Pages:
Author

Topic: GRIDSEED G-BLADE Overclocking 7Mh/s, improvements and repair - page 7. (Read 74040 times)

sr. member
Activity: 346
Merit: 260
Impressive thread guys, but how would this affect the warranty offered by the mining company you bought the gridseed from? Do you think they would care?

Warranty, what is that?

You would be lucky to get 30 days on a unit from these manufactures.
These blades are coming to the end of their life as the new units that
will be out soon are go up to 500Mhs and I am sure 1Ghs is in 2015!

That being said you could get 8 for around $2.4k and lightly mod (900Mhz)
to get around 48Mhs at 200 watt each unit. (approx 1600w max)

But the A2 28NM Scrypt Miner 28Mh is only $1.6K and @ 250watts so
about <1/3 the power draw in watts.
newbie
Activity: 28
Merit: 0
The answer is simple : They don´t care about this! Any modifications , even if this include just a higher resistance for the original 33 kOhm Resistor exclude any warranty.
I don´t know what kind of warranty some firms like Zoomhash offer , who sell modifed Blades , but if there is any , it is restricted...
There are surely some Sellers that offer some better conditions but have in mind that the manufactor exclude any warranty only if you replace or remove the original fan, this was the same for the smaller usb gridseeds who don´t really need a fan in scrypt mode.
sr. member
Activity: 266
Merit: 250
Impressive thread guys, but how would this affect the warranty offered by the mining company you bought the gridseed from? Do you think they would care?
newbie
Activity: 28
Merit: 0
Here are some examples how the temperature influence the drain current in Ampere. (In this example for the Dual Cool NEX Mosfet from Texas Instruments)

At 50 C° ---> 100 A
At 75 C° ---> 100 A
At 100 C° ---> 95 A
At 125 C° ---> 80 A
At 137.5 C° ---> 40 A
At 150 C° ----> 0 A !!!

This one has a max operation temp of 150 C° the IRF ones has 175 C° , but they react similar.
And this is the reason why the temps of the mosfets affects the HW´s.
As you can see the critical case temperature is at something near 90 C° , then the drain current begins to decrease.
I hope this helps you a little bit to plan your modifications.
newbie
Activity: 28
Merit: 0
They surely mean the temperature inside the Chip at maximum, so if you measure the temp on the heatsink , it´s better to remain much lower than 100 C° or 212 F or 373 Kelvin.
One secure Outside Temp for long run is surley something about 70 degrees Celsius.But this can be realize only with a large Heatsink on the PCB and active cooling. Maybe is better to keep in mind , that one faulty "part" may cause more damage to other parts ,  for example on my PCB the AME8805 was faulty and gives me 5 V on the output , instead 3,3V ,in worse case the Stm Chip may die due the Overvoltage. So I recomend to be very carefull if you go to the limits. But when i count how much time and money for parts i invest in this Blades, i´m not sure if there will be ever a R.eturn O.f I.nvestment point ever. Maybe if we can sell this huge calculation power for other application , or as a heater for the winter. The only winners in this game are in my opinion only the producers, they can mine at first with this things at a time when this is rentable, and then they sell those things at very high pre order prices. If i spend this time as a taxi driver , i surely earn much more money..... Undecided And who cares about, that now I can repair some G-Blades , no one wants to have them , at the current prices.So much money and time lost.... Angry
Sorry about this , it´s out of topic and i know that this is the mining game , but it´s really annoying.
sr. member
Activity: 346
Merit: 260
If you Mean the 3 Mosfets on power Side,  the original ones from IRF (IOR), mostly all mosfets produced by IRF in SMD have the max operating Temperature up to 175 C°.
And for Soldering max 300 C° at 1.6mm from case for a maximum of 10 seconds. The better sollution is to replace the original mosfets with Dual Cool ones. Like the ones in my list or in the list from "J4bberwock".His list is really the best , if you want to go to the limits. This ones have 2 heat pads over/under , so the heat is better get outside the case. And adding a Fan for the powerside counts a lot.
But its highly recomended to replace the 16V Cap with a more serious one , if you don´t have.


Ok, thanks at around 6.6Mhs the max my mosfet chips got to was 143F peak on the
heatsink and dropped down to 139F once the unit stabalized at 1024 share difficulty.
I switched out my 16V cap and put in a 35V one instead, 0 ohm resistors in place of
the ferrite beads.
newbie
Activity: 28
Merit: 0
And for those who are interresed in, my blade still works  Grin

After some unexplicable USB errors , like cgminer has unexpected ended , now both blades works for 2 days. No problems no errors, so the soldering interventions seems to be succsessfull.The replaced Mosfet seems to work, the replaced AME8805 seems to work, and the STM32 ,too. They accumulate a lot of heat , but as i told , i apply a backwards fan, so i hope this is enough. In Germany are now 19 C° Outside and inside are 28 C° , even all my windows are open.And I only run the 2 Blades and an Antminer S1.
newbie
Activity: 28
Merit: 0
Quote
Ok so finally got the STM32 off, what a bugger to get those LQFP IC's off, eventually had to use a bigger nozzle on the hot air and heat from directly above and apply slight sideways pressure with the tweezers, came right off, now just need to clean the excess solder and wait for the new part to arrive.

Styson , are there some news about your fight against the stm32 ? And where did you get the firmware?
newbie
Activity: 28
Merit: 0
If you Mean the 3 Mosfets on power Side,  the original ones from IRF (IOR), mostly all mosfets produced by IRF in SMD have the max operating Temperature up to 175 C°.
And for Soldering max 300 C° at 1.6mm from case for a maximum of 10 seconds. The better sollution is to replace the original mosfets with Dual Cool ones. Like the ones in my list or in the list from "J4bberwock".His list is really the best , if you want to go to the limits. This ones have 2 heat pads over/under , so the heat is better get outside the case. And adding a Fan for the powerside counts a lot.
But its highly recomended to replace the 16V Cap with a more serious one , if you don´t have.
sr. member
Activity: 346
Merit: 260
The max I have turned the blade up to so far is 950Mhz @ 6.45Mhs and the
chips heat sink measured was around 142F and unit drawing around 220 watts.

Anyone know what is the max temperature for the 3 main power chip on the board?
Want to see how much room I have to crank it up a bit more, going to get a PCIe
cable to solder to the board directly tomorrow and ditch the stupid plug in jacks.
sr. member
Activity: 346
Merit: 260
I finally got a chance to take some time from work and modify my blade.
Here are some pics of the mods, temperatures and hash rate at the current
900Mhz freq.

A 10 turn variable pot was used to increase the resistance from the stock
33k to now 36k currently. The average hash rate is 6.125 and the temp
measured on the back side heat sink is around 107F and on the chips
heat sink it was around 126F.

In adding the larger heat sink on the chips I had to remove the plug and
connector on the board for the fan (on both boards) and solder directly to
the circuit board. I have now noticed that the power plugs are starting to
get warm so any further increase in freq/power I would solder directly a
PCIe female plug cable directly onto the circuit board.

All heat sinks where thermal epoxy mounted to the back board and the
chips to get max heat transfer. I would also look at adding larger finned
heat sinks to help keep the temps down. Also the lower voltage capacitor
was replaced with a 35V unit and raised off the board for better cooling.

Back Board Heat Sink


Chip Heat Sink & Adjustable 47k Pot (Set at 36k)


Thermal Probes both Heatsinks


Back Heatsink Temperature


Front Heatsink Temperature


Hashrate
sr. member
Activity: 412
Merit: 250
Did someone checked, blades are now 300$, price is really low but still if you buy 10 of them that is 50Mhash and for 3.000$ you soure get alot more than 50Mhash
newbie
Activity: 28
Merit: 0
Quote
As far as I know, the 5v USB circuit isn't related to the 12v power.

But if your ferrite AFB26 is damaged, it could be the reason why you have the failure.
You can replace it with a 0ohm resistor, at least for testing purpose.
I also have some spare ferrites. I can send you 1 or 2 from France for the cost of shipping, shouldn't be more than 1.5€

Ferrites are used as a filter for bad quality power source.
When shorted, they can act somehow like a fuse, even if they aren't very good at this. They often damage the surrounding components or the copper tracks when melting.

I'd also check your USB power source. Maybe something is wrong with its voltage.

Thank you for your suggestions and offer , I replace this with O-Ohm i try to see if it works for longer time ,i use a good PC - Powersource , so maybe there are no negative effects , i hope so.
newbie
Activity: 28
Merit: 0
Hi,

I have some news, too.
My ordered parts arrived now and i replaced them so far.

This is what i replaced :





And don´t be afraid to order from Mouser , if you are not in the USA . I was also a little bit afraid , after i had some bad experiece with my orders from USA to Germany,
but in this case it was really a dream! Just 2 days to arrive , no extra costs for customs or other fees. Even if i order directly in Germany this would not be faster, and in other shops it is really difficult to find the parts.

Some conclusions (for beginners)  Wink:

  • The upper Mosfet (this was absolutly the reason why the power source shut down - Thanks to Styson and Barssguy) was really hard to desolder with hot air , i have to remove the Cooler from the PCB and it takes a lot of time. Surprisely it was also hard to solder the new
    one , it fits perfect to the older size , but i was afraid to do this with hot air , not to damaged it , and i have to use a lot of flux but it was really difficult to apply it. i made it finally , not perfect but it works so far. Mybe to use soldering Paste which has a lower melt point isn´t a bad idea
  • Be very carfull not to scratch the PCB , when working with sharp tools, especially when the protecting surface is hot.
  • The Ame8805 was easy to solder , but everytime you have to clean with isoprpyl before apply and fix it with tweezers. You also have to be very carefully with the heat not to apply to long , some seconds more than expected over 300 C and you destroy it. And have in mind that this parts are very moisture-sensitive. in the bag is an indicator , if this is not blue you have to bake them. And also not apply to much solder !
  • For the AFB26 ( maybe this just a liitle bit melted part cause the fault to the AME8805 - thanks to J4bberwock) just remove the older one carfully not to damage the tracks under , in my case the tracks was not damaged from melting. And a 0- OHM Resistor ( i take the one with high power) works definitley. For a long time period i don´t know but i think it looks good at the moment.
    For those ones who has more real FB on the Board , replace them with 0 Ohm.It´s more secure , than to have the posibility to damage the PCB.My PCB was originaly send form Hong Kong and I have only 0- OHm Resistors in the Place for the FB´s directly form the Factory , just the AFB26 was a real FB.
  • The Capacitor was harder to desolder than i thought , with hot air , may the plastic part begin to melt , and again be carefull not to pull them to much.
    I don´t know if the one i choose ( 220uf 35 V ) is really the best choice , may some one has a better alternative?.
  • And don´t forget , it´s highly recomended to have some basic things like desoldering wire/stripe , a hot air station, and a Flux Pencil. Also try to remove the old solder that remain on the pads and apply new one which is recomended for SMD.It´s better than to damage more. This was my mistake too , so i thought : "Let´s start to solder SMD without any tools , just a soldering Iron." ---> Bad idea. After some impressions i highly recomend , if you are not a proffesional , to think twice , the things are not so easy than they appeard and just one wrong move and your investment is gone... the current prices for Scryptmining are sad enough , so don´t make them sadder.If you have a little bit experience in this things , so my recomend is not to hesitate and prepare everything calm, just a few days without mining are not so expensive like a valueless device. Also don´t forget to apply coolers on any important part, like Mosfets , Choke R , and backside of the PCB , I also apply a small cooler on the Capacitor which was original the 16V one and only to be absolutley sure one small cooler on the AME8805 from USB side.But the most important thing :  USE A FAN to cool the powersource part backwards! Just to show how important this is : my powerplugs ,i don´t apply screw terminal ones , are purley melted and destroyed under "normal" overclocked conditions with 39Kohm and 938 Mhz!!!!, just even if i used good copper stripes 2.5mm, so to replace this with screw terminals are much more better.Keep in mind that only with the front Fan , the airflow decrease significantly backwards , and it has allready accumulate all the heat from the cooler part , so you try to "cool" with "hot" air. This was not a big invention from Gridseed, under summer temperature , you may have heat problems und normal conditions , without any overclock. They try to compensate this with the more powerfull Fan in the front, but this is a loud and inefficient sollution. I use now 2 Arctic Fans with Temp Sensor, one in the front and the other one i mount backwards with screws on the original cooler, so the hot air from the front fan AND the power -part , is blowing away to the side.And if you apply the fan on the side , the Blades are able to place one above the other.



So i hope this time it works better and longer , after replacing the AME and the AFB26 the USB Communication works again.
If it works i hope that the price drop don´t continue like this , that i don´t have to use my 2 G Blades only for heating my home in the winter.

And thanks to all the guys ,who spend they time to made this thread.

Has anyone an idea how to opperate with cgminer 3.72 and win xp and not tho have to unplug the Blades form USB and Power when I change them Pool or freq?
I try the newest BFG Miner it has not this fault , but it don´t works so fine like CGMiner, after some times the stratum is interrupted.  



 
newbie
Activity: 20
Merit: 0
Ok so finally got the STM32 off, what a bugger to get those LQFP IC's off, eventually had to use a bigger nozzle on the hot air and heat from directly above and apply slight sideways pressure with the tweezers, came right off, now just need to clean the excess solder and wait for the new part to arrive.
hero member
Activity: 840
Merit: 1000
I asked myself how they separate the usbinput and the 12v input why this works yesterday only when i plug the 12V too? Maybe the AFB26 is like a separator? And what happens when i change this to a 0 Ohm resistor? Has anyone replace this with 0 Ohm? Normaly the communication part has to work when only the usb is plugged in. But when only the 12V is plugged in the STM must be unpowerd.


As far as I know, the 5v USB circuit isn't related to the 12v power.

But if your ferrite AFB26 is damaged, it could be the reason why you have the failure.
You can replace it with a 0ohm resistor, at least for testing purpose.
I also have some spare ferrites. I can send you 1 or 2 from France for the cost of shipping, shouldn't be more than 1.5€

Ferrites are used as a filter for bad quality power source.
When shorted, they can act somehow like a fuse, even if they aren't very good at this. They often damage the surrounding components or the copper tracks when melting.

I'd also check your USB power source. Maybe something is wrong with its voltage.
newbie
Activity: 28
Merit: 0
I asked myself how they separate the usbinput and the 12v input why this works yesterday only when i plug the 12V too? Maybe the AFB26 is like a separator? And what happens when i change this to a 0 Ohm resistor? Has anyone replace this with 0 Ohm? Normaly the communication part has to work when only the usb is plugged in. But when only the 12V is plugged in the STM must be unpowerd.
newbie
Activity: 28
Merit: 0
Hi,

My AME is dead again, it works for 25 hours perfectly , then i change the usb drivers to try bfgminer insted cgminer , and it works for 30 seconds and than again bfgminer detect a dead device. I switch again the drivers to try with cgminer again but it still remains dead. Really annoying . It seems that the AME8805 is broken again , when i plug only the usb , i have of course the unknown device and input on AME 2,88V and output 2,88 V , no shorten pins without voltage. When i Plug the 12V too , then the input rise to 5 V but the output is 5V too, instead 3,3V and the red LED lights permanently. So no comunication through usb.

Any ideas ? 

Ok maybe the AME are  a little bit sensible but I think it must be something with the usb unpluging and pluging. Because this happens when i unplug usb to change the driver , and when i plug in again.This is beacause i work with windows and cgminer requires to unplug the device compleatly and plug again in ,when i change the pool or the freq.  Maybe the AFB26 ? This part is just a little bit melted , but it seems to work . I don´t really know the function of this but maybe it cause some overvoltage or something like this? Or the capacitor for the AME between pin 2-3? How to check this?  For this time i don´t use any Usb Hub only the device directly , so a faulty usb hub can´t be the reason.
Any suggestions are welcome... Smiley


 
newbie
Activity: 28
Merit: 0
Quote
Looks like the STM32 is shorted, so have to wait for a replacement, and then hopefully I can get the firmware on there  Huh

I wish you good luck that it works. This is really annoying... Undecided
newbie
Activity: 28
Merit: 0
I finally wake up to life one PCB! Grin

The parts was send by mouser with fedex and amazing fast just 2 days from the USA to Germany ! But i wasn´t home... Sad
So I can´t wait any longer and because my hotair station was allready there I try again to replace the AME8805. But I have only the other ones from the lower board , so i try to remove one from the board with the broken upper mosfet. The first mistake that i made was not to remove the PCB from the coller , because the lower ones are on the side where the cooler is ... so i wonder why this hot air don´t works  Shocked Roll Eyes , all the heat goes instantly to the cooling part... So i removed the PCB carfully , just a little bit and with the hot air finally it works , not so easy that i thought , this was my first time with hot air , but it works... So maybe it would be better to do some exercise befor try on things that are more expensive , but i just can´t wait...
I´ve made to little scratches on the PCB , but I remove this f---ing AME.
And this time it seems not to be broken after the heat. So with little problems , i finally place the AME on the right place, but then the next missfunction, i plug the usb and do some measure ...and again -- unkonown device, and Input form USB was 2.88 V and Output , again 2.88 V instead 3.3V ,my first thought was that there was more burned parts , or i made wrong soldering , but just to give them a try i plug the 12V Input  too, and voilá .... the STM  Virtual COM Port appears on the computer and the AME begun to work correctly input 5V -- Output 3.3V. So maybe if i don´t try the 12V too, i never find out that it works. If some one else has the same , try to plug the 12V too!
Now i´m carfullier and set only 875 Mhz , and it works now for an hour , but still with a little bit to much HW´s , 8 % but i am happy when it works , the HW´s was too high from the beginning , so maybe there are some faulty chips...  Hope that it still works tomorrow..

Now i have to wait until i have the parts and try to reanimate the second pcb,too. I am a little bit afraid that i don´t handle to remove the upper mosfet but I hope i´m able to do this. And now I have to apply two AME8805s , but i think to solder them ist easier than to desolder. And I have tor replace the AFB26 ,too. I will try this with a 0 Ohm resistor.If the second PCB works , after replace this 4 Parts  , i let you guys know.
Greetings
Pages:
Jump to: