Pages:
Author

Topic: GridSeed 5-chip USB miner voltage mod - page 30. (Read 157001 times)

member
Activity: 84
Merit: 10
April 24, 2014, 12:33:05 PM
phobya are good readily availbale and fairly cheap. probably one of the best thermal conducivity wise. certainly what ive found anyway. Although thermal paste is even better, but more messy.

I don't think that we need to contradict thermal paste and thermal pads...Just opposite...

IMHO, ...BOTTOM side should be the best with thin layer of thermal paste...
         ...TOP - with   (Phobya) thermal pads...and if your mechanics are so bad...supplemented with copper shims as needed on the top of the thermal pads...

Your mileage may vary...

ZiG

agreed. remeber to cover any exposed copper contacts on the underside though otherwise you risk a short! i remember the rest button having two contact on the underside and a few others.
sr. member
Activity: 378
Merit: 250
April 24, 2014, 12:30:42 PM

WRONG!
12V powers the processors and the fan is basically inconsequential where power draw goes. Wink
5V powers USB UART, controller and all logic circuits and if you want, my 5V fan mod as well.
You cannot hash with only 12V or 5V seperately. Both voltages MUST be present in order to hash! PERIOD!
Yes you need a supplemental 5V power supply if you are going to run more than 1 seed from a standard USB port. PERIOD!
You need and can safely use 13.8VDC 'standard output voltage' from any 'good' working power supply.
If you are only mining with 1 seed, you can do so without supplemental power pretty safely I would assume. I haven't done it since I started out with 4 pods, and now have 6 modded pods running and soon will have more modded pods hashing away making me money 24/7 Wink Cool!
Good luck!
sure check the schematic. the 12v powers the fan.   only the fan the processors arent powered by 12v the 12v is also converted to 5v and 3.3v via chip u7 which are fed to the processors via other bits and pieces.
the 5v from usb is just for that the uart com link. nothing more it doesnt draw any power from usb - unless there is the wolfy fan mod.

Here, let's try this.
Plug in the 5V USB power alone and see if the seed will hash and produce shares. The answer WILL be NO!
Next, try 12V power without 5V USB power. Same result. The answer WILL be NO!
Then plug in your 12V power and 5V USB power and see if they will hash now. The answer WILL be YES!
You're either mistaken or confused. The 12V is needed for processing to occur. It also powers the fan, yes but the fan can be connected to 5V USB power with very little current draw i.e. 78mA nominal.
The 5V feeds power to the logic circuits that run the chips, not the actual switching side of the chips that do the calculations etc. 12V IS what powers that!
Also, think of it this way. Why else would we need 12V 20 to 30 AMP power supplies to feed power to just the fans?
Get it?
Okay, hope this straightens things out for you! Wink
Whew!
ZiG
sr. member
Activity: 406
Merit: 250
April 24, 2014, 12:30:23 PM
phobya are good readily availbale and fairly cheap. probably one of the best thermal conducivity wise. certainly what ive found anyway. Although thermal paste is even better, but more messy.

I don't think that we need to contradict thermal paste and thermal pads...Just opposite...

IMHO, ...BOTTOM side should be the best with thin layer of thermal paste...
         ...TOP - with   (Phobya) thermal pads...and if your mechanics are so bad...supplemented with copper shims as needed on the top of the thermal pads...

Your mileage may vary...

ZiG
hero member
Activity: 826
Merit: 1000
°^°
April 24, 2014, 12:26:16 PM
phobya are good readily availbale and fairly cheap. probably one of the best thermal conducivity wise. certainly what ive found anyway. Although thermal paste is even better, but more messy.
nothing beats thermal paste (>10W/mK)

but like one said here, you need to mill down a part of the heatsink, and i dont have such tools
hero member
Activity: 826
Merit: 1000
°^°
April 24, 2014, 12:25:01 PM
How many HW errors...?

BTW, Did you measure the Voltage supplied to the Gridseed chips...What is it...?

EDIT ...: What about cooling...Stock or modded...I see Phobya pads...anything else...?
6 in 2h, trying 1175 MHz now

1,69V

nope, just cleaned the chips and put 1,5mm Phobya on - next time will try 1mm
member
Activity: 84
Merit: 10
April 24, 2014, 12:20:11 PM

WRONG!
12V powers the processors and the fan is basically inconsequential where power draw goes. Wink
5V powers USB UART, controller and all logic circuits and if you want, my 5V fan mod as well.
You cannot hash with only 12V or 5V seperately. Both voltages MUST be present in order to hash! PERIOD!
Yes you need a supplemental 5V power supply if you are going to run more than 1 seed from a standard USB port. PERIOD!
You need and can safely use 13.8VDC 'standard output voltage' from any 'good' working power supply.
If you are only mining with 1 seed, you can do so without supplemental power pretty safely I would assume. I haven't done it since I started out with 4 pods, and now have 6 modded pods running and soon will have more modded pods hashing away making me money 24/7 Wink Cool!
Good luck!
sure check the schematic. the 12v powers the fan.   only the fan the processors arent powered by 12v the 12v is also converted to 5v and 3.3v via chip u7 which are fed to the processors via other bits and pieces.
the 5v from usb is just for that the uart com link. nothing more it doesnt draw any power from usb - unless there is the wolfy fan mod.
member
Activity: 84
Merit: 10
April 24, 2014, 12:11:52 PM
phobya are good readily availbale and fairly cheap. probably one of the best thermal conducivity wise. certainly what ive found anyway. Although thermal paste is even better, but more messy.
sr. member
Activity: 378
Merit: 250
April 24, 2014, 12:07:30 PM


Well, all I can say is, find and clear up all of your misunderstood words and concepts!
Ever heard the term 'heat soak'?
Fan forced air is just that. FORCED!
The fans on our pods don't cool the chips directly 'obviously'...They cool the fins that pull the heat off of the chips just fine. Particulate buildup on the fins is minimal and of course one should keep them clean with regular maintenance. Sorry. Dead-spot, schmed-spot! Doesn't matter in this case IMO....
Next?


"Pull heat off the chips" so then you just push the heat back down? seems counter really, carry on pulling it out. Heat rises.
Particulate buildup over time can and will impact the cooling. of course as you said maintance can solve that, but its easier to clean from the outside than dissambling it to remove it all from inside.

https://www.youtube.com/watch?v=UyC3lZ5WFMk
skip to about 4 mins youll see the dust build up and DEAD spot - ok on a radiator fan system same concept though same results.
And the dead spot is in the center / centre right where the huge block of heatsink is, the bit thatll conduct most of the heat before it gets to the fins. so yeah big issue really.

All im saying on this for now - might bring it back up again if i can find my thermal diode should have some spares. then ill do an experment of push vs pull and check the temps on the heatsink next to the chips.

on to overclocking and stock . Judging from the reporting watted of overclocked grids, they appear to be drawing the same amount as if hashing in SHA. That wattage is converted to heat and dissapated via the heatsink and fan. Now considering the heatsinks and fan where desgined - at stock - to deal wiht the SHA hashing, overclocking them leaving in stock they should cope fine. If it goes higher than the desgined watted - was it 60 or 70 watts?Huh) then recommended to upgrade the thermal pads and cooling. Our overcloking is equiv to hashing bitcoins. Just an after thought i had. so yeah maybe not need to upgrade the cooling but wont hurt, if nothing else peace of mind. besides its what we modders do Smiley

Right. Okay, Cool, have fun! Wink
ZiG
sr. member
Activity: 406
Merit: 250
April 24, 2014, 11:57:34 AM
besides, im doing 1,2Ghz stable so far with only 47k
but i replaced the thermal pad with Phobya 7W/mK

Super Xer0,

This is probably the sweet spot of OC performance we can get...

How many HW errors...?

BTW, Did you measure the Voltage supplied to the Gridseed chips...What is it...?

ZiG

EDIT ...: What about cooling...Stock or modded...I see Phobya pads...anything else...?

Thanks...
member
Activity: 84
Merit: 10
April 24, 2014, 11:51:18 AM


Well, all I can say is, find and clear up all of your misunderstood words and concepts!
Ever heard the term 'heat soak'?
Fan forced air is just that. FORCED!
The fans on our pods don't cool the chips directly 'obviously'...They cool the fins that pull the heat off of the chips just fine. Particulate buildup on the fins is minimal and of course one should keep them clean with regular maintenance. Sorry. Dead-spot, schmed-spot! Doesn't matter in this case IMO....
Next?


"Pull heat off the chips" so then you just push the heat back down? seems counter really, carry on pulling it out. Heat rises.
Particulate buildup over time can and will impact the cooling. of course as you said maintance can solve that, but its easier to clean from the outside than dissambling it to remove it all from inside.

https://www.youtube.com/watch?v=UyC3lZ5WFMk
skip to about 4 mins youll see the dust build up and DEAD spot - ok on a radiator fan system same concept though same results.
And the dead spot is in the center / centre right where the huge block of heatsink is, the bit thatll conduct most of the heat before it gets to the fins. so yeah big issue really.

All im saying on this for now - might bring it back up again if i can find my thermal diode should have some spares. then ill do an experment of push vs pull and check the temps on the heatsink next to the chips.

on to overclocking and stock . Judging from the reporting watted of overclocked grids, they appear to be drawing the same amount as if hashing in SHA. That wattage is converted to heat and dissapated via the heatsink and fan. Now considering the heatsinks and fan where desgined - at stock - to deal wiht the SHA hashing, overclocking them leaving in stock they should cope fine. If it goes higher than the desgined watted - was it 60 or 70 watts?Huh) then recommended to upgrade the thermal pads and cooling. Our overcloking is equiv to hashing bitcoins. Just an after thought i had. so yeah maybe not need to upgrade the cooling but wont hurt, if nothing else peace of mind. besides its what we modders do Smiley
hero member
Activity: 826
Merit: 1000
°^°
April 24, 2014, 11:34:18 AM
besides, im doing 1,2Ghz stable so far with only 47k
but i replaced the thermal pad with Phobya 7W/mK
sr. member
Activity: 378
Merit: 250
April 24, 2014, 11:29:36 AM
I'd hate to piss in the fire here, but at one time a company did make what was called a TMD fan which minimized the center deadspot by moving the fan motor(s) to 4 corners of the fan.  I had one, it was VERY powerful for a 70mm fan, had a LOT of torque and produced very good pressure.  Trouble is it didn't last very long.  They don't make them anymore that I'm aware of.  Back when I built my watercooling setup I had looked for some but couldn't find any.

ah...found a link to a wiki:  http://en.wikipedia.org/wiki/Tip-magnetic_driving
and a youtube :  https://www.youtube.com/watch?v=kKofleGNE7g

i did find this spinning fan :

http://www.geek.com/chips/sandias-floating-spinning-heatsink-promises-30x-better-cpu-cooling-1498703/

That's really cool....but since the heatsink is spinning and floating above the base....how is it transferring the heat from the cpu to the floating spinning heatsink???  

Ahh! FIRE! FIRE! Shocked
Cool video and product! I like it! Science, engineering and invention are so cool, eh?
It works like a hurricane or tornado effect. Sucks air in and up from the processor's upper face.
Simple fluid dynamics Wink Excellent! Simply further evidence that Simplicity = Efficiency! Smart stuff!
sr. member
Activity: 378
Merit: 250
April 24, 2014, 11:23:00 AM
While on the subject of thermals if you really wanted to get down to the root of the cooling issue you'll have to either use thin copper shims or mill out part of the heatsink so the taller components don't prevent full contact with the gc3355 chips.  If you chose to mill out part of the heatsink you will also notice many of the capacitors around the gc3355 chips are the same height if not just a tad higher so those will need to be insulated to prevent electrical contact with the heatsink.

Shims are easier Wink

NSt this is what I am going to do and also put a thin layer of paste on the on the bottom...

Zig made a good point I TOTALLY forgot about the short out part, If I lay a shrim directly on the tope gridseed wouldnt this short it out? Wouldnt it be soo soooooo close to the pins that a single touch and fry it? how are people with shrims putting these on the top, Id love to see a picture

In order to avoid shorting out any components, thereby ruining your miner, you need to have a pad / insulator between the top of the chips and the shim. Letting the shim touch the bottom plate of the top heat sink is inconsequential, electrically speaking.
Be aware that the more shim thickness you put in there, the higher the risk of crushing a chip or breaking a pin/s where they connect the chip to the copper traces.
Torque the screws evenly and do not over tighten them what ever you do.
That's all I did with the stock pads in there, was evenly tighten and torque the screws.
Honestly, I know some of the pods are pretty messed up requiring some intervention but I believe it is unnecessary in most cases to even mess with changing out the pads or adding heat sink paste. It has nothing to do with HW errors in any case, IMO. The pads are non-conductive by default. It's the gap between certain components that cause a difference in a minute amount of emf capacitance between the bottom heat sink plate and the chips on the card. When it's a certain gap, it possibly causes HW errors at certain frequencies due to this capacitive issue. When it's at the ideal gap, the capacitance issue hence HW errors go away.
That's what I think is going on.  Shocked

Ok So what Ill do then is use a blade and cut the original pad in the shape of the shim, and then let the shim touch the big heat sink bare back .


Id STILL LOVE to morph two pods in to the one.. Not sure if people saw my other post where  I did a photo edit and morphed two pods into one removing one of the Top sinks and replacing it with one fan sharing the 2 boards. Basically starting from bottom up, Large Sink > Small Sink (Fan removed on bottom pod) 2nd pod > small sink with 2nd board facing up > Attach the cross member onto the 4 gold screw stands then comes the 1 Fan..ALl the screws line up and its just a matter of putting it together... The pod looks a little longer than the original pod But has 2 boards in there 10 chips and only uses 1 fan..  This -= Less wattage = Less space...

Only thing I wonder is the heat and air flow....

I would love to try this

Yep, sounds like a plan. I think you've got it figured out.
I like the morphing idea. And bathing the chips with air from the fan will work well but sucking air from the chips will probably not work well at all. They need to be bathed/covered/soaked/drenched in cooling air in order to be properly cooled.
Same with blowing instead of sucking air through the fins on the top heat sink. That's the way all heat sinks are cooled. By FORCING air through them, not sucking it. Well, unless they are ducted, then perhaps sucking air through would work.
So.....Nice idea though.
Bed time! Nighty night yall!


sorry wolfy but your wrong here matey.

Some heat sinks remove the hot air from them allowing the cool room temperature air to flow in, create a negative pressure difference - ie graphics card remove the air, i know for a fact the sapphire vapor x does it this way. its more effective to create a negative pressure. Nowadays with processor power some top end heatsinks and fans use a push pull system, to really improve the air flow.
and sorry dont know the link, but there was a review i read while building my water cooled system on different configurations for cooling of the radiator. the most effective cooling was to use a push pull system with the fans mounted in a collar 5-10 cam away from the radiator. the second most effective way was to pull the air through the radiator, again fan mounted 5-10 cm away from radiator.

Just looks at the heat sink, tell me air can cleanly and clearly flow to the chips!? nop no chance, turbulence all the way. will it work yes, will it be efficient and effective not really. to get max cooling and efficiency from your fan flip it over, so it is pulling the air out the system, and lift it away about 5cm or so in a shroud. itll mean lengthening the cable a bit though. fans mounted directly to heatsinks are ineffective, the very nature of a fan cause this, it creates a dead spot right in the middle of the fan, just where the motor is where little or no air hits. lifting the fan removes this dead spot. this is a well known fact, that i assume most serious overclockers know about, but its a common "fault" on factory heatsinks - its all about factory cost, for the AVERAGE user, not an overclocking user.

Basically the grids cool well in stock for stock frequencies. start overclocking these and you should consider doing something to improve cooling, before 3-6 months down the line youve burnt your system out. your pushing the chips and other components past their safe limits, it puts more waer on them. just like your car engine cools fine, but start turbo evrywhere and riding the revs high itll not be long before youve blown the engine.

Overclocking generates more heat and this extra heat need  to be removed from the system.
And heck even just say that the chips run really cool anyway overclocked, what harm is it going to do? NONE! if anything itll only help, a cooler chip is a happy chip. itll def last longer.

Although you could always just stick the grids next to a big fan or two, 14cm fans push pull system.

Stock heatsinks in this trade are all CRAP. yeah big enough, but allow fro properly air flow no. look at the usb bitcoin miners and the heatsinks, then think how you would put a fan to blow air over them, then consider where the actual air flow is going. from above it gets blocked by the hub or usb port, from the side the first fin blocks the air flow, as most are vertical fins. simple and cheap. effective yes efficient NO

WTF? Are you on drugs?
No, I am not wrong about how fan forced cooling works. Geesh, read a book on the subject and get some experience too! Geesh!
read all the books on it at university - bsc applied physics with computering. and considering these are cheap crap noisey fans, not optimsied for pressure differecne etc like modern fans, i say pull.

look read all the books and google all you want for every reason for push the is a counter and valid reason for pull on fan. push or pull its a long running debate. and i cant find the link it was old.

whichever way you go put a fan shroud on it. No one can argue that a fan shroud wont improve its effiecnieny even you wolfy are disagree on that!??? it removes the dead spot - you do know about that it was mentioned in your books?

But i can give you the best reason for putting the fan in pull as opposed to push - dust!
in push all the dust will get pushed into your heatsink, and after months or years depending on how dusty your area is - ie living next to main road windows open etc will produce more dust - will leave all that dust inside your heatsink. dust reduces the the effectiveness of a heatsink - common problme for laptop failing overheating through dust being drawn in as people use them on their laps bed etc.
In pull the air in pull into it and the heatsink will act as a largish filter and will trap some of the dust before it gets in.



Well, all I can say is, find and clear up all of your misunderstood words and concepts!
Ever heard the term 'heat soak'?
Fan forced air is just that. FORCED!
The fans on our pods don't cool the chips directly 'obviously'...They cool the fins that pull the heat off of the chips just fine. Particulate buildup on the fins is minimal and of course one should keep them clean with regular maintenance. Sorry. Dead-spot, schmed-spot! Doesn't matter in this case IMO....
Next?

sr. member
Activity: 252
Merit: 254
April 24, 2014, 11:22:30 AM
I'd hate to piss in the fire here, but at one time a company did make what was called a TMD fan which minimized the center deadspot by moving the fan motor(s) to 4 corners of the fan.  I had one, it was VERY powerful for a 70mm fan, had a LOT of torque and produced very good pressure.  Trouble is it didn't last very long.  They don't make them anymore that I'm aware of.  Back when I built my watercooling setup I had looked for some but couldn't find any.

ah...found a link to a wiki:  http://en.wikipedia.org/wiki/Tip-magnetic_driving
and a youtube :  https://www.youtube.com/watch?v=kKofleGNE7g

i did find this spinning fan :

http://www.geek.com/chips/sandias-floating-spinning-heatsink-promises-30x-better-cpu-cooling-1498703/

That's really cool....but since the heatsink is spinning and floating above the base....how is it transferring the heat from the cpu to the floating spinning heatsink??? 
member
Activity: 84
Merit: 10
April 24, 2014, 11:13:48 AM
I'd hate to piss in the fire here, but at one time a company did make what was called a TMD fan which minimized the center deadspot by moving the fan motor(s) to 4 corners of the fan.  I had one, it was VERY powerful for a 70mm fan, had a LOT of torque and produced very good pressure.  Trouble is it didn't last very long.  They don't make them anymore that I'm aware of.  Back when I built my watercooling setup I had looked for some but couldn't find any.

ah...found a link to a wiki:  http://en.wikipedia.org/wiki/Tip-magnetic_driving
and a youtube :  https://www.youtube.com/watch?v=kKofleGNE7g

i did find this spinning fan :

http://www.geek.com/chips/sandias-floating-spinning-heatsink-promises-30x-better-cpu-cooling-1498703/
member
Activity: 84
Merit: 10
April 24, 2014, 11:08:50 AM

WTF? Are you on drugs?
No, I am not wrong about how fan forced cooling works. Geesh, read a book on the subject and get some experience too! Geesh!

It needs cool air brought to it, and needs warm air taken away. Chicken/Egg issue.... It really depends which side of the fan you are on!

 Grin Grin Grin Grin Grin Grin
sr. member
Activity: 252
Merit: 254
April 24, 2014, 11:06:32 AM
I'd hate to piss in the fire here, but at one time a company did make what was called a TMD fan which minimized the center deadspot by moving the fan motor(s) to 4 corners of the fan.  I had one, it was VERY powerful for a 70mm fan, had a LOT of torque and produced very good pressure.  Trouble is it didn't last very long.  They don't make them anymore that I'm aware of.  Back when I built my watercooling setup I had looked for some but couldn't find any.

ah...found a link to a wiki:  http://en.wikipedia.org/wiki/Tip-magnetic_driving
and a youtube :  https://www.youtube.com/watch?v=kKofleGNE7g
member
Activity: 84
Merit: 10
April 24, 2014, 10:58:33 AM
While on the subject of thermals if you really wanted to get down to the root of the cooling issue you'll have to either use thin copper shims or mill out part of the heatsink so the taller components don't prevent full contact with the gc3355 chips.  If you chose to mill out part of the heatsink you will also notice many of the capacitors around the gc3355 chips are the same height if not just a tad higher so those will need to be insulated to prevent electrical contact with the heatsink.

Shims are easier Wink

NSt this is what I am going to do and also put a thin layer of paste on the on the bottom...

Zig made a good point I TOTALLY forgot about the short out part, If I lay a shrim directly on the tope gridseed wouldnt this short it out? Wouldnt it be soo soooooo close to the pins that a single touch and fry it? how are people with shrims putting these on the top, Id love to see a picture

In order to avoid shorting out any components, thereby ruining your miner, you need to have a pad / insulator between the top of the chips and the shim. Letting the shim touch the bottom plate of the top heat sink is inconsequential, electrically speaking.
Be aware that the more shim thickness you put in there, the higher the risk of crushing a chip or breaking a pin/s where they connect the chip to the copper traces.
Torque the screws evenly and do not over tighten them what ever you do.
That's all I did with the stock pads in there, was evenly tighten and torque the screws.
Honestly, I know some of the pods are pretty messed up requiring some intervention but I believe it is unnecessary in most cases to even mess with changing out the pads or adding heat sink paste. It has nothing to do with HW errors in any case, IMO. The pads are non-conductive by default. It's the gap between certain components that cause a difference in a minute amount of emf capacitance between the bottom heat sink plate and the chips on the card. When it's a certain gap, it possibly causes HW errors at certain frequencies due to this capacitive issue. When it's at the ideal gap, the capacitance issue hence HW errors go away.
That's what I think is going on.  Shocked

Ok So what Ill do then is use a blade and cut the original pad in the shape of the shim, and then let the shim touch the big heat sink bare back .


Id STILL LOVE to morph two pods in to the one.. Not sure if people saw my other post where  I did a photo edit and morphed two pods into one removing one of the Top sinks and replacing it with one fan sharing the 2 boards. Basically starting from bottom up, Large Sink > Small Sink (Fan removed on bottom pod) 2nd pod > small sink with 2nd board facing up > Attach the cross member onto the 4 gold screw stands then comes the 1 Fan..ALl the screws line up and its just a matter of putting it together... The pod looks a little longer than the original pod But has 2 boards in there 10 chips and only uses 1 fan..  This -= Less wattage = Less space...

Only thing I wonder is the heat and air flow....

I would love to try this

Yep, sounds like a plan. I think you've got it figured out.
I like the morphing idea. And bathing the chips with air from the fan will work well but sucking air from the chips will probably not work well at all. They need to be bathed/covered/soaked/drenched in cooling air in order to be properly cooled.
Same with blowing instead of sucking air through the fins on the top heat sink. That's the way all heat sinks are cooled. By FORCING air through them, not sucking it. Well, unless they are ducted, then perhaps sucking air through would work.
So.....Nice idea though.
Bed time! Nighty night yall!


sorry wolfy but your wrong here matey.

Some heat sinks remove the hot air from them allowing the cool room temperature air to flow in, create a negative pressure difference - ie graphics card remove the air, i know for a fact the sapphire vapor x does it this way. its more effective to create a negative pressure. Nowadays with processor power some top end heatsinks and fans use a push pull system, to really improve the air flow.
and sorry dont know the link, but there was a review i read while building my water cooled system on different configurations for cooling of the radiator. the most effective cooling was to use a push pull system with the fans mounted in a collar 5-10 cam away from the radiator. the second most effective way was to pull the air through the radiator, again fan mounted 5-10 cm away from radiator.

Just looks at the heat sink, tell me air can cleanly and clearly flow to the chips!? nop no chance, turbulence all the way. will it work yes, will it be efficient and effective not really. to get max cooling and efficiency from your fan flip it over, so it is pulling the air out the system, and lift it away about 5cm or so in a shroud. itll mean lengthening the cable a bit though. fans mounted directly to heatsinks are ineffective, the very nature of a fan cause this, it creates a dead spot right in the middle of the fan, just where the motor is where little or no air hits. lifting the fan removes this dead spot. this is a well known fact, that i assume most serious overclockers know about, but its a common "fault" on factory heatsinks - its all about factory cost, for the AVERAGE user, not an overclocking user.

Basically the grids cool well in stock for stock frequencies. start overclocking these and you should consider doing something to improve cooling, before 3-6 months down the line youve burnt your system out. your pushing the chips and other components past their safe limits, it puts more waer on them. just like your car engine cools fine, but start turbo evrywhere and riding the revs high itll not be long before youve blown the engine.

Overclocking generates more heat and this extra heat need  to be removed from the system.
And heck even just say that the chips run really cool anyway overclocked, what harm is it going to do? NONE! if anything itll only help, a cooler chip is a happy chip. itll def last longer.

Although you could always just stick the grids next to a big fan or two, 14cm fans push pull system.

Stock heatsinks in this trade are all CRAP. yeah big enough, but allow fro properly air flow no. look at the usb bitcoin miners and the heatsinks, then think how you would put a fan to blow air over them, then consider where the actual air flow is going. from above it gets blocked by the hub or usb port, from the side the first fin blocks the air flow, as most are vertical fins. simple and cheap. effective yes efficient NO

WTF? Are you on drugs?
No, I am not wrong about how fan forced cooling works. Geesh, read a book on the subject and get some experience too! Geesh!
read all the books on it at university - bsc applied physics with computering. and considering these are cheap crap noisey fans, not optimsied for pressure differecne etc like modern fans, i say pull.

look read all the books and google all you want for every reason for push the is a counter and valid reason for pull on fan. push or pull its a long running debate. and i cant find the link it was old.

whichever way you go put a fan shroud on it. No one can argue that a fan shroud wont improve its effiecnieny even you wolfy are disagree on that!??? it removes the dead spot - you do know about that it was mentioned in your books?

But i can give you the best reason for putting the fan in pull as opposed to push - dust!
in push all the dust will get pushed into your heatsink, and after months or years depending on how dusty your area is - ie living next to main road windows open etc will produce more dust - will leave all that dust inside your heatsink. dust reduces the the effectiveness of a heatsink - common problme for laptop failing overheating through dust being drawn in as people use them on their laps bed etc.
In pull the air in pull into it and the heatsink will act as a largish filter and will trap some of the dust before it gets in.

sr. member
Activity: 378
Merit: 250
April 24, 2014, 10:42:57 AM
I am just about at my wits end with these stupid gridseeds.
Nothing I do works properly. I am just about to throw these in the trash and try to find peace in my life.

The only software that seems to work for me is bfgminer.
I have tried cpuminer (works I think, but who knows)
cgminer with zadig drivers (can't even find the devices)
The raspberry pi that came with them (wouldn't even boot with the provided SD card, had to load a copy of scripta)
The pi will boot, but goes down after about 10 minutes.

So I am basically back to BFGminer. It's the only one I can get to work, if I modify the bat file that was included with my info.

So, here I am, back to bfgminer, I have modded 2 of 10 gridseeds, but I can't figure out how to run 8 of them at thr stock 850 and increase the clock of the 2 modded units.
When I add "--set-device gridseed@XX72117B5XXX:clock=950" nothing works at all, the command window opens then closes right away.

Yeah, Gridseed and Dualminers are PITA...  Huh You should put Your serial numbers instead. BFGminer versions are still experimental, BTW pr3 is more stable for me than clean-version. I'm using CGwatcher for running BFGminer because of crashing and unsuccessfull starts You are experiencing too. Here is my command line with my serial numbers and my serial ports, so You can't just copy/paste the line:
Code:
bfgminer -S noauto -S dualminer:all -S gridseed:all --set-device dualminer@\\.\COM25:clock=800 --set-device dualminer@\\.\COM27:clock=850 --set-device dualminer@\\.\COM23:clock=850 --set-device gridseed@XXX2117B5355:clock=875 --set-device gridseed@XXXX4F8D5148:clock=900 --set-device gridseed@XXXX336B5148:clock=850 --set-device gridseed@XXXX5F7E5148:clock=925
I'm using also standard bfgminer.conf with pools definitions.

Thanks for all the help. I have now modded 10 with the 47k resistor. I have them all working and have each set to the lowest frequency that gives the least HW errors for each one using BFGminer PR3.
I think all is working bout as good as I can make them. Each one is running about 470khs.

Now, all that I would like to do is set up BFGminer to run in CGwatcher like i use for my GPUs.

Has anyone done this? I think it could be done, except I have no idea how to setup the Conf file.

Does anybody have an example Conf for BFGminer to run in CGwatcher that shows how to setup for multiple Gridseeds using the ID of the gridseed and the frequency for it?

Thanks!

Hi, I'm a little bit shocked!
How in the world did you get stuck using 47k on your mods? 49.9k is the optimum value for least power usage and maximum hash rate with extremely minimum / nearly 0 HW errors?
If you want to see how it's done, look up my posts on here via Search. You can start here https://bitcointalksearch.org/topic/m.6098563 to see how it's done if you don't already know, which you probably do Wink
Also, unless you're not using Windows 7 32 or 64 bit, you can use cpuminer which is the best and most stable and easiest mining program to use as well as the most profitable at the pool!
I believe if you are getting any HW errors, like a lot, it's due in part to using the 47k and bfgminer.
If you can and are willing, I suggest upgrading to the 49.9k or 50k is okay too and change over to cpuminer using the STM virtual comm port drivers and a good USB hub.
You can also covert do my 5V USB fan mod while you're at it. It will run ultra quiet while soaking the cooling fins with enough air to keep them running at room temperature.
Good luck in what ever you decide!

I'm afraid that I did not have any 49.9k at my disposal. I did try 51k because I had those, but no luck with them.
Today I am working on adding 3k in series with my 47k to get as close to 49.9k as I can.
So far I have done this to 2w of them and it seems to be working. Much more stable. Tyhey are at about  49.6k measure accross both resistors, so I am pretty close.
I am going to do this to all 10. This should put me right where I want to be.
Since I can't really get BFGminer to work right with CGwatcher, I am going to abandon it all together.
From what I am reading, most people are recommending CPUminer.
I see there is a new version out that will Autotune the gridseeds! That is awesome and exactly what I am looking for!
Also everyone says that it report poolside hash rate properly as well.

So I am going to give this a try.
I don't have any experience with CPUminer though.

Can anyone post an example .BAT file that I could copy that will show how to setup Autotune function for each miner?
I have looked at the included AUTOtune .BAT example that came with cpuminer, but it only shows setup for 1 gridseed. I have to make it run 10 of them.

Thanks for all the help so far, and for all the great work done to help figure all this out. I am getting closer every day to a properly working setup!

RE: 49.6K ohm combo... that'll work! Wink Happy hashing!
sr. member
Activity: 378
Merit: 250
April 24, 2014, 10:31:19 AM
While on the subject of thermals if you really wanted to get down to the root of the cooling issue you'll have to either use thin copper shims or mill out part of the heatsink so the taller components don't prevent full contact with the gc3355 chips.  If you chose to mill out part of the heatsink you will also notice many of the capacitors around the gc3355 chips are the same height if not just a tad higher so those will need to be insulated to prevent electrical contact with the heatsink.

Shims are easier Wink

NSt this is what I am going to do and also put a thin layer of paste on the on the bottom...

Zig made a good point I TOTALLY forgot about the short out part, If I lay a shrim directly on the tope gridseed wouldnt this short it out? Wouldnt it be soo soooooo close to the pins that a single touch and fry it? how are people with shrims putting these on the top, Id love to see a picture

In order to avoid shorting out any components, thereby ruining your miner, you need to have a pad / insulator between the top of the chips and the shim. Letting the shim touch the bottom plate of the top heat sink is inconsequential, electrically speaking.
Be aware that the more shim thickness you put in there, the higher the risk of crushing a chip or breaking a pin/s where they connect the chip to the copper traces.
Torque the screws evenly and do not over tighten them what ever you do.
That's all I did with the stock pads in there, was evenly tighten and torque the screws.
Honestly, I know some of the pods are pretty messed up requiring some intervention but I believe it is unnecessary in most cases to even mess with changing out the pads or adding heat sink paste. It has nothing to do with HW errors in any case, IMO. The pads are non-conductive by default. It's the gap between certain components that cause a difference in a minute amount of emf capacitance between the bottom heat sink plate and the chips on the card. When it's a certain gap, it possibly causes HW errors at certain frequencies due to this capacitive issue. When it's at the ideal gap, the capacitance issue hence HW errors go away.
That's what I think is going on.  Shocked

Ok So what Ill do then is use a blade and cut the original pad in the shape of the shim, and then let the shim touch the big heat sink bare back .


Id STILL LOVE to morph two pods in to the one.. Not sure if people saw my other post where  I did a photo edit and morphed two pods into one removing one of the Top sinks and replacing it with one fan sharing the 2 boards. Basically starting from bottom up, Large Sink > Small Sink (Fan removed on bottom pod) 2nd pod > small sink with 2nd board facing up > Attach the cross member onto the 4 gold screw stands then comes the 1 Fan..ALl the screws line up and its just a matter of putting it together... The pod looks a little longer than the original pod But has 2 boards in there 10 chips and only uses 1 fan..  This -= Less wattage = Less space...

Only thing I wonder is the heat and air flow....

I would love to try this

Yep, sounds like a plan. I think you've got it figured out.
I like the morphing idea. And bathing the chips with air from the fan will work well but sucking air from the chips will probably not work well at all. They need to be bathed/covered/soaked/drenched in cooling air in order to be properly cooled.
Same with blowing instead of sucking air through the fins on the top heat sink. That's the way all heat sinks are cooled. By FORCING air through them, not sucking it. Well, unless they are ducted, then perhaps sucking air through would work.
So.....Nice idea though.
Bed time! Nighty night yall!


sorry wolfy but your wrong here matey.

Some heat sinks remove the hot air from them allowing the cool room temperature air to flow in, create a negative pressure difference - ie graphics card remove the air, i know for a fact the sapphire vapor x does it this way. its more effective to create a negative pressure. Nowadays with processor power some top end heatsinks and fans use a push pull system, to really improve the air flow.
and sorry dont know the link, but there was a review i read while building my water cooled system on different configurations for cooling of the radiator. the most effective cooling was to use a push pull system with the fans mounted in a collar 5-10 cam away from the radiator. the second most effective way was to pull the air through the radiator, again fan mounted 5-10 cm away from radiator.

Just looks at the heat sink, tell me air can cleanly and clearly flow to the chips!? nop no chance, turbulence all the way. will it work yes, will it be efficient and effective not really. to get max cooling and efficiency from your fan flip it over, so it is pulling the air out the system, and lift it away about 5cm or so in a shroud. itll mean lengthening the cable a bit though. fans mounted directly to heatsinks are ineffective, the very nature of a fan cause this, it creates a dead spot right in the middle of the fan, just where the motor is where little or no air hits. lifting the fan removes this dead spot. this is a well known fact, that i assume most serious overclockers know about, but its a common "fault" on factory heatsinks - its all about factory cost, for the AVERAGE user, not an overclocking user.

Basically the grids cool well in stock for stock frequencies. start overclocking these and you should consider doing something to improve cooling, before 3-6 months down the line youve burnt your system out. your pushing the chips and other components past their safe limits, it puts more waer on them. just like your car engine cools fine, but start turbo evrywhere and riding the revs high itll not be long before youve blown the engine.

Overclocking generates more heat and this extra heat need  to be removed from the system.
And heck even just say that the chips run really cool anyway overclocked, what harm is it going to do? NONE! if anything itll only help, a cooler chip is a happy chip. itll def last longer.

Although you could always just stick the grids next to a big fan or two, 14cm fans push pull system.

Stock heatsinks in this trade are all CRAP. yeah big enough, but allow fro properly air flow no. look at the usb bitcoin miners and the heatsinks, then think how you would put a fan to blow air over them, then consider where the actual air flow is going. from above it gets blocked by the hub or usb port, from the side the first fin blocks the air flow, as most are vertical fins. simple and cheap. effective yes efficient NO

WTF? Are you on drugs?
No, I am not wrong about how fan forced cooling works. Geesh, read a book on the subject and get some experience too! Geesh!
Pages:
Jump to: