GRIDSEED G-BLADE Overclocking 7Mh/s, improvements and repair

[styson]

Hey Freeman,

The Larger size FB's are Ferrite Beads, 300 ohm 100Mhz 3A 1206, the smaller ones are 300 ohm 100Mhz 1.5A 0805

A lot of the guys just bypass the ferrites with a 0 ohm resistor or bridge with solder, that's if you have a descent power supply.

Just also check that the blown ferrite beads have not damage the tracks beneath them.

[Freeman2000yes]

Hi,

Thanks for your answer,

When I understand right , I can replace AFB26 with a 0 Ohm resistor. Thanks now now i can see clearer : "FB" stands for "F"errite "B"eads , only the "A" has irritate me ...
I was a little bit confused , because all the other FB ´s on my PCB are allready 0 Ohm Resistors form Factory (like Fb 27 -29 and 23-25 ) only this AFB 26 was different.
So this was a good decision that i order some 0 Ohm resistors in different sizes.
Thanks again from Germany and I hope my Blades will work again i am able to replace the upper mosfet i order a hot air solderstation expecially for this.
I have ordered my Blades from Zoomhash one Overclocked one normal  , which I overclocked by myself.  I have a lot of trouble with delay time to germany and I my resume is , that i better save my money for the overclocked version next time , the Resistor was unprofessional placed and they forget to place the cooling pads there was only some glue but the coolers was gone. I have to complete this by myself. And the "normal" one was very fast from Honkong to Germany just 2 Days! but it was a little bit faulty form the beginning something like 7% HW on 800 Mhz!. Then I have apply the 39Kohm Vmod by myself and it works fine but again with 5% Hw at 950 Mhz. But I warn anybody who Overclock! The 39 kohm is surley a saver one then above 40 -- and of course I placed a PGA Cooler on the back of the PCB and also on the front dirctly to the Mosfets and chocke. But there is a huge difference between Standard 33 kohm and 39 kohm , this was the reason why i felt "save" , with the 33 kohm the blade was only hand warm , very few heat production even though the high summer temeratures but with the 39 kohm and 938 Mhz it was a huge difference. So now i´m wiser and i strongly recomend to work with active coolers on the back. I simlpy screw the old fan from a 5 chip girdseed.
http://

[Freeman2000yes]

https://www.dropbox.com/s/woh3ajsv164scmr/DVCI0162.JPG
https://www.dropbox.com/s/xzov5dp2tic7oyp/DVCI0163.JPG

hm , i don´t know why the image attach don´t work

[Freeman2000yes]

One more question i forgot:

Is the burned upper Mosfet the reason why i have a short circuit when i plug the PCB to 12 V? There are sometihing like 50 Ohm between the measure points , so its completly far away from infinite and definitly damaged, but is this the reason why the power source goes instantly down?  Or maybe there are more burned parts ?

[Brassguy]

One more question i forgot:

Is the burned upper Mosfet the reason why i have a short circuit when i plug the PCB to 12 V? There are sometihing like 50 Ohm between the measure points , so its completly far away from infinite and definitly damaged, but is this the reason why the power source goes instantly down?  Or maybe there are more burned parts ?

Probably. Power supply is detecting a dead short and shutting down.

[styson]

One more question i forgot:

Is the burned upper Mosfet the reason why i have a short circuit when i plug the PCB to 12 V? There are sometihing like 50 Ohm between the measure points , so its completly far away from infinite and definitly damaged, but is this the reason why the power source goes instantly down?  Or maybe there are more burned parts ?

Yes it's definitely the reason for the power supply going off I had the same happen to my blade, replace the upper mosfet with CSD16321Q5

Also be carefully to not pull the mosfet off, let the hot air station reach temperature for a while, then apply heat to the left of it, it should be able to slide right off, do not pull it, if the solder is not molten it will pull the tracks from the board and cause damage. ;-)

[styson]

Ok so I removed the ame8805 by the USB and there is still a short between 3.3v and ground, any idea's ?

Will be check the STM32 for shorts tonight 

[Freeman2000yes]

Yes it's definitely the reason for the power supply going off I had the same happen to my blade, replace the upper mosfet with CSD16321Q5

Also be carefully to not pull the mosfet off, let the hot air station reach temperature for a while, then apply heat to the left of it, it should be able to slide right off, do not pull it, if the solder is not molten it will pull the tracks from the board and cause damage. ;-)

Thank you so much for the information. I have to wait now for the parts , I think it´s a little bit tricky to change the upper mosfet , because there are the 2 small parts , mabe capacitors , so i have to be very carefull to not blown them away... Did you cover the other parts to protect them or you apply the hot air whithout any "preparations"? What´s your exeperience for how long to apply the air 3-5 minutes?
 Thanks

Ok so I removed the ame8805 by the USB and there is still a short between 3.3v and ground, any idea's ?

Will be check the STM32 for shorts tonight  Huh

I also try to replace the Ame8805 from the other PCB , because it seems to be working there. But i think this parts are really sensible to hight temps, i don´t wait for the hot air station and i have tryed with a soldering iron, but it was not so easy because the whole body is solded on the pcb. The replace was successfully , but the high temp damaged the Ame8805. Before replacing this i have with the old one : Input 5V Output 1,2 V and the problem with unknown device, i think thats because the low voltage, after replace 5 V input , 0.1V Output, so this fails , i think with hot air this works better. Maybe you can try to find out if there be any shorts between the pins.there is a very small place between the big pad and the pin pads.So the space between the pins may look ok but under the part itself may be a little solder bridge to the gnd pin,it would be better to desolder the part again and check whithout them if there is still a short and be carefull , if there is to  much solder on the big pad it owerflows when you apply the AME8805 and maybe it cause a short to a pin pad, the problem is this is invisible because the bridge is under the part. I´ve checked this before i plug the usb and measure between pin 1-2 and 3-2 if there are any shorts. Between pin 1 and 3 must be something like infinte Ohm, my broken one has 25  Ohm, so i think the high temp from the soldering iron damaged it. Maybe you can try also to find some diferences with the working Ame8805 on the lower board there are 5 or 6 from  them.  And after all this has a build in overtemperature cutoff and over current, but this resets after the temperture lows

Here the description from Datasheet:

The AME8805/8810 family of CMOS regulators con-
tain a PMOS pass transistor, voltage reference, error am-
plifier, over-current protection, and thermal shutdown.
The P-channel pass transistor receives data from the
error amplifier, over-current shutdown, and thermal pro-
tection circuits.  During normal operation, the error am-
plifier compares the output voltage to a precision refer-
ence.  Over-current and Thermal shutdown circuits be-
come  active  when  the  junction  temperature  exceeds
150
o
C, or the current exceeds 600mA.  During thermal
shutdown, the output voltage remains low.  Normal op-
eration is restored when the junction temperature drops
below 120
o
C.
The AME8805/8810 switches from voltage mode to cur-
rent mode when the load exceeds the rated output cur-
rent.  This prevents over-stress.  The AME8805/8810 also
incorporates current foldback to reduce power dissipa-
tion when the output is short circuited.  This feature be-
comes active when the output drops below 0.8 volts, and
reduces the current flow by 65%.  Full current is restored
when the voltage exceeds 0.8 volts.

[Freeman2000yes]

Ok i overread that you only remove the AME , so there is a short whithout the part , sorry.
Maybe the caps for the AME cause the short?

But maybe this above helps someone else...

[styson]

Yes the 2 parts to the right of the upper mosfet are capacitors, that's why it is easier to apply the heat from the left side pins, and lift carefully with tweezers from the left side, the solder on the pad should flow quickly and you can take the part off without blowing the capicitors off. The air should not be applied for long, I could not find anything in the datasheet though about the temperature and time you can apply the heat for. for the AME8805 it can be 300 degrees celsius for 10 seconds.

Thanks for your suggestions, will keep them in mind.

Looks like the STM32 is shorted, so have to wait for a replacement, and then hopefully I can get the firmware on there 

[Freeman2000yes]

I finally wake up to life one PCB!

The parts was send by mouser with fedex and amazing fast just 2 days from the USA to Germany ! But i wasn´t home...
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  , 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

[Freeman2000yes]

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...

[Freeman2000yes]

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...


 

[Freeman2000yes]

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.

[J4bberwock]

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.

[styson]

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.

[Freeman2000yes]

Hi,

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

This is what i replaced :

• For the AME8805 : This one ---> AP1115BY33L-13 Link : http://de.mouser.com/Search/ProductDetail.aspx?R=AP1115BY33L-13virtualkey62110000virtualkey621-AP1115BY33L-13
• For the Upper Mosfet ( IOR 5302 C349P LL2C) : This ----> CSD16321Q5C (DualCool N-CH) Link: http://de.mouser.com/Search/ProductDetail.aspx?R=CSD16321Q5Cvirtualkey59500000virtualkey595-CSD16321Q5C
  
• For the AFB26 Ferrite Bead :  ----> Zero - Ohm Resistor High power Size 0805 CRCW08050000Z0EAHP Link: http://de.mouser.com/Search/ProductDetail.aspx?R=CRCW08050000Z0EAHPvirtualkey61300000virtualkey71-CRCW08050000Z0EAH
  
• For the Capacitor 220uf 16V (this wasn´t damaged , but just to be secure i replaced them too) : -----> SMD Cap 220uf  35 V (This one is higher, but fits perfect) Link : http://de.mouser.com/Search/ProductDetail.aspx?R=PCR1V221MCL1GSvirtualkey64700000virtualkey647-PCR1V221MCL1GS
  

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)  :

• 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.  



 

[Freeman2000yes]

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.

[Blisk]

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

[racebyu]

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