MinerEU.com 850USD THunderX3 80 USD Blizzard , 4*Blizzard @ 60 USD each - page 20.

JoostB

sr. member

Activity: 391

Merit: 250

Quote from: beanhead on June 23, 2014, 01:21:01 PM

Capacitors have the score marks in the top for a reason, to allow them to blow when shit goes wrong.

Cheap components, cheap capacitors will mean the stress that those components can take is less. That's why your psu is fubar.

Buy a new one and get a decent one.

BTW A2 will take 100 days at 0.002 btc per MH per day to ROI, calculated at an average of 88MH/s. That's not including power costs. These units will never ROI at 10k each if you buy one now. GAW and KNC will have their bad boy units out at the end of September I reckon, even they will struggle to ROI because of all the 80-90MH units being bought currently.

But don't listen to me, I'm just a cynical bastard after pre-ordering a KNC titan and paying 12k after Vat and now KNC have dropped batch 2 by 3k! And will be shipping those just 1month after batch 1, and let's not mention that batch 1 is 250-300MH and batch 2 is 400MH with no hardware change apparently!

Only one winner in mining hardware these days, the hardware manufacturers....

There is some truth in this, i know. But I first have to see whether KnC and GAW will come with a 500Mh/s machine which consumes low power.

GAW doesn't even make miners, it is only a reseller. That guy Josh has a lot of blah blah but is a mere amateur with respect to mining.

JoostB

sr. member

Activity: 391

Merit: 250

I ordered two Meanwell PSU's they are not very cheap, but they fit in the same spot as the current one.

http://www.meanwell.nl/Images/pdf/datasheets/HRP-450-spec.pdf

They claim 88% efficiency for the 12V version.

beanhead

newbie

Activity: 44

Merit: 0

Capacitors have the score marks in the top for a reason, to allow them to blow when shit goes wrong.

Cheap components, cheap capacitors will mean the stress that those components can take is less. That's why your psu is fubar.

Buy a new one and get a decent one.

BTW A2 will take 100 days at 0.002 btc per MH per day to ROI, calculated at an average of 88MH/s. That's not including power costs. These units will never ROI at 10k each if you buy one now. GAW and KNC will have their bad boy units out at the end of September I reckon, even they will struggle to ROI because of all the 80-90MH units being bought currently.

But don't listen to me, I'm just a cynical bastard after pre-ordering a KNC titan and paying 12k after Vat and now KNC have dropped batch 2 by 3k! And will be shipping those just 1month after batch 1, and let's not mention that batch 1 is 250-300MH and batch 2 is 400MH with no hardware change apparently!

Only one winner in mining hardware these days, the hardware manufacturers....

james42

sr. member

Activity: 434

Merit: 250

Yes I used the same fan, on your recommendation. What PSU did you get?

JoostB

sr. member

Activity: 391

Merit: 250

Quote from: james42 on June 23, 2014, 12:17:19 PM

Chips should use 12W each at 1200Mhz, so 96W per blade theoretically. So assuming everything is efficient (which it isn't), 192W for both blades, <10W for Pi, and ~20W for fan. So should be just over 200W really

Maybe the chips take 12W each, but there is also a lot of power lost in the DC-DC converters that are powering the chips.
And I replaced the fan with the Scyte Ultra Kaze which only takes 7W. The current PSU efficiency is maybe 75% or less.
This week I will get the new PSU's which have 87% efficiency, they will consume about 50W less from the wall outlet.
And also important, less heat.

james42

sr. member

Activity: 434

Merit: 250

You were lucky not to have a fire

gyverlb

hero member

Activity: 896

Merit: 1000

Quote from: james42 on June 23, 2014, 11:24:24 AM

Do you have pictures of the defective area?

This is a view of the capacitors, you can't see them whole through the vents but you can see they are all opened instead of having a flat metal surface. The one on the top left shows the material best.

This is the material still clinging on the vents.

This is the projection of the material mostly on the edge of the controller board in front of the vents:

james42

sr. member

Activity: 434

Merit: 250

Chips should use 12W each at 1200Mhz, so 96W per blade theoretically. So assuming everything is efficient (which it isn't), 192W for both blades, <10W for Pi, and ~20W for fan. So should be just over 200W really

JoostB

sr. member

Activity: 391

Merit: 250

Just measured the current through the blades of the A2Mini, for 2 blades it takes 22A. that is 264W.
I wonder where the other 100W goes, maybe dissipated in the PSU, seems like a very bad efficiency of 72.5%.
I assume the controller wont take much.

emdje

hero member

Activity: 686

Merit: 500

I have connected everything directly to 12 volts of a ATX PSU, but be aware because in my case the + and - of the connectors on the blade are reversed from the ones from my PSU.

james42

sr. member

Activity: 434

Merit: 250

Update from Interlink/DPD, they have my A2 Mini in their grubby hands and it'll be with me tomorrow :-) A little faith restored in MinerEU at last

james42

sr. member

Activity: 434

Merit: 250

Do you have pictures of the defective area?

gyverlb

hero member

Activity: 896

Merit: 1000

Quote from: james42 on June 23, 2014, 10:43:58 AM

It's a standard ATX connector as far as I am aware with the same pinouts. So should have both +12V and -12V if it is fully wired, on yellow and blue cables respectively.

The hashing boards are directly wired to the 12V supply. If your controller board has blown (or that little ATX adapter), I would put it down to running the fan through it. It's quite a high current fan, would have been more sensible to hook it straight to the 12V power supply

Currently the PSU is clearly defective: there's no way a PSU can run correctly with 3 large capacitors blown so I can't test much until I swap it (I can only hope it didn't take anything else with it).

james42

sr. member

Activity: 434

Merit: 250

It's a standard ATX connector as far as I am aware with the same pinouts. So should have both +12V and -12V if it is fully wired, on yellow and blue cables respectively.

The hashing boards are directly wired to the 12V supply. If your controller board has blown (or that little ATX adapter), I would put it down to running the fan through it. It's quite a high current fan, would have been more sensible to hook it straight to the 12V power supply

gyverlb

hero member

Activity: 896

Merit: 1000

Quote from: james42 on June 23, 2014, 10:30:49 AM

No the power supply provides only 12V, there is a subsequent transformer actually on the ATX plug. This is what powers the interface board, the fan and the Pi. It could well be that which is blown, at least partially

Are you sure? There are yellow wires for all 12V related components (including the controller board) and a red wire from a different connector on the PSU connected to the controller board (through the plug you speak of IIRC). I assume the red wire is 5V as it's the standard coloring for this voltage (at least on PC equipments) and I don't see the reason why 2 separate 12V lines would be connected to the controller board with 2 different colors.

Edit: maybe there are different a2mini generations with different power supply solutions?

james42

sr. member

Activity: 434

Merit: 250

Kind of like this http://www.mini-box.com/picoPSU-80

james42

sr. member

Activity: 434

Merit: 250

No the power supply provides only 12V, there is a subsequent transformer actually on the ATX plug. This is what powers the interface board, the fan and the Pi. It could well be that which is blown, at least partially

gyverlb

hero member

Activity: 896

Merit: 1000

Quote from: james42 on June 23, 2014, 09:56:38 AM

IIRC correctly the controller board has headers for at least three fans

Yes, I saw at least a second one on close inspection. My initial diagnosis was wrong though: the Raspberry is connected to the controller board and is working fine. But I realized that the Raspberry should use a 5V power line (when it's powered by USB this is what is available to it) and not the 12V used by the fan and blades.
I looked more closely at the power supply and the cause of death is now quite obvious: through the vents one can see 3 blown capacitors and a dust-like material clinging to the vents and projected on the controller board in front of the vents. By the look of it the material is the same as what is protruding from the blown capacitors. Most likely scenario: these capacitors are part of the 12V transformer and the 5V transformer is independent on this PSU.

The PSU replacement seems the only option...

These are run in ideal conditions : 23°C ambient temperature but I fear the two other miners PSUs might not survive (they seem identical)...

JPS2K5

full member

Activity: 266

Merit: 100