2 to 3.4TH unless they risk it via Over volting to get the 3.7-3.9TH and void warranty.  If this was the case with the S7, everyone would be bitching and asking for compensation.  Why is no one asking for compensation for the Avalon 6?  My brother in law is not impress with Avalon 6 and after seeing this post, he stops trying to get a PSU for the over volting so he's just basically have a weak 3.2TH system.  300GH missing is almost 10% difference!

You can get in the 3.2 to 3.4 area with 11.9. I bought an EVGA atx just so I was using same PSU's on both miners (I like to be tidy).   But they run 11.9-12.1 all seems to depend. Looking back I should have got some nice server ones instead.

Likely if you are using the "12V DC ±0.2V" I think you could get the 3.5T side of things around the 12.2 side.  But as everyone has seen you should not go above that.

[BTC_ISTANBUL]

I want to ask a question.I have connected my avolon miners but I have overloaded a line.The PSUs are burnt out and also a miner.Miner shows that boards do not have burnt chips but do not hash.What is the not hashing situation?Is it because of rasberry pi or is there a smd on the card?
If this not the appropriate topic pls send me a pm and I can delete my message.

[VirosaGITS]

This is where Avalon support stands out.  Well done by Block C.  Although, 1 thing I don't think is right is that Avalon 6 advertise the Hash rate at 3.5 to 3.65TH everywhere but most will only get between 3.2 to 3.4TH unless they risk it via Over volting to get the 3.7-3.9TH and void warranty.  If this was the case with the S7, everyone would be bitching and asking for compensation.  Why is no one asking for compensation for the Avalon 6?  My brother in law is not impress with Avalon 6 and after seeing this post, he stops trying to get a PSU for the over volting so he's just basically have a weak 3.2TH system.  300GH missing is almost 10% difference!

You can get in the 3.2 to 3.4 area with 11.9. I bought an EVGA atx just so I was using same PSU's on both miners (I like to be tidy).   But they run 11.9-12.1 all seems to depend. Looking back I should have got some nice server ones instead.

Likely if you are using the "12V DC ±0.2V" I think you could get the 3.5T side of things around the 12.2 side.  But as everyone has seen you should not go above that.

Is there a particular thing to do, to get 3.2 to 3.4THs? I get wildly different temps and different hashrate and no where near 3.4THs. Is it a overvolt thing only? I'm on ATX psu but i usually get 12.1v at the ui;



I see "smartspeed" is on, not sure if i need to do manual tuning or something?

[philipma1957]

This is where Avalon support stands out.  Well done by Block C.  Although, 1 thing I don't think is right is that Avalon 6 advertise the Hash rate at 3.5 to 3.65TH everywhere but most will only get between 3.2 to 3.4TH unless they risk it via Over volting to get the 3.7-3.9TH and void warranty.  If this was the case with the S7, everyone would be bitching and asking for compensation.  Why is no one asking for compensation for the Avalon 6?  My brother in law is not impress with Avalon 6 and after seeing this post, he stops trying to get a PSU for the over volting so he's just basically have a weak 3.2TH system.  300GH missing is almost 10% difference!

well I have paid for the replacement board.  I am awaiting the parts.

The replacement board which I was given a decent discount.

Plus one doa blue adpater which should be free as It arrived doa.

Once parts arrive I will post.


_____________________
++++++++++++++++++
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

Hey Philipma1957-

I know you don't normally perform software updates unless it fixes something from your posts in other threads, but was wondering if you had upgraded your Ras Pi to the latest image? (openwrt-brcm2708-bcm2708-sdcard-vfat-ext4.img   25-Dec-2015 03:37   76M)

Only reason I ask is I'm using Jabber/Finksy BB also, and the volts on the GUI read 11.8~11.9 now no matter what PSU I use. I have 32 DPS-2000BB PSUs and the same ones that I received a readout of 12.1~12.2 on the GUI before the update, now show 11.8~11.9. So I'm wondering if the software is reading low on the volts now? Instead of you running at 12.6, you could of actually been running a lot higher...

Just a thought from my observations. The LCD screen on my PSUs right now reads 12.35, but GUI shows 11.8

no stock 11-20 or so firmware.

my lcd screen reads 12.75 volts my gui read 12.4 to 12.7 on the three different avalon's

BTW  the back probe at the avalon side all read about 12.5 volts  as the voltage drops with the draw.






_____________________________
@@@@@@@@@@@@@@@@@@
VVVVVVVVVVVVVVVVVVVVVVVVVV

I want to ask a question.I have connected my avolon miners but I have overloaded a line.The PSUs are burnt out and also a miner.Miner shows that boards do not have burnt chips but do not hash.What is the not hashing situation?Is it because of rasberry pi or is there a smd on the card?
If this not the appropriate topic pls send me a pm and I can delete my message.

if you post here


https://bitcointalk.org/index.php?board=40.0


on the support board  I could try to troubleshoot what is wrong.

This is where Avalon support stands out.  Well done by Block C.  Although, 1 thing I don't think is right is that Avalon 6 advertise the Hash rate at 3.5 to 3.65TH everywhere but most will only get between 3.2 to 3.4TH unless they risk it via Over volting to get the 3.7-3.9TH and void warranty.  If this was the case with the S7, everyone would be bitching and asking for compensation.  Why is no one asking for compensation for the Avalon 6?  My brother in law is not impress with Avalon 6 and after seeing this post, he stops trying to get a PSU for the over volting so he's just basically have a weak 3.2TH system.  300GH missing is almost 10% difference!

You can get in the 3.2 to 3.4 area with 11.9. I bought an EVGA atx just so I was using same PSU's on both miners (I like to be tidy).   But they run 11.9-12.1 all seems to depend. Looking back I should have got some nice server ones instead.

Likely if you are using the "12V DC ±0.2V" I think you could get the 3.5T side of things around the 12.2 side.  But as everyone has seen you should not go above that.

Is there a particular thing to do, to get 3.2 to 3.4THs? I get wildly different temps and different hashrate and no where near 3.4THs. Is it a overvolt thing only? I'm on ATX psu but i usually get 12.1v at the ui;



I see "smartspeed" is on, not sure if i need to do manual tuning or something?

I am not sure  what your question is.  the screen shot above is showing 3.5th  is that you?

[notlist3d]

Is there a particular thing to do, to get 3.2 to 3.4THs? I get wildly different temps and different hashrate and no where near 3.4THs. Is it a overvolt thing only? I'm on ATX psu but i usually get 12.1v at the ui;



I see "smartspeed" is on, not sure if i need to do manual tuning or something?

(shortened post quite a bit)

I am not sure  what your question is.  the screen shot above is showing 3.5th  is that you?

I think hes asking what causes the lower then 3.5 area, I'm guessing he has not experienced it.   The most common reason I would say is the PSU puts out less then 12.0 volts.

Not all PSU's are the same on putting out exactly 12 or 12.1 even.  I have some EVGA's that worked great for everything else but with A6's they tend to be 11.9-12 most of time.

[VirosaGITS]

Is there a particular thing to do, to get 3.2 to 3.4THs? I get wildly different temps and different hashrate and no where near 3.4THs. Is it a overvolt thing only? I'm on ATX psu but i usually get 12.1v at the ui;



I see "smartspeed" is on, not sure if i need to do manual tuning or something?

(shortened post quite a bit)

I am not sure  what your question is.  the screen shot above is showing 3.5th  is that you?

I think hes asking what causes the lower then 3.5 area, I'm guessing he has not experienced it.   The most common reason I would say is the PSU puts out less then 12.0 volts.

Not all PSU's are the same on putting out exactly 12 or 12.1 even.  I have some EVGA's that worked great for everything else but with A6's they tend to be 11.9-12 most of time.

Yes, sorry for being unclear, i'm wondering why the hashrate varies so much. For example, it look like the first unit, in the screenshot (yes that is me) can do 3550 to 3650 depending on uncertain factors. While some units does 3.2 to 3.4 and the GUI still report 11.9v to 12.0v.

I am under the impressing these prefer running hot? because generally unit #1 does 3650 and the second one does 3550. But i am not sure why such a varriance&difference in hashrate and also temps, they are each using a EVGA G2 1300w.

Now the real problem i have right now is, if i set the minimum fan speed to 60, it fix the issue of the fan going from 2.5k to 5k all the time (i find it insanely annoying). The fan does 3-3.2k all the time which is great. But if i do that unit #1 seem to run too cool(?) and lose 100 GH/s?

I'm just trying to ascertain what is going on and see if someone has a workaround. Maybe controlling the fan of each miner individually?

[philipma1957]

Is there a particular thing to do, to get 3.2 to 3.4THs? I get wildly different temps and different hashrate and no where near 3.4THs. Is it a overvolt thing only? I'm on ATX psu but i usually get 12.1v at the ui;



I see "smartspeed" is on, not sure if i need to do manual tuning or something?

(shortened post quite a bit)

I am not sure  what your question is.  the screen shot above is showing 3.5th  is that you?

I think hes asking what causes the lower then 3.5 area, I'm guessing he has not experienced it.   The most common reason I would say is the PSU puts out less then 12.0 volts.

Not all PSU's are the same on putting out exactly 12 or 12.1 even.  I have some EVGA's that worked great for everything else but with A6's they tend to be 11.9-12 most of time.

Yes, sorry for being unclear, i'm wondering why the hashrate varies so much. For example, it look like the first unit, in the screenshot (yes that is me) can do 3550 to 3650 depending on uncertain factors. While some units does 3.2 to 3.4 and the GUI still report 11.9v to 12.0v.

I am under the impressing these prefer running hot? because generally unit #1 does 3650 and the second one does 3550. But i am not sure why such a varriance&difference in hashrate and also temps, they are each using a EVGA G2 1300w.

Now the real problem i have right now is, if i set the minimum fan speed to 60, it fix the issue of the fan going from 2.5k to 5k all the time (i find it insanely annoying). The fan does 3-3.2k all the time which is great. But if i do that unit #1 seem to run too cool(?) and lose 100 GH/s?

I'm just trying to ascertain what is going on and see if someone has a workaround. Maybe controlling the fan of each miner individually?

leave the fans at 25 they will ramp up less.

since i have a 1/2 unit i will try a different fan and some other fun.

[notlist3d]

Is there a particular thing to do, to get 3.2 to 3.4THs? I get wildly different temps and different hashrate and no where near 3.4THs. Is it a overvolt thing only? I'm on ATX psu but i usually get 12.1v at the ui;



I see "smartspeed" is on, not sure if i need to do manual tuning or something?

(shortened post quite a bit)

I am not sure  what your question is.  the screen shot above is showing 3.5th  is that you?

I think hes asking what causes the lower then 3.5 area, I'm guessing he has not experienced it.   The most common reason I would say is the PSU puts out less then 12.0 volts.

Not all PSU's are the same on putting out exactly 12 or 12.1 even.  I have some EVGA's that worked great for everything else but with A6's they tend to be 11.9-12 most of time.

Yes, sorry for being unclear, i'm wondering why the hashrate varies so much. For example, it look like the first unit, in the screenshot (yes that is me) can do 3550 to 3650 depending on uncertain factors. While some units does 3.2 to 3.4 and the GUI still report 11.9v to 12.0v.

I am under the impressing these prefer running hot? because generally unit #1 does 3650 and the second one does 3550. But i am not sure why such a varriance&difference in hashrate and also temps, they are each using a EVGA G2 1300w.

Now the real problem i have right now is, if i set the minimum fan speed to 60, it fix the issue of the fan going from 2.5k to 5k all the time (i find it insanely annoying). The fan does 3-3.2k all the time which is great. But if i do that unit #1 seem to run too cool(?) and lose 100 GH/s?

I'm just trying to ascertain what is going on and see if someone has a workaround. Maybe controlling the fan of each miner individually?

leave the fans at 25 they will ramp up less.

since i have a 1/2 unit i will try a different fan and some other fun.

Out of curiosity what does the miner seem to run like that has 1/2 of boards now?  Does it seem to run fine just has only 1 board?

Was not sure with smartspeed what it was set to do if anything different when just 1 hashing unit.