Topic: Avalon Clone Assembly Service Europe - page 2. (Read 20432 times)

A closer look reveals just one affected board.  It looks very blown.  All 10 of the Avalon chips have holes in their packages.

As for the other boards, I got rid of the big screws and they're all mining again now, including the "#2" board that I couldn't get to work before.

(I installed them onto connectors 1-7 leaving #8 empty and set MinerCount = 31.)

I can send the one board back, but I'm sure the Avalon chips are no good.  It would only be useful for failure analysis, or putting new Avalon chips onto.

I do have 8 unsoldered chips left that I could send with it.  Wish I had 10...

Any value above a few ohms is about a good sign.
Double check that there are no short circuits:
1 - between the leads of the capacitor C1 (1.2V) should be 56-64ohms
2 - between the leads of the capacitor C5 (12V) a few Mohm
3 - pin 1 and 2 of the lower row black socket (3.3V) a few Mohm
And it's important test a short circuit between 1.2V and 3.3V. The easiest way to check this by measuring first pin on black socket and + on C1
If no shorts, are sufficient to wash the PCB with isopropyl alcohol (IPO) preferably in an ultrasonic cleaner. For cleaning you can use a toothbrush or brush with hard bristles.

Measuring with a meter on the 20Mohm range, it does not appear to be conductive.  Neither your silver compound, nor the white stuff I added.

I think it's the screw head thing.  This image (of the #2 board) shows how the one screw head cut through the insulating layer on the top side of the PCB and made contract with the top plane:

https://i.imgur.com/fHBsNqo.jpg

I measure about 6 MOhm on one unit and about 8 MOhm on another unit. What's the expected value?

Are you going to have own service at some point ?

Bogart: Can you test with a multimeter if the compound is conducting? I don't see any screw heads in contact with anything else in your pictures.

That's good support.  I appreciate that.

I'd like to move forward with diagnosing what's left and finding out what still works, but I'd really like to know the cause of the failure first so I don't repeat it.

Do you think it was the excess thermal compound conducting where it shouldn't, the large screw heads conducting where they shouldn't, or something else?

crazyearner: I do assembly service, I assemble other peoples chips, not my own chips. I don't think I have time to design one of those liquid cooled units.

I'm planning on keeping a small stock of spare parts, there seems to be some interest for this, and I need to be able to repair customer's units as well...

Local pick up is perfectly ok.

Bogart: This is in large part my mistake, my assembly process has not been very good, only the last few units I have assembled have been done properly, I was in too much of a hurry to get units out the door fast. If you send the broken units back I will refurbish them, I think most of the chips are still ok! I hope I can compensate you in some way.

As I told you in the messages please confirm that I want to buy 1 module unit

Tell me the cheapest shipping through pm and how much % of the price you want to secure the order in btc, or as it may be a small amount this first payment you can accept it through paypal and then I pay the rest in btc

If you got any chips remaining please tell me price 

Measure the resistance between 12V IN and 1.2V on transistor Q1 http://imgur.com/7sacImC

Hmm.  I did add some screws where they were not present, and my screws did have larger heads.  I made sure they didn't make contact with any components, but I didn't think that they might penetrate the insulating layer and make contact with the PCB copper.

That may have been what happened.

Hi stormbom. Nice project you have I have a few questions.

1 Will you be doing any other cooling designs ie liquid cooled units or something like this.


2 Will you be providing full units ready to hash with complete design?

3 Will you be providing units that can buy separate parts like spare parts and or modulus to add to current units?

4 Will you be providing bigger units than 60 to 80GH units.

5 Will you be doing buy now or will you be doing pre order.

6 Where will you be shipping to and what estimated times do you think will take to arrive.

7 Will you be providing local pick up option and do order part payment to secure unit or units and then pay rest on arrival?

8 What payments will you be accepting?

Thx

In my opinion thermal  compound do nothing here unless you have used compound  with silver addition.
I think that you are use bolt with to big head. In effect short 1.2V from top layer to ground GND (small ring around bolt hole coming from bottom layer)  Usually the inverter switch off if short  detected. In this case, could be puncture inside Q1 MOSFET and chips were cooked by 12V. 

LukeStokes also had a disaster.  He documented it fairly well, and ended up selling his Avalon.

Well I don't know what to do now.  I think I've ruined the module.

I started troublshooting yesterday and took all the boards off of the heatsink, and found that the thermal compound didn't cover a lot of the PCB surface area behind the chips.

I decided to add compound to all the boards to fix that.  I think my mistake was in using cheap white "Stars" brand compound, and in assuming that it was nonconductive.

I also used too much, and some got forced through the vias and ended up on the topside of the boards.  Especially badly on board #2.

After all this the #5 board was working again, but now #2 (the one with the most excess goop) wouldn't work.  I cleaned a lot of excess compound off with isopropyl alcohol swab pads, but a bunch got up under the QFN48 chips, and I wasn't able to make #2 work again.

I thought I would run the module with the remaining boards overnight to see if I had any trouble with #5 again.  Well this morning the machine was reporting zero hashes and the fans weren't spinning.  I turned the PS switch off and then on again, and right away I heard arcing and saw yellow sparks/fire from inside.  I turned it back off and took the new module out, and found places where apparently the excess compound had formed a conductive path and caused arcing and carbon tracks.

The Avalon still runs now with just the original 3 modules, thank "bob".

Pics: http://imgur.com/a/K1r2M

I don't know what to do now.  I can't really ask strombom to fix my mistake.  I imagine some of the chips and some whole boards are still good, but I don't know how to clean this junk off of them...

Did you study the Avalon schematics?  If you did, do you know how to number their modules?

ProfMac: I'm not sure I understand the question?

Yes, and that is good.  However, I have an Avalon.  I just hoped that you might have comprehended the Avalon schematics during your travels.

ProfMac: There are 4 modules with 8 units each. Module 1 has unit 1-8, module 2 has unit 9-16 etc, module 3 unit 17-24, module 4 unit 25-32. So module 29 would be the 5th unit in module 4 if I'm not mistaken? There is numbering on the silkscreen of the back planes, so it's easy to see.