[ANN] Technobit HEX8A1 -240 GHs Coincraft A1 board- Finished production

[eestimees]

back in business 

determined that chips do not get juice (power).. when PSU is off, cgminer displays and acts the same.. something must be wrong on the board..

Try setting 220 for frequency & 880mv for voltage

still same

[eestimees]

[Zich]

That's it, both polyfuses leg should soldered to board. Maybe they damaged on shipping.
You need to consult marto then.

[eestimees]

That's it, both polyfuses leg should soldered to board. Maybe they damaged on shipping.
You need to consult marto then.

jep, it's strange all of them are loose/broken.. I'd underestand one or two, but even those that are under the sink...

[Zich]

jep, it's strange all of them are loose/broken.. I'd underestand one or two, but even those that are under the sink...

Yeah,  re-solder it to original pad is impossible since the pad also detach from pcb.

This is what i did  

[eestimees]

jep, it's strange all of them are loose/broken.. I'd underestand one or two, but even those that are under the sink...

Yeah,  re-solder it to original pad is impossible since the pad also detach from pcb.

This is what i did  

I can weld metal, do woodwork, build a house, write programs, run win/linux servers, configure cisco device  but I'm not good at soldering.. considering that I have all pads detach'es.

[iFA88]

I have the usb hub what u recommended, but this issue is not here, is at my friend.
I can not understand this, why is tplink to better than RPI?

Thanks for advice:)
BR:
-iFA88

Hy again!:)
The five hex8 now work nice with 52% CPU use. But when i connect one hexC or hexB then the CPU spin up to 100% and the performance decreases on all board. While this the hex8 board red light pulses.
Now i have two RPI. One for 2xHexB 1xHexC and for the other 5xHex7.
Any idea?
Thanks!

I have RPI A, and when i use 5 Hex8 miner, then the cgminer process goes up to 100% cpu use and therefore decreases the performance. With 4 work it 70% and all is nice.
Is there any solution to config less work to the cgminer?
Thanks!

Sure --set-dif-to-one 0
Check my spelling though it is written in chnagelog. And make sure you are using latest version -you will know it when you see pool reject rate % in screen session

Get a TP-Link TL-3020 it kicks the ass of the RPi.

Red light of hex 8 means it has no work. USB writes are taking too long for some reason. It seems some USB issue. Try with other hub or use the one I have recommended ....
And finally tplink if all fails
However if all is good at two pi's follow golden rule if it works do not touch it

[Gator-hex]

The five hex8 now work nice with 52% CPU use. But when i connect one hexC or hexB then the CPU spin up to 100% and the performance decreases on all board. While this the hex8 board red light pulses.

I can not understand this, why is tplink to better than RPI?

Because I have 2x Hex8A1 + 6x Hex16B = 700GH in mode 1 and the load average is only 27%  

Here's the broken legs on my green-orange light one. They seem to come away from the circuit board with nothing left to solder them back to.

I've noticed in all our pictures it always seem to be the same side. Marto is this a common ground? Is there somewhere else on the board we can solder these legs to?

I bet if Marto made these boards without polyfuses they would run at 0.9W/GH instead of 1.1W/GH.
I've been thinking about snipping mine off and shorting them to increase power efficiency. What do you think?

[marto74]

actualy already made this but with some other changes in components

[foobar31]

What I'm missing here:

cgminer monitor keeps restartin cgminer, and not hashing a thing.. using router MR3020

Did you change "Handle USB detect errors as (Default: hexA)" to Hex8?

Well, I had a very similar issue.

cgminer restarting every ~ 3 minutes.
At first sight, it hashes at good rates (which makes my issue a bit different of yours), whatever the core_speed / voltage I choose, but with a high reject rate (20-30% reported on 2 of the testing pools I use...) => since it restarts every 3 minutes, it appears normal to me to get these stats.

Using MR3020 freshly flashed with the new 0.2.2 firmware, tried a lot of different settings (), without any success.
Also tried a powered USB hub to power both the router & the miner...

I'm convinced that using a MR3020 might be the best and simple solution for most of us but since I have a couple of servers running in the same server cabinet, I decided to try using one of them... Simply compiled cgminer 4.0.0 with the patch provided within the MR3020 firmware archive
Guess what ? It works like a charm ! So far, ~ 240/250 GHs at 240/900 and drawing 301W.

So I don't know what's wrong with the MR3020 solution but at least, I'm sure that the issue is not on the miner side

For your information, and since I asked the question here before, I tried to use an old BFL Single's stock PSU, worked fine at 240/900mV but it looks not as stable as with the Corsair 750W ATX PSU...

[marto74]

Are you powering the hub(if any ) and TP-link from the board it self?
Just power the tp-link with 5 V separately.
It think this is your issue with router and dd-wrt/cgminer setup

[askhimthatsoon]

I got my HEX8A1 to operate this morning, but it was an adventure:

* Tried OZC 550w PSU yesterday, default settings: 260 clock,1000mV => Miner started hashing but as soon as I checked hashrate in terminal (next room) I saw it already disabled itself. Went back to the PSU, HEX8A1 is off. PSU is dead.
* Next try: El cheapo CP 750 Plus PSU. The label says 2*12V rails with 30A each = 720W. I used only the 2 PCI-e power cords, no other load. I tried default settings again. Miner hashes, with very high hashrate (way over 260GHs) but after 20 minutes, PSU switched off. PSU in failsafe mode.
* Couple of additional tries ending with PSU switching off.
* Here is what finally worked (stable for a few hours now): 255 clock, 980mV. Results in fairly stable 260.5GHs. I didn't measure the actual power draw at the wall yet, but I am scared already.

http://thatapp.de/hex8a1.png

[Gator-hex]

I got my HEX8A1 to operate this morning, but it was an adventure:

* Tried OZC 550w PSU yesterday, default settings: 260 clock,1000mV => Miner started hashing but as soon as I checked hashrate in terminal (next room) I saw it already disabled itself. Went back to the PSU, HEX8A1 is off. PSU is dead.
* Next try: El cheapo CP 750 Plus PSU. The label says 2*12V rails with 30A each = 720W. I used only the 2 PCI-e power cords, no other load. I tried default settings again. Miner hashes, with very high hashrate (way over 260GHs) but after 20 minutes, PSU switched off. PSU in failsafe mode.
* Couple of additional tries ending with PSU switching off.
* Here is what finally worked (stable for a few hours now): 255 clock, 980mV. Results in fairly stable 260.5GHs. I didn't measure the actual power draw at the wall yet, but I am scared already.

I had a similar issue, 1 Hex8A1 was ok, but 2 would put PSU into shutdown, even though it was well within power limits of the PSU. All it needed was something on the 5V line of the power supply to fix it. I connected a 12V fan the the red/black wire of a molex connector. But some people use a hard drive/floppy drive/CD or DVD drive. But I find another fan more useful.

[Gator-hex]

actualy already made this but with some other changes in components

Is there anywhere else on the board we can solder these legs to Marto? When they come off they take the circuit board contract with them so they cannot be soldered back to their original place. Zich's solution of soldering to the nearest good leg wont work for someone with Eestimee's problem where all the poly-fuses have a bad left leg. Maybe if you fix Eestimee's board you can take a photo to show us what you did to it? Thanks.

[Zich]

I got my HEX8A1 to operate this morning, but it was an adventure:

* Tried OZC 550w PSU yesterday, default settings: 260 clock,1000mV => Miner started hashing but as soon as I checked hashrate in terminal (next room) I saw it already disabled itself. Went back to the PSU, HEX8A1 is off. PSU is dead.
* Next try: El cheapo CP 750 Plus PSU. The label says 2*12V rails with 30A each = 720W. I used only the 2 PCI-e power cords, no other load. I tried default settings again. Miner hashes, with very high hashrate (way over 260GHs) but after 20 minutes, PSU switched off. PSU in failsafe mode.
* Couple of additional tries ending with PSU switching off.
* Here is what finally worked (stable for a few hours now): 255 clock, 980mV. Results in fairly stable 260.5GHs. I didn't measure the actual power draw at the wall yet, but I am scared already.

Why did you test your board with try & error method?

The PSU clearly had specification about 12V rail, how many rail & what is maximum amp per rail.
I already post the HEX8 consumption on 260/1000 setting, it's 33 A on 12V rail.

You said "El cheapo CP 750 Plus PSU. The label says 2*12V rails with 30A=720W", clearly you see the 720W not the limit 30A per rail.

About OCZ 550W, i don't know which type you use.
OCZ Fatal1ty 550W is 2 rail, max 25 A per rail which clearly will overload


OCZ ZS 550W is single rail, max 38 A & OCZ ZT 550W is single rail, max 45 A should do the job

[Zich]

actualy already made this but with some other changes in components

Is there anywhere else on the board we can solder these legs to Marto? When they come off they take the circuit board contract with them so they cannot be soldered back to their original place. Zich's solution of soldering to the nearest good leg wont work for someone with Eestimee's problem where all the poly-fuses have a bad left leg. Maybe if you fix Eestimee's board you can take a photo to show us what you did? Thanks.

Agree, i would love to see the via  
I already try measure with connectivity but not found any near smd capacitor.

[marto74]

Technobit HEX boards family simple setup guide
http://www.megafileupload.com/en/file/506573/TECHNOBIT-HEX16--Setup-pdf.html

[QuiveringGibbage]

Mini-USB connectors on these board? or standard?

Cheers,
QG

[marto74]

mini on all hex

[askhimthatsoon]

You said "El cheapo CP 750 Plus PSU. The label says 2*12V rails with 30A=720W", clearly you see the 720W not the limit 30A per rail.

About OCZ 550W, i don't know which type you use.
OCZ Fatal1ty 550W is 2 rail, max 25 A per rail which clearly will overload


OCZ ZS 550W is single rail, max 38 A & OCZ ZT 550W is single rail, max 45 A should do the job

You are right, I saw the 720W, because I didn't expect the two PCIe power connectors to be connected to the same voltage regulator, which seems to be the case.

The OCZ was a ZS. May it rest in peace. No clue why that happened.