There is a thick copper "Ground Plane" and the small holes are called "vias" the ASIC chips are designed to pass 80% of their heat (watt dissipation) thru the board to the large sink. CPU and GPU Dissipate thru the Tplate or Top Plate covering the silicon.
Those tiny copper H/S are failsafe, and do 20% of the work.
Well, it may work with other kinds of devices, but it doesn't appear to be working here! |
|
|
|
The design isn't ideal as there are multiple points of failure, especially those tiny heatsyncs on the bottom (shown in the picture) - it wouldn't take much time with a fan down, for those to get incredibly hot and burn up the chip.
How are the copper sinks mounted to the board? I dont see any clips or fasteners.
I don't see any fasteners, the boards are positioned vertically so if you use certain kinds of thermal paste, the paste starts become liquid when heated, so what could happen is these heat sinks just fall off! That is why CPU have screws to fasten down heat sinks! You need constant pressure. |
|
|
|
One of the conversations I recall having with Josh in the parking lot of this office is that he was having trouble finding the right thermal paste to work, thus being aware that an issue was at-hand.
In the CPU space you can easily get failures if you don't put the right amount of thermal paste and have it securely attached using screws with springs. For something that is supposed to run 24 hours 7 days a week, these tiny heat sinks that are secured just using thermal paste doesn't look like it'll work. Every high end design like CPUs and GPUs have the heat sink attached from the top of the package and never via some conduction on the PCB as some other claim to work. |
|
|
|
 The rig AMT shipped to me. Notice how the left and right side are all bent out of shape. Notice the power connecter on the right side is completely gone. Notice that the some of the boards looked warped out of shape. Notice the right most heat sink in bent out of shape. Notice the white plastic on the right.. that's the stripped out power connector. Notice on the second to the left board, there's this black piece.... that a piece of the disintegrated fan. Notice the paper on the right, that's what AMT considers as padding. Well it didn't help avoid having the rightmost heat sink get bent out of shape! UNBELIEVABLE! Tip to AMT, use bubble wrap first around the entire unit, then pack in foam peanuts or styrofoam on all eight corners. Paper is not padding!! |
|
|
|
This is pic of 1 of my boards that doesnt run. you can see the bent copper heat sinks and paste all over.  I don't understand the heat sink configuration here though? Why are the heat-sinks on the A1 chips so small? What is the bigger heat sink on the other side connected to? Because 30 % of the heat can be disposed from one side of the chip 70% from the other This seems to be done right CONGRATS. AMT  This is useless of course as all my other posts right? CPU and GPUs have their heat sinks on the side of the chip and there is a conduction path from chip thermal paste to heat sink. I don't understand this kind of a configuration, there is no conduction path from the chip through the board to the massive heat sinks. |
|
|
|
This is pic of 1 of my boards that doesnt run. you can see the bent copper heat sinks and paste all over. Damn dude. Those are bent to shit. I only have one little prong bet. It looks like someone was kicking your box down the road. Now hopefully the cable is the only issue with mine. once I can finally check the information about the chips and see if they are working. If you look carefully its the compound. That happened well before it left the facility for shipping. Its all over the leads and contacts. Thats really really bad. Could be shorting the board. Bent heats-ink pins/fins while bad is not as bad and can still result in a working board (presuming the heat-sink didn't tear the chip from the board itself slightly in transit. Again a result of shit packing Strange that the heat sinks are bent. I am not sure what kind of material is used to paste the heat sinks onto the A1. From my limited knowledge, typically conductive paste heats up and starts flowing, so there is usually something to bolt down the heat sink. In this case, there's nothing bolting it down. Now I notice that the paste is white, usually this kind of paste is the slightly sticky kind and doesn't flow when it gets heated up. Still, I don't know why the heat sinks are so small. |
|
|
|
This is pic of 1 of my boards that doesnt run. you can see the bent copper heat sinks and paste all over. I don't understand the heat sink configuration here though? Why are the heat-sinks on the A1 chips so small? What is the bigger heat sink on the other side connected to? |
|
|
|
More PowerSupply Rail info, if you do have one of these multi rail jobbers you can puzzle it out. Corsair HX series single rail does not shut down from ripple, it is way to heavily built for that, you can weld with them. $$$ also. ========== Quote from: Hark33 on April 12, 2014, 08:26:23 AM Hello, I have technical problem with 1 TH/s Desk. I received 2 1 TH/s Desks yesterday. I'm using 2 1600W Lepa power supply. Miners firstly turn off after 5 minutes, I saw, that 1 original fan was set to air flow in, one to air flow out. A set both fans to IN. So now it working for approximately 30 minutes, then one temperature sensor reach 50 degrees - miner shut down. I have miners in room, where is 21 degrees. So I add 2 fans at front side, now it working for 1 hour, but its not a lot. So, where is a problem? I must put it to fridge? --- This is the same issue we observed with the Platimax 1500W: some PCIe connectors share the same rail, with that you are feeding two hashing units over the same rail and overloading it - the Platimax uses to shut down after a while. Since the Lepa's PCIe pinout looks similar to that one, here is the correct way to set up your CCD with the chosen PSU: 1) take a look at bottom of datasheet page 2 ( http://www.lepatek.com/files/LE_ProductBasic_eng/PB_File/G1600-EN.pdf) 2) notice that the distribution of the 6 12V rails is as follows: 12V1:MB, 12V2:MB, 12V3:PCIe1+4, 12V4:PCIe2, 12V5:PCIe5+6, 12V6:PCIe3 3) ensure that you connect 4 hashing units to PCIe connectors 1, 2 , 3, and 6 (left 4 and 5 unconnected) 4) buy or build yourself an adapter from MB-20 (12V1) or MB-16 (12V2) connector to PCIe and feed the 5th hashing unit with that Bottom line: ensure to feed only one hashing unit through one 12V rail. Even if the PSU uses a common rail, powering two hashing units through a single rail will either cause voltage ripples or over-current protection shutting down the PSU. Good luck Thanks a lot Zefir, you have right! I did it exactly how you said, I connected 4 hashing units to separate PCI-E rails and then I modified CPU power supply connector to PCI-E connector, now working stable! Thank you very much again Thanks... may be very important info. |
|
|
|
The AMT administrative office is not open on Saturdays or Sundays so that would not be possible. Surely someone will get back to you on Monday regarding your request for pickup.
We're still working and shipping out over the weekend
Yes, you can handle pickups. You owe it to your customers, since you are horrendously late delivering their miners. Well you owe to your customers, not only are you ridiculously late, you keep shipping damaged merchandise! |
|
|
|
AllCrypt now lists IXC https://www.allcrypt.com/market?id=531This is an exchange based out of Baltimore, Maryland. The founders seem pretty legit. Met them at the crypto currency convention. Please move your Vircurex trades over to AllCrypt. |
|
|
|
1.2 miner DIY Kit has arrived and the packaging is awful. The chassis is bent and some of the cards don't have screws holding them in place.
2/5 cards have 2 screws
2/5 cards have 1 screw
1/5 cards have no screws
and no instructions on how to get this thing up and going.
Good job amt. mine was bent too. It's looking less and less like a mistake and more like a cool new feature! Well consider yourself lucky that it is only bent! I have two boards with the power connectors completely torn off! Going to have to figure out how to get this replaced. |
|
|
|
You are going the kit route right?
If so let me know if you get any documentation with it for set up. Also if it has any issues locating hashing modules. mine seems to not be finding them where they are looking for them and as I can't kill cgminer I'm not sure it's even launching. I can kill python though.
Yea I went the kit route. I will make a point of asking for documentation on the device. If anything I will make a PDF and post it (with their permission). Might help them some too. And it doesn't take me much time to do that as I would be doing it anyway. This is B.S.! Even some rudimentary manual or some email or post here on how to hook this up should be made available! This is like amateur hour in every aspect. |
|
|
|
Oh this is not a good thing. Damn for all that trouble I may as well go pick mine up. Im close enough that I can. We welcome pickups, if you'd like we can change your status to pickup and we can schedule a time to pick it up. If you welcome pick ups please email me with a reply on this. I sent 2 separate emails requesting this. I can go today just about any time really since you are shipping out miners anyway. Honestly it would be simpler to just pick these things up and have them arrive intact. As I am doing a review on it would not be a good thing for you to have the unboxing show these things in pieces. As I did arrange originally to pick up my order anyway, this is not anything lost here. But I would like the courtesy of a reply at this point to make the arrangements for pick up. If had know how poorly they pack these things, I would have picked up the miner myself! |
|
|
|
AMT,
I have not received new backplanes from you. You said they shipped to me, However; I doubt they would take over 5 days to get here, much less 2 days. What is the status on this?
STATUS UPDATE on mining: At my facility, I have the 1500W plugged into a singular 20A circuit. However, even with my patchwork backplane, it is still only allowing 1-2 CARDS to be detected at 200-300 GH/s. The others show 0 GH/s.
Nearly two weeks INOP now. Much less, I cannot get both miners to work on separate (isolated) 20 A circuits. What is my next step?
~Casey
BTW, do you know if all the cards work in a 2 card configuration? I got my miner yesterday and it has 3 cards that are physically damaged from the poor packaging. The two cards on both ends have their power supply input modules damaged. Also, do you have a wiring diagram? What gets hooked where? |
|
|
|
It wasn't even a box, it was cardboard taped together.
Yes it seems fedex did a number on that one. We'll go ahead and schedule a pickup to RMA that miner and have another back to you as soon as we receive it. This is the first instance of a 1.2Th arriving like that, surprised the foam protection on the sides didn't keep it safe like all the other miners shipped. We'll arrange for an RMA pickup for Monday and have a new miner back to you in a jiffy. Thanks. I have a similar damage package. Not only is the package damaged by the components are damaged. One fan completely destroyed. Metal rack all bent up. Two boards appear to have their power connectors destroyed. One board appears to be bent. Am I supposed to send this all back? Why did you folks not ship this unit in a proper package? This is so upsetting. I wait for a unit that is delayed for over a month and I get this shoddy packaging! I can't believe they skimped on packaging for something that cost $6,000. This is ridiculous. I've shipped lots of hardware and you need to have a lot of foam and bubble wrap. AMT shipped the unit with ZERO bubble wrap or foam, they just had paper to fill up some space. Otherwise, there was zero padding. Now I get a damaged system! It is so dumb of them to cut corners on the packaging! No paper in mine just bubble wrap. I can't get it up and running and I'm starting to think the components might have been damaged during shipping. The heavy cage wa son top of the pie and main board as well as the back plane. 2 of the 5 cards look undamaged, so I'll try to see if I can get a system working with those. Meanwhile, I want to send back just the damage cards and get something returned immediately. The cage is bent up, but I think I can hammer it together to partially straighten the out. Also, one fan totally got disintegrated, need a replacement for that too. |
|
|
|
It wasn't even a box, it was cardboard taped together.
Yes it seems fedex did a number on that one. We'll go ahead and schedule a pickup to RMA that miner and have another back to you as soon as we receive it. This is the first instance of a 1.2Th arriving like that, surprised the foam protection on the sides didn't keep it safe like all the other miners shipped. We'll arrange for an RMA pickup for Monday and have a new miner back to you in a jiffy. Thanks. I have a similar damage package. Not only is the package damaged by the components are damaged. One fan completely destroyed. Metal rack all bent up. Two boards appear to have their power connectors destroyed. One board appears to be bent. Am I supposed to send this all back? Why did you folks not ship this unit in a proper package? This is so upsetting. I wait for a unit that is delayed for over a month and I get this shoddy packaging! I can't believe they skimped on packaging for something that cost $6,000. This is ridiculous. I've shipped lots of hardware and you need to have a lot of foam and bubble wrap. AMT shipped the unit with ZERO bubble wrap or foam, they just had paper to fill up some space. Otherwise, there was zero padding. Now I get a damaged system! It is so dumb of them to cut corners on the packaging! The miner I got from Josh and shipped via UPS also didn't make it intact, with the box damaged (just learnt of it now). I even stood over the gal packaging it to make sure it was as secure as possible for a 54 lb device. Luckily, I insured it, thus the buyer WILL be made whole, albeit I'm not sure how long that process takes. I feel I'm covered since they packaged it, for that's one aspect of what they do - a package/shipping outfit that uses UPS. I shipped before a 100lb server and it got lost by UPS. Anyway, I filled a claim and got compensation for the lost package. I don't think the buyer gets paid, its the shipper that gets paid and the buyer obviously gets a refund by the shipper. Anyway, hope AMT learns their lesson from this. There is a reason why they invented bubble wrap and styrofoam....USE IT!!! |
|
|
|
It wasn't even a box, it was cardboard taped together.
Same here, I was shipped units in a cage, no case. The box looked like a makeshift box pieced together from other boxes. |
|
|
|
It wasn't even a box, it was cardboard taped together.
Yes it seems fedex did a number on that one. We'll go ahead and schedule a pickup to RMA that miner and have another back to you as soon as we receive it. This is the first instance of a 1.2Th arriving like that, surprised the foam protection on the sides didn't keep it safe like all the other miners shipped. We'll arrange for an RMA pickup for Monday and have a new miner back to you in a jiffy. Thanks. I have a similar damage package. Not only is the package damaged by the components are damaged. One fan completely destroyed. Metal rack all bent up. Two boards appear to have their power connectors destroyed. One board appears to be bent. Am I supposed to send this all back? Why did you folks not ship this unit in a proper package? This is so upsetting. I wait for a unit that is delayed for over a month and I get this shoddy packaging! I can't believe they skimped on packaging for something that cost $6,000. This is ridiculous. I've shipped lots of hardware and you need to have a lot of foam and bubble wrap. AMT shipped the unit with ZERO bubble wrap or foam, they just had paper to fill up some space. Otherwise, there was zero padding. Now I get a damaged system! It is so dumb of them to cut corners on the packaging! |
|
|
|
|
Received shipment.
Damaged box. Not a lot of padding.
3 of five boards physically damaged.
Shoddy shipment!
|
|
|
|
That is exactly what I said do not knowing Jim is a real person.
You can use google to translate it if you wish. And I clearly said that I do hope he gets better in most sincere way
However for me some things just do not go buy so I will say again and agin what amt are doing because they will continue to do it in the feature. Someday you can go back 200 pages and found my post staying here
Basically you have two options to get rid of me
1. Ask mods to ban me permanently
2. Use ignore button
3 both one and two
1.
Agreed, And I have done both. Likley just remove your posts but I will just put you on ignore. I had you on it until recently. Clearly you are still just as useless as before. +1 Agree... utterly useless. |
|
|
|
|
Show Posts
