I didn't send you my boards but the reason I suggested a car battery is because I attempted to hook the board to a 1300 watt bench power supply and it still overloaded it without blowing the mosfet. Car battery did the trick though. DC power is DC power and as long as you don't overvolt the board there's no issue. Considering the choice is between dead board and maybe alive board I figured it was worth the risk.

Sounds good. I'm sure it's fine. Generally there's a fair bit of play in the allowable input voltage and the chip voltage is regulated internally on the board. I don't recall if you did this already but when you attach it back to the terraminer power supply does it still crap out the same way?

Have you attempted swapping the power supplies? Orange light means either an AC issue (problem with power supply itself) or DC issue (usually a short in the board). Read back a couple pages for my posts on this. I provide instructions for how to handle both scenarios. In your case because the board lights up it's probably an AC issue which means you need to replace the faulting power supply.

That's only when charging. Nominal voltage on SLA cells is between 2 and 2.15 volts fully charged. Generally I read around 12.2 on a battery that's been sitting for a few hours. My instructions say to always test voltage first and don't do it while the car is running because then you are correct. Charging voltage is usually around 13.5.

also the sizzling is a sign of a bad connection. If the connector sparks and you make contact again in the same location the connection usually isn't as good which can lead to the sizzling you heard. If you hook back up to the regular power supply now what happens?

definitely something wrong with that board. Has it ever worked properly or was it that way since you received it? If it's always been that way I would say try switching the polarity. I can't say from experience but I have heard reports of these shipping with reverse polarity and flipping the cables helped. D45 light means you have a dead chip but all LEDs should only flash on for a second and then begin cascading.

Carefully check the board again for bits of solder sticking out where they shouldn't. When mine blew it melted out the bottom and oozed out from under the surface mount mosfet. It doesn't sound like you had the same problem but it's worth checking anyway. If you have time you can try testing what components you can with a multimeter. You can get a good idea of the status of the inductors, resistors, mosfets and capacitors without removing them. Just look for a reading that's significantly off from the others. Then replace or remove the faulty component and try again. It's not a sure fire test but it's way easier than breaking out the rework station.

Besides that I'm running out of ideas. It's likely there is a serious issue with that board that's beyond my ability to troubleshoot remotely or even at all.

OK. Sorry about that. If you are getting 12V at the input connection then swapping board location probably wont help you. It sounds like an issue with the board. Arctic silver 5 is non conductive but it also sucks. Noctua is much better IMO. Do any of the status LEDs light up at all?

I wouldn't expect any improvement. CGMiner uses a custom driver for Cointerra devices and considering they went under I doubt there have been any improvements since the latest firmware release. Furthermore the problem was never really with CGMiner in the first place but rather the trashy firmware on the terraminer. If yours are running fine, pray and don't look to hard at them. That reliability can change at the drop of a hat. 

It sounds like your problem may be different if you are reading 12V at the big connectors. Do you have any orange lights on your power supply?

To answer your question though the reason you need to use a car battery (or a monster bench power supply) is because you have a short in the mosfet (source to drain is most common). This short causes the board to draw too many amps from the power supply and the overload protection kicks in. I believe the terraminer has a 1100 watt power supply so if we assume that one channel is drawing all 1100 watts then that's roughly 91 amps. If you are going to blow out the short you need to supply more than that. It's possible a supernova 1300 could supply enough but the easiest and cheapest thing to do is just get some #12 primary wire and hook it to a car battery. Most car batteries are capable of supplying in excess of 600 amps at 12V plus there's no overload protection so even if you draw all 600 for a few seconds it's not going to kick off. That will be more than enough time to overheat the shorted mosfet.

Now on to what I think your problem is. I suggest swapping power supplies to rule that out as a possibility. If you take the retaining screws out from the back of the terraminer and depress the release clip they slide right out. Just switch the 2 you have and see what happens. If it follows the power supply then that's bad. If not then you have a board problem. I still suggest trying the car battery approach to rule that out as well but it's likely something else failing on the board. Usually catastrophic failures can be spotted visually. Remove the board and carefully inspect all hardware. It's also possible that the board is not detecting the water pumps. You can try swapping the pump cables from the board that works just to see if it will run. DO NOT start mining like this because you will overheat it almost instantly but it will at least help you narrow down the issue.

Let me know how all that works and we'll move on from there.

Glad the new thermal compound worked out for you. As for your totally dead board, power it up on a car battery using my instructions a couple pages back. One big spark, one pop and a few seconds later and you will see the board light up save for one dead core. If you want you can then replace the bad mosfet otherwise just run it on 7 cores. You won't get the orange light anymore after you short it. Just make sure you get the polarity correct.

@Cefalu

There is a guy on ebay who has some spare parts for these at reasonable prices. His ebay user is express_computer. Last I heard he had a complete set of working radiators and pumps.

I would swap the pumps around on your units first to make sure the problem follows the pump otherwise you may have a board issue or controller problem. If it follows the pump though you can just swap the unit and you're back off to the races.

You can also use regular water cooling equipment. I believe I read the dimensions of the water block screws match that of an Intel LGA1156. Most store bought water cooling setups will have a bracket to match that. You will probably need to come up with your own replacement screws though as the length required may be different. I recommend spring loaded screws. That way you get nice even pressure and that takes most of the guess work out of it. Also if you have custom loop water cooling equipment laying around those water blocks generally fit without much fuss. I started setting mine up that way before I sold but it was going to be so expensive I gave it up.

Incidentally the machines I sold are running great for the guy. He's happy and I'm happy they're gone. I get a little OCD when I can't make things work right so they had to go.

@FuryFever

If you have a board not powering on please read back a page to my post about shorted boards. If it's not coming on your probably have a short that you will need to 'blow out' before you can start the rest of the board. It's strange though that you have 2 green lights on the power supply so that board may actually be cooked.

Arctic silver thermal compound sucks. The burn-in period is 200 hours so it's really not good for a 100% continuous utilization application. If you are going to use an ordinary non-conductive thermal compound I recommend Noctua nt-h1. Also use TONS of it when applying. You want to fill all the air pockets in the spacer ring and lastly wiggle the water block down to make sure you get any bubbles out. Each of those steps should yield about 10-15C decrease in temps which combined will leave you with a decent working miner.

If you have more questions let me know. I'll try to help.

Lastly antminer is the way to go. I got my terraminers running well and sold them to buy 2 antminer S4s.

Upon further investigation it appears that pasting everything inside of the spacer ring has the potential to increase cooling performance. The air that is normally trapped inside the ring acts as an insulator. Thermal paste heat transfer is somewhere between 200 and 400 times better than air so it stands to reason that you would want to minimize the amount of air trapped by the water block. Tomorrow I will attempt to fill all air gaps and also wiggle the water block down rather than pushing it straight down. This should squeeze more of the paste out from the top of the chips while still allowing me to fill in the surrounding areas.

Fingers crossed. More updates to follow.

Another update on the cooling situation. If you read my post from last night you will know that LiquidPro worked on one machine but not the other. I was very particular about how I applied it and I'm confident I was getting good contact where it was needed. The strange thing is on the machine that it worked well on I actually only coated one chip. Despite that fact, BOTH chips were significantly cooler after running the equipment and both are STILL running cool. (This furthers my suspicion of how accurate the temp sensors actually are) Keep in mind I only touched one chip/heatsink. No arguments from me though. I'll take it. Until I get a rework station and replace the bad mosfets I am happy with how this machine is running.

Now for the bane of my existence machine. I attempted LiquidPro and was not happy with the results at all. At best it yielded no improvement and at worst it ran hotter than before. So I put it back together with the Noctua and it ran pretty much back where it was. On the hot side but still under 100C and still running reliably.

This afternoon a thought occurred to me. Every video I have seen of the paste being redone on these machines shows a TON of thermal paste on the chip from the factory. So far I was just using little gobs on each core like I would for a GPU or CPU. I decided to apply the paste at the volume that I saw in the videos. I figure there are still many of these machines out there that are still operating at good temps in factory form so there must be something to it. I applied so much paste that it covered the sides and even the channels between the chips. I still don't think I used as much as the factory but it was easily 3X what I was using before. I turned on the machine and immediately noticed about a 15C improvement. One of the chips is approaching about 20C improvement after running for a couple hours.

I believe this tells me that the top surface of the chip is not the only area that needs thermal conductivity and adding heat transfer to the sides and channels between does yield additional cooling benefits. Basically I believe anything that has the potential to draw heat away from the chips and increase the surface area contacted by the heatsink will yield positive results. I am going to run the unit like this for 24 hours and if the results are still encouraging I may apply more paste between the chips and to the spacer ring around them. I will post my results if/when I perform the test.

I tried this but I don't see any stats on the temps. Is there something special I need to do? All I see is the hash rate and accepted/rejected.

I got liquidpro last night and tried it out. It worked great on one of the machines. I'm able to operate at full power and good temps. The other machine it didn't help at all. I may try to reapply or grind down the spacer more. I'm guessing the spacer is still holding me too far away. Its back on the noctua for now.

Do any of you know where I can get firmware and what is a good version? I would like to rule that out as a possibility for the strange behavior of the one miner.

Thanks. I'll try that out. I don't think it's hitting 120 though because it seems to top out around 95 as best I can see. It could be the UI isn't keeping up though. I really don't like the cointerra UI. I have a couple pi's and a failing controller now so I may swap the failing beaglebone with a pi and see what happens.

With regard to the spacer I ground mine down to even but not below. I checked the pattern of the thermal paste after doing so and there was a significant difference in the contact pattern of the block against the chips. Before I had visible high spots in the paste but after grinding the spacer things were much more even. After running for a couple days I think this change yielded 10-15C improvement.

I am planning on either removing it with a razor or grinding it down further the next time I have them apart (probably this weekend). I will post results for everyone with pics.

So far with the changes I have made one board has dropped an average of 40C and the other 60C and that's on noctua thermal paste. I still have some work to do because I am only running at power level 6 and cannot run higher at this time without the boards exceeding 80C.

Definitely not. I have clear lines so I would see any air. Plus I purged it all when I re-ran the lines on that board.

I'm going to take a straightedge and my feeler gauges to it as soon as the liquidpro comes in. I'm willing to bet the chips don't sit straight so sometimes I get lucky and they cool properly but most of the time they do not. That's probably why LiquidPro works so well for people. Until then I will deal with it because I'm getting tired of taking it apart every night. I will post my updated results as soon as my shipment arrives.

Well I was all excited to get the now famously running terraminer off of the extension cords and into a more permanent location tonight but upon relocating the unit a couple chips decided they wanted to be hot again. I'm contemplating the possibility of faulty temp sensors. I was gentle when moving the unit and most of the cores are still reading great temps. It's not out of line but definitely hotter than I would like. I have the unit in a (near) freezing cold environment so it should hold until the LiquidPro arrives later this week (hopefully). Another interesting thing I have noticed is that the temps start off really high and then settle down significantly lower after running for a few minutes. Do the rest of you experience this behavior when mining starts?

Another interesting thing, the machine with the resurrected battery-shorted board was running 20C hotter last night than it is running today. I'm not complaining but I really wish I could understand the wild fluctuations in cooling performance (or supposed cooling performance if the sensors can be trusted)

More to follow after the LiquidPro treatment over the weekend.

Another update. I fired up the other board in the unit and I am now hashing at about 1.3TH total and my hottest chip is 66C. The fans are barely running and the errors are well within range. I'm calling this a win. If the unit runs stable overnight I will move it to a more permanent location and let it eat.

I think what happens is the coolant in the system degrades and little bits of solids hinder the pump performance. I initially drained the system into a cup and there flakes of solids in there. Small ones but solids just the same. We will see how much temps change after the ethylene glycol. I also expect cooling performance to improve after the thermal paste has time to burn in. I have already seen significant improvement in just this short time.