Topic: GridSeed 5-chip USB miner voltage mod - page 57. (Read 156991 times)

artic cooling mx4 - use it for my cpu gpu basically everything, also comes in 20g good if youve got a lot to sipply thermal paste to. its cermic thermal compound so non conductive. probably one of the best, if no the best cermic thermal compund youll get. dering

EDIT : and yes maybve a bit overkill for the grim idseed chips , if it where only thoose BUT there is also the voltage regulators at overclocking with be pushing more through thoose and they can get pretty hot. WE need to consider keeping all these parts cool not just the chips - appears they dont get too hot themselves, but the surrending parts can - im assuming i not taken any temperture measures or touched them. anyone got an IR thermo they can check the temps on the parts??

@wolfey2014

Did you forget to say 'Amen'?

This isn't even worth quoting much less commenting on! But, before anyone goes throwing their miner's out the window, here's my take on the above.

*Fact: Most folks pay .10 per KWh, not .20!
*Fact: I pay NOTHING extra for my added KW usage! So my extra hashes ARE indeed extra $profitable!$
*Fact: Bitcoin difficulty has not been halving 2X per month!
*Fact: On average, difficulty has been incrementing at ONLY around 2.3% on average! (Jan '14 - to current)
*Postulate: Second generation ASICs coming out in Q2/Q3 will have a significant impact on the value of our beloved Gridseeds by causing their price per unit to drop significantly meaning their actual profit value will increase several times which will make our mods and modded units even more profitable! That is, it will make our investment in modifications even more valuable than it is right now!
*Fact: There is plenty of time to recoup the investment in over-volting / overclocking due to the fact that ROI is - in most cases - only a matter of days i.e. less than 1 month!

I feel sorry for anyone having to pay .20 per KWh and I know some in other countries are paying upwards of .75 per KWh! Wow! One can hardly compete much less profit under conditions like that!

So, do not let the merchants of chaos and fear mongers / market manipulators talk you out of your due profits and profitability!
The modification IS profitable. The modification IS harmless. And for most folks, the modification will always be an improvement over stock! Always!

Oh and one other point. If you're going to mine crypto-coins, go at it with long term investment in mind, years, not short term and certainly not on a quarterly basis. Bitcoin is not going anywhere. It's here to stay. So is the entire crypto-currency market! It's here to stay! Don't let short sightedness hinder or stop you from progressing, from risking and especially from WINNING PROFIT$!

Did some thinking about this, and think there is some room for interpretation of the results.
Some points of interest,
Im paying only $0.10/KWH where I live (Arizona). I know rates vary all over the US as well as internationally, so the degree to which the wattage increase will affect you strongly depend on those local rates.
Im a bit concerned about the difficulty factors as well. For sure, no-one knows what will happen with difficulty, which is why this is open to large-scale debate. I won't presume to fault your numbers, but I want to offer an alternative viewpoint (which may be wrong, I admit).
My feeling is that the GSD devices don't offer a significant change in the overall hashrate compared to GPU's. They are in the same ballpark for hash-rate, as well as price/KHs. At-least roughly speaking. I mean that from the perspective of GSD at around 0.4 $/KHs and GPUs in the 1-2 $/KHs.
So, from my point of view, I don't expect to see any radical change in the global hashrate in the NEAR future due to the GSD marketplace.
That said, when some of the "second-generation" ASIC's hit the market later this summer (Fibonacci, KNC, Alpha, etc), those do offer the potential for a significant increase in the global hash-rate. It's when those ASICs hit the market that I would expect to see the difficulty start to spike.
I also know it's imossible to predict the future with any accuracy, so I would like to offer this alternative viewpoint on the difficult factors...
From Jan-now, LTC went from a difficulty around 3K to around 6K. If you divide that into the number of difficulty adjustments which occured, you find the average difficulty increase for 2014 (so-far) has been about 2.3% per increment. Im hoping this trend continues untill around July-Aug when the next-gen ASICs hit the market at which point it will spike much quicker (in my opinion).
That said, I think we have a window of time untill July to try and recoup as much as possible from the GSDs. After that, the chances of any ROI will dwindle rapidly.
Anyhow, just my opinion

arctic mx4? i'm afraid your just wasting your money and time by buying high-end thermal paste, messing with expensive cables etc. you're just pushing your theoretical roi further. although, most likely you never meet ROI using these fish cans anyway.. do it just for fun if you feel you need to. 

Thanks guys, I hadn't been googling for "Non-Electricity Conductive", I didn't know the terminology until your latest posts.

That is interesting. I have the original 3 mods on a pair of my units which allows them to clock to 1075Mhz stable.  This is yielding 454Kh/s pretty much both client and pool side.  How are you clocked 150Mhz more and not really getting any better performance?  I would expect at that clock 500+Kh/s easy client side.

I will be reverting one unit and trying this mod tomorrow after the resistors come in. We will see what changes.

You are right indeed! I was looking for something for OP and first thing came in to my head was MX-4 

That one works too, but no need for any expensive one, first price non conductive/capacitive will do. It's not like its a monster GPU that gets to 90 degrees 

www.newegg.com/Product/Product.aspx?Item=N82E16835186038

If you check the description, its says non-electricity conductive.

google

http://www.newegg.com/Product/Product.aspx?Item=N82E16835186020

https://www.google.com/search?q=non+conductive+thermal+paste&oq=non+conductive+ther&aqs=chrome.1.69i57j0l5.6387j1j7&sourceid=chrome&espv=210&es_sm=122&ie=UTF-8#q=non+conductive++capacitive+thermal+paste&safe=off

Yes well that brings us full circle back to my original problem: every paste I see says that it is "conductive" - so I'm asking if someone can show me a non-conductive paste that is suitable to use in the gridseeds.

Any paste will do, as long as it is non conductive. So nothing with "silver" "metal" etc..

Almost precisely the reason why I reverted my mods back to default without fans (except for a single 120mm fan running at 5v cooling 14 GS).

My simple reasoning was this. Why would I make the mod to double the hashrate when it quadruples the power consumption? Which produces more heat requiring the need for more cooling?

I admit, the mod is a great endeavor if the cost of power is factored out of the equation, ie gamers with their graphics cards.  Those gamers push for more gpu cycles irregardless of power costs.

Thanks but we have been advised not to use Arctic Silver @ https://bitcointalksearch.org/topic/m.6027770 or is the one you are inking to different somehow?

This can be realistic cenario indeed. IF that happens, ill just undo the mod in 3 months from now.

Profit and Loss

Looking at my Gridseeds and wondering if I should mod the hardware, and if so by how much, it seemed to me that the alarming increase in power usage as clock speed increases might have undesirable effects on the profitability. So I decided to get out my calculator. The results below are based on some tests performed by myself on the latest version of the Gridseed 5 chip module (the one with surface mount LED's). I modified one by changing R52 to 39k and another by changing R52 to 47K. The modules all had the 5V fan mod and were clocked at 850MHz, 1000MHz and 1200MHz, these values were chosen as the maximum which gave zero (or close to it) errors. The power cost is calculated from what I pay at about $0.20/KWHour and the modules using 4W, 15W and 25W. All figures relate to one Gridseed module running for one day and based on a bitcoin value of $460. Earnings figures are based upon a return of 0.004386 BTC/MHash/day.

R52	Power Cost	Earns Today	Profit Today	Earns Month 2	Profit Month 2	Earns Month 3	Profit Month 3
33k(standard)	$0.019	$0.72	$0.701	$0.36	$0.341	$0.18	$0.161
39k	$0.072	$0.857	$0.785	$0.4285	$0.3565	$0.21425	$0.14225
47k	$0.12	$1.029	$0.909	$0.5145	$0.3945	$0.25725	$0.13725

So assuming that the income from mining halves each month (seems likely it will) we can see that in just 2 or 3 months time it will be best to have unmodified Gridseeds.
I must emphasise that these figures are far from definitive and should only be viewed as a rough estimate. Your modules might give different results and your power cost is probably different from mine.
Overall I think my best bet is to leave mine standard with no fans at all rather than go to all the trouble of modding them only to have to put them back in a month or two.

Well done guys!
Also i would recomend some hot clue on top of that soldered spots. To make sure that resistor doesnt move and break the soldered spots and possibly damage the lines on PCB.

Nice work there Simon!
I see our work and pics have helped you determine you could do the mod yourself.
Enjoy your overclocking profits!
We've already determined the 'sweet spot'. But please let us know what you come up with too.

*bump* So much talk of thermal paste and still no one here can recommend one? (??)

p.s. sorry to bump.