Custom Innosilicon A2 Terminator image - Anx Edition

[quantumgravity]

Thanks for the info! Very helpful!

[citronick]

Hi Mark,

I have been mining with A2 110MH for about 2.5 months now.
I am saving to get another 110MH and would like to try out your firmware.
Can you send me the latest version download link?
Also, is there any improvements that I can do to the stock 110MH in areas of power savings and less heat.
The fans are very loud but pushes tons of hot air -- is this worth improving?

Thanks
TMT

[emdje]

Hi Mark,

I have been mining with A2 110MH for about 2.5 months now.
I am saving to get another 110MH and would like to try out your firmware.
Can you send me the latest version download link?
Also, is there any improvements that I can do to the stock 110MH in areas of power savings and less heat.
The fans are very loud but pushes tons of hot air -- is this worth improving?

Thanks
TMT

First post of this topic has a link to the latest version

[MarkAz]

I have been mining with A2 110MH for about 2.5 months now.
I am saving to get another 110MH and would like to try out your firmware.
Can you send me the latest version download link?
Also, is there any improvements that I can do to the stock 110MH in areas of power savings and less heat.
The fans are very loud but pushes tons of hot air -- is this worth improving?

As emdje said (and it's the same for his firmware), the first post on the thread always has the latest version so you can grab it from there.

In terms of improvements in terms of power savings, there really aren't that many that would make a huge difference... The reality is the vast majority of the power goes towards hashing, probably 1000 of the 1100 watts, then the rest is the 5 fans and the Pi takes up almost nothing.

From the noise perspective, you can do some things if temperature isn't an issue for you - meaning you're in a cool environment.  I actually swapped out the fans on one of my 88's with these:

http://amzn.to/1irg7Ua

And it definitely reduces the sounds, but it also reduces the airflow.  The 88's generally run much cooler than the 110's, and it made basically no difference for the temps on the boards.  On a 110, I would imagine it's going to run a bit hotter with these fans, but in a cool environment shouldn't be an issue.  You might also consider seeing if you can jury-rig something to put a 6th fan in the open slot in the back - if you did that, then I wouldn't be surprised if you ended up with a lower temp than the stock fans.  Keep in mind that most of the fans you buy, you'll need to either get an extension or splitter to lengthen the cables, since the farthest fan's stock wire won't reach that far (and the middle fan might not also).

[The_Prof]

I have a query that I would like to check before I do anything. On my stats list of the A2 it has the ID as 0 to 4. Am I correct in assuming it should go up to 5 and one of the boards is either disconnected somehow or dead? Or is there something else I have missed.

You're talking about under the realtime stats, correct?  If so, all of the Terminators I have have 6 blades, 0 through 5.  Now, this isn't to say that InnoSilicon didn't make some 5 blade versions, there seems to be about a million variations of the hardware from them.  I would say this first - assuming you're running my firmware, and running at 1200MHz, on that realtime stats page, what's the SUMMARY MHS av/5s?  Those are what you're actually hashing at, so if you have a 110, and that number is in the 90's, then you have a blade that's down.

Next, look in through the front, if one of the blades is having problems connecting, in most cases the green LED's on the front will flash.  When you reboot the machine, they flash on all the blades, and as they each are initialized they stop flashing.  So if you see one of the blades flashing, then you know which blade is giving you grief.

Then once you know the culprit, I would open the case and make sure all the cables are seated tightly on that blade, both on the blade and back to the controller.  The connectors are super finicky on the A2's, so making sure they're in nice and tight and getting good connections (obviously do all this when it's turned off).

Finally, if that doesn't resolve it, then try swapping the data cable with one of the ones next to it, just swap the part that connects to them - this way if the blade STILL has the issue, you know it's the blade and not something else.

You can buy replacement blades from InnoSilicon - it's a pretty painless process and they aren't overly expensive.

Hope that helps and let me know what you discover.

I forgot I posted here, thanks for the detailed response. It is the blade. Swapped power and com cables. Still flashing the light while the others are mining. It not a major issue, might get another one. BUt thanks again for you input and help.

[QuintLeo]

Is there somewhere to get this other than that mega.nz site?

 That site just loads a logo and then just sits there forever for me, seems to be BADLY broken.

[emdje]

Is there somewhere to get this other than that mega.nz site?

 That site just loads a logo and then just sits there forever for me, seems to be BADLY broken.

Link works for me, have you tried a different browser?

[QuintLeo]

If a reasonably recent update of Firefox isn't good enough for them, THEY have an issue.

 I suspect more that they do broken wierd stuff like Github (that I also have MAJOR issues trying to use) that doesn't use STANDARD download capabilities.

[emdje]

If a reasonably recent update of Firefox isn't good enough for them, THEY have an issue.

 I suspect more that they do broken wierd stuff like Github (that I also have MAJOR issues trying to use) that doesn't use STANDARD download capabilities.

I have firefox and can download the image just fine. This is the link to the newest version from page 1: https://mega.nz/#!3VkjgaJT!PH4ixphRQFT23tgTwpnSwrkJPG0gB2VXsmN9BxagftI

[QuintLeo]

Gotta wonder why you bothered reposting the same link I've already mentioned DOES NOT WORK FOR ME.

[emdje]

Gotta wonder why you bothered reposting the same link I've already mentioned DOES NOT WORK FOR ME.

Chill dude, I am just trying to help.
Have you tried downloading the link with firefox in safe mode?

[mxnsch]

Gotta wonder why you bothered reposting the same link I've already mentioned DOES NOT WORK FOR ME.

The Link work for me, too. Usually, the Browser itself, an extension or AdBlocker is the problem in those cases. As emdje wrote, start Firefox in Safe Mode or switch to Chrome just for this Download.

Here is a version from portableapps you can use without installation:
http://portableapps.com/de/apps/internet/google_chrome_portable

[citronick]

in the Mega site, there is a link just below "download using browser" - which i use to download Mark and emdje image files - no issues.

[ZeroGee]

I'm using this image on my 110, and I decided to do some voltage modification. I figure I'd post this here as well. I edited index.php to get different clock speeds.

So I decided to take a crack at overvolting and overclocking my A2 110. The guide provided earlier in this thread does not cover the V4 boards, and I could not find one on the forums so I decided to post it here.

The A2 110 V4 blade buck controllers are based on Linear Technology's LTC3856. Datasheet. Following Pg 29, the feedback resistors will be connected to FB (pin 2), and dictate the output voltage via the equation Vout=0.6*(1+R_B/R_A). The controllers are set up to use the chip's diffout configuration, but the equation remains the same.

There are 3 different controller layouts on this board:
One side
Other side
End of blade

As we see in the photo, there are two resistors labeled 01C and 66B, 10kΩ and 4.75kΩ respectively.
This yields a stock voltage of 0.6*(1+4.75/10)=0.885V

I figure I want around 0.925V, following the target voltage earlier in this thread.

There are two ways to handle this. We can decrease R_A or increase R_B. I had a lot of 24kΩ and 27kΩ resistors left over from a previous project. It just so happened that if I used one of each value in parallel with each R_A and R_B, 27kΩ||10kΩ=7.297kΩ and 24kΩ||4.75kΩ=3.965kΩ yields 0.926V.

So I soldered a 27kΩ across the resistor marked 01C and a 24kΩ across the resistor marked 66B. I ended up with ~0.933V on average.

If you want to only solder one resistor, R_p in parallel with R_A (01C), it would follow this equation: Vout=0.6*(1+(4750/(10000*R_p/(10000+R_p)))). Solve for R_p given a target Vout.

I've yet to do long term overclock testing. I'm testing 1200MHz stability for now. On average the power consumption is up ~2A to ~15A on the 12V rail. My stock power supply is not able to supply more than 4 blades stock, much less 4 overclocked blades. I just split the load half and half between the stock power supply and another 1kW power supply. I should have some numbers soon.

So far 24 hour testing of 1320MHz has been successful. With 5 blades overvolted and clocked at 1320MHz, and one at stock voltage and 1200MHz, I'm getting an average 110MH/s pool side.

[2x2lcallingcq]

I'm using this image on my 110, and I decided to do some voltage modification. I figure I'd post this here as well. I edited index.php to get different clock speeds.

So I decided to take a crack at overvolting and overclocking my A2 110. The guide provided earlier in this thread does not cover the V4 boards, and I could not find one on the forums so I decided to post it here.

The A2 110 V4 blade buck controllers are based on Linear Technology's LTC3856. Datasheet. Following Pg 29, the feedback resistors will be connected to FB (pin 2), and dictate the output voltage via the equation Vout=0.6*(1+R_B/R_A). The controllers are set up to use the chip's diffout configuration, but the equation remains the same.

There are 3 different controller layouts on this board:
One side
Other side
End of blade

As we see in the photo, there are two resistors labeled 01C and 66B, 10kΩ and 4.75kΩ respectively.
This yields a stock voltage of 0.6*(1+4.75/10)=0.885V

I figure I want around 0.925V, following the target voltage earlier in this thread.

There are two ways to handle this. We can decrease R_A or increase R_B. I had a lot of 24kΩ and 27kΩ resistors left over from a previous project. It just so happened that if I used one of each value in parallel with each R_A and R_B, 27kΩ||10kΩ=7.297kΩ and 24kΩ||4.75kΩ=3.965kΩ yields 0.926V.

So I soldered a 27kΩ across the resistor marked 01C and a 24kΩ across the resistor marked 66B. I ended up with ~0.933V on average.

If you want to only solder one resistor, R_p in parallel with R_A (01C), it would follow this equation: Vout=0.6*(1+(4750/(10000*R_p/(10000+R_p)))). Solve for R_p given a target Vout.

I've yet to do long term overclock testing. I'm testing 1200MHz stability for now. On average the power consumption is up ~2A to ~15A on the 12V rail. My stock power supply is not able to supply more than 4 blades stock, much less 4 overclocked blades. I just split the load half and half between the stock power supply and another 1kW power supply. I should have some numbers soon.

So far 24 hour testing of 1320MHz has been successful. With 5 blades overvolted and clocked at 1320MHz, and one at stock voltage and 1200MHz, I'm getting an average 110MH/s pool side.

You're rocking it!!!

[scrouch]

What model of Raspberry Pi do I need for this image to work? I think I have a bad Pi and need to purchase another.

[ZeroGee]

Raspberry Pi 1 model B revision 2. It's marked with:

(Rpi Logo)
Raspberry Pi
(c)2001.12

Other easy identifiers: Yellow capacitor and 2 through holes by the voltage regulator.

https://en.wikipedia.org/wiki/Raspberry_Pi#Connectors

[QuintLeo]

in the Mega site, there is a link just below "download using browser" - which i use to download Mark and emdje image files - no issues.

 I never see anything about "download using browser" at all.

 It DOES appear that the Mega site needs you to use an extension to work - but I can't even download the extension, SAME issue.

[ZeroGee]

in the Mega site, there is a link just below "download using browser" - which i use to download Mark and emdje image files - no issues.

 I never see anything about "download using browser" at all.

 It DOES appear that the Mega site needs you to use an extension to work - but I can't even download the extension, SAME issue.

There should be...

Just in an attempt to help, I put it on tempsend for a month starting today. AKA This link will expire in a month.

[citronick]

Mark - what is the process of getting a replacement blade from Innosilicon.... I think one of my board just got toasted.
The A2-110MH now just shows 4 ASICS instead of 5
I have disconnect power from that board but after booting up with stock FW, only 3 boards now is ok and one is having problems connecting (flashing green light) - does disconnecting the bad board affected the other boards?
Appreciate any suggestions.