Topic: GridSeed 5-chip USB miner voltage mod - page 5. (Read 156980 times)

I actually just used a very small dab of Arctive Ceramique and let it set.  Once it's been run through a cycle of hot/cold it seems to hold very good.  I have the same Ceramique that has been holding a heatsink on the cpu of a lan card for years, the lan card stands vertical and even without glue the Ceramique holds the heatsink on it.

Heatshrink is a good idea to insulate that hazard waiting to happen, I had thought of placing a heatshrink on the axial resistor (VMOD3) but decided against for now.

As for the fan ripping away, I have stowed them miners away in a ventilated cabinet which takes away some of the noise. Power wise there's not a lot saving from the slower fan, so since the noise factor is reduced, I don't mind them running full speed for now. I was planning to put a 39ohm too after seeing yours but a 1/2w with the heatshrink, but I guess I will let the fan blow full speed until I figure a heatsink for the MOSFET's. I dug out some TO-220 sleek heatsinks, will see if it will work, or will order some like chanbergs (saw some on Amazon sold by cosmos in a bag of 20s).

BTW, did you use adhesive thermal paste or thermal pad to stick the modded heatsinks over the MOSFETs?

I actually went back after I took that picture and slid a little heatshrink over the resistor/wire to insulate it better.  I'd highly recommend doing this if you go that route as it will prevent any accidental shorts against the inside heatsink.  Also, just for reference the resistors I used were mostly 39 Ohms.  I may have had a couple 45Ohm in the mix, but it lowers the fan speed enough that it still blows a good amount of air but isn't as annoyingly loud as the straight 12v.

Indeed chanbergs heatsinks are cool.

nst - you gotta love them Dremel - handy for tinkering! I will dig out my stash. I like what you have done to drop fan speed, very safe!

@chanberg - I like those heatsinks.  If I had some of those I would have used them.  I just used what I had on hand though and used a dremel to cut them to size, and a grinder to remove any uneven-ness.

Here's a couple pics. 

I have no idea where the heatsink on the left came from, but the long thin heatsink actually came from an old motherboard from it's mosfet cooling.  The other two were from old Pentium based northbridge systems (as you can see they've been cut down - they used to be a lot larger and square).

Here's a shot of one like the motherboard mosfet heatsink on the gridseed:

Just checked the rest of my seeds, and it seems only one out of the batch was having an issue with the burned mosfets...

But nonetheless, i ordered a pack of these guys:
http://www.ebay.com/itm/291128184875?_trksid=p2059210.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT


I'm guessing it can't hurt haha

Do share the pics, mucho gracias!! I might also have some lying around and with the help of the Dremel, will fit some on the MOSFET's, which I believe really need the heatsinks with the mod.

BTW, whats the package size? Let me look up on Mouser as well...

Well...I'm not sure of the exact dimensions of the heatsink....I'm sort of a collector of such things since you never know when or what you might need one for.  I've saved them all from past motherboards.  The one I used was an old northbridge heatsink.  I can take some pics and measurements today and post those.  The fins are MAYBE 2-3mm high.  Taller can be used...maybe 4-5mm....so whatever you use in that location will need airflow to really be effective.

What heatsinks fit the mosfets? Can you share that please? It might be useful for lot of folks who have modded their gridseeds.

I have knocked off the thermal pads (bottom and top). Bottom I have used thermal paste and direct contact to the heatsink (insulated the troublesome areas), since anyways the ground-pour in the bottom is connected to the heatsink (via the brass nuts). On the top, I have added copper shims and thermal paste and direct heatsink. The copper shims were 15x15mm which I split into 4 (7.2x7.2) using my trusty Dremel. The heatsink on the top (non fan side) get slightly warm @1200 for the GS.

I think the copper shim works good in the way I have done, since the shims on top ensure a proper contact in the bottom...

I am using a 47k axial resistor mod.

Here's a pic:
http://i62.tinypic.com/14bm5wy.jpg

EDIT: And before anyone tells me about the hazards of the silver thermal paste... this is CoolerMaster ThermalFusion 400 (Electrially Non-conductive, high thermal efficiency paste...) :-)

No pinched fan cables or anything. They have been running fine overclocked for months now since mid march. Out of the 22 only this one has a burned MOSFET. So I'm guessing I had a bad one in the batch. But yeah a little heat sink won't hurt nobody hah Oh and I'm using these with custom cables to an ax1200.

X-cuse me yooze guys,..

You're talkin about our little 5 chip seeds, right? How the F**k can they be blowing up FB's and FET's and Hash chips like that?
I'm just amazed at how many seeds are taking a dump for no good reason.
Have any of you attempted mods different than those specified where known proper and workable alterations go? You know, the ones where the top heat sink is left on with fan running at either full or a bit less than half speed?

My guesses are: inadequate cooling via alterations that have backfired under normal Scrypt overclocking.
The entire card is meant to be FAN FORCE COOLED even though they are not directly sandwich style heat sinked the same way the main hash chips are. I know this for a fact because I recently took apart and dusted off 6 of them as part of converting 5V USB to 5V regulated via 12V source after months of 24/7 cooling at 5V fan speed.

Perhaps it was a batch of bad FET's and or Hash chips or even FB's...It HAPPENS! And it is quite possible we are witnessing one of those coincidences right here on our forum. More than one person in the group got stuck with a faulty seed or seeds from one of those bad batches. There are always a small % of DOA's in most batches, I am sure.

Some yet discovered mistake being made during any of the VMOD's?

Reasons for my amazement:
Scrypt should not cause an over-strain on any components! Period!
After all, they are made to run SHA-256 at 60+W, for Aunt Jamima's sake!
Scrypt at full blast won't even touch that figure if even by half! NEVER! Or something is very wrong with the components.

I've had 6 overclocked seeds running for months now virtually 24/7 with no adverse hardware issues  or firmware issues, whatsoever! Just like it should be. I also have 20 more that have been running faithfully for weeks same same.

Worrying about using more powerful FET's is a waste of time for if you have to, then there is something wrong with your seed/s! Something else on there is backing up current flow where it shoudn't be and causing failures. They are defective! So you're either stuck with a patched up overheating POS now or shyt can them and buy some new ones. Maybe you can swing a discount, that is - if you can get your supplier to feel sorry for you

I believe Hashra intends to offer replacement chips for the seeds at some point. Probably after they start selling more of the latest super duper S, SN, X11 etc. hashers coming out starting now....
Contact support for more info. Tellum Wolfey sent ya.

I'm so lucky I don't have any failed seeds... I'd hate to have to deal with those hassles!
Damn things better work as advertised, or else I'm gonna be pissed 

If that's the case then the heatsinks that will fit in that small location won't do much good at all unless they have airflow.  Time to look into some better cooling options than the small hs I put on them.

I suppose repeated high current spikes and overheat may help damage the mosfet, I think those little LFPAK rated for 76AMP might be possible but to allow continues drain @76AMP the pad need to be able to handle the heat dissipation of SMD mosfets. Which explain why they are overheated and blow. For same reason anything to cool this baby down will help bigtime!!

Wow...no pinched fan cables or anything?  How long has that one been mining up until that happened?  What type of power supply were you using?  What's the ambient temp where it's mining?

Sorry for the questions, but I've got 6 that have the 49.9k/50k mod.  3 have been running non-stop 24x7 for 2 months, the other 3 have been running non-stop for about 3 weeks now.  When I've modded them, I did add some small heatsinks to the mosfets because a little extra cooling never hurts when overclocking anything.

Does anyone know if it is possible to buy the GC3355-Q64 chips themselves? Has anyone replaced one before?

Just regular scrypt mining on a 47k resistor mod.

Think I'm gonna get some small heatsinks and attach it on all my seeds.

The picture above is what I found on google matching my problem but this is what happened to mine

That's what I replaced mine with yes.  It's close to the original specs, but not exactly.  I tried a couple different ones from TI and they didn't work.  The 2Q5A ended up working so I left it with that.


What happened for yours to look like that?  I tried dual mining with a 47k voltmod...won't do that again.

what is the replacement part for this blown mosfet again?



Is it this one? http://www.ti.com/product/CSD17552Q5A

That's too expensive, I need to pay an additional $49.99 shipping since i'm in europe.
I was gonna buy from GAWMiners but they removed the product.
Has anyone used sevengnomes.com? Is it legit?