Pages:
Author

Topic: GRIDSEED G-BLADE Overclocking 7Mh/s, improvements and repair - page 8. (Read 74040 times)

newbie
Activity: 20
Merit: 0
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  Huh
newbie
Activity: 28
Merit: 0
Ok i overread that you only remove the AME , so there is a short whithout the part , sorry. Cheesy
Maybe the caps for the AME cause the short?

But maybe this above helps someone else...
newbie
Activity: 28
Merit: 0
Quote
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

Quote
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.
newbie
Activity: 20
Merit: 0
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
newbie
Activity: 20
Merit: 0
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. ;-)

full member
Activity: 139
Merit: 100
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.
newbie
Activity: 28
Merit: 0
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 ?
newbie
Activity: 28
Merit: 0
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 Roll Eyes , only the "A" has irritate me ... Huh
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://
newbie
Activity: 20
Merit: 0
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.
newbie
Activity: 28
Merit: 0
I´ve checked some measure points with the excellent guides from this thread and find you that not only the afb26 was burned... Sad
On one PCB the AU5 VR is dead , Input from USB is 5 Volt but on V OUT there are only 1,2 Volt , so i think this can be the cause , why the PCB seems to work when 12 V is pluged in ( The Chips get warm and all the other VR works fine) but only a unknow device appear when plug in to the computer.
On the another PCB the upper Mosfet seem to be broken , there are something about 50Ohm between the measure points and this thing named AFB26 near by the USB Plug is exploded.
On the first PCB the afb26 is only a little bit melted , but it seems to work, with unplugged board it has a 0.8 kOhm resistance , when the board is pluged in , the resistance goes down an fluctuate from 0-600 Ohm. May this can be a NTC Thermistor or a Varistor???
I ordered now some parts from Mouser , like the mosfets and some 0 Ohm Resistors and the broken VR AU5 or A8805 and also some NTC Thermistors for the AFB26 , but I am not sure if this was really a NTC.
Is there any posibilty to find out what kind of part this is? Maybe from a working PCB? I think its possible that figures like a fuse or something like that so a NTC make sense. But i´m not sure if is possible to replace that part with a 0 OHM resistor?
Or I´m compleatly wrong?

Thanks.
newbie
Activity: 28
Merit: 0
Hello guys, Smiley

I need urgently some help from anyone,

One of my Gblades ( overclocked with Vmode 39K Resistor) , burned up instantly , after it works fine for some days at 950 Mhz .
It cause a short circuit and my powersource goes down when i try to replug the Gblade.
My powered Usb hub was broken too, after the fault .
When i opened the case i saw that the AFB26 directly on the right side of the USB Plug was burned down, the half of them was burned down.
The problem is that I have no idea what kind of part this is , a resistor ? diode?
Maybe with a stroke of luck this is the only broken part and if i replace this , maybe it works again. I think my Gblade send some overvoltage through the usb port because my hub is dead too, or mybe the Hub was faulty ....
It would be nice if someone can tell me what kind of part the AFB26 is...
Did anyone have the same issue?

Thank you in advance and sorry for my bad english. Cheesy
member
Activity: 88
Merit: 10

This is a "big" smd part, so even with a fine tip iron, you can easily replace it.
The harder thing will probably be to remoove the failed one.

AME8805AEFT (the AEFT thing tells the output voltage and package, so it needs to be AEFT) can be ordered from aliexpress/alibaba, hard to find for me in usual shops.
datasheet:
http://pdf.datasheetcatalog.com/datasheet/ame/ame8805.pdf

It could be replaced with the following pin to pin compatible, but not tested (it took some time to dig them out)
some of them are available from mouser

AP7215-33YG-13
AP1115BY33
AX1113BF
AX6647-33-E
AX6647-33-F

When ordering, check that these are SOT89 package with the following pinout:

pin 1 is ground
pin 2 is the input (center tab) +5V
pin 3 is the output +3.3v

Thanks for the infor J4bber, really appreciate the part comparison. I went ahead and ordered a few AP7215-33YG-13 from Mouser, and for the hell of it I ordered 50x AME8805AEFT from Utsource (couldn't pass up a good deal!).

Unfortunately (or maybe fortunately) the seller decided to give me a refund on the blade if I ship it back to him, so I now no longer need these chips. Oh well, at least if someone else needs them, I'll have stock in the US in a few weeks.

Thanks again for the help, and if anyone needs any AME8805AEFT's in the US, shoot me a PM.
newbie
Activity: 20
Merit: 0

here is the picture with what you should measure.

You simply need to plug the USB, nothing else


you can use any ground point on the board. the easier one to use is this one. square pad in the connector.


Thanks J4bberwock! I tested the chip as described, and on the non-working blade I do get 5V on the input, but no 3.3V on the output. Just a 0v read. Cry

The working blade shows the correct 5V input and 3.3V output.

I'm assuming this means that the AME8805 is fried. I'm certainly no good at SMD soldering, but do you think I could have this replaced by a pro? Could this failure have caused damage to any of the other components?

Thanks for any other info. Hopefully I can get half my money back from the seller.

Just make sure that you use a high enough wattage soldering iron and be careful not to force the AME8805 off the board or you can rip the tracks up, best way to do this kind of work is with a hot air station, most cellular phone repair people have them and they can replace the part for you in 5 minutes.
sr. member
Activity: 412
Merit: 250

This is a "big" smd part, so even with a fine tip iron, you can easily replace it.
The harder thing will probably be to remoove the failed one.

AME8805AEFT (the AEFT thing tells the output voltage and package, so it needs to be AEFT) can be ordered from aliexpress/alibaba, hard to find for me in usual shops.
datasheet:
http://pdf.datasheetcatalog.com/datasheet/ame/ame8805.pdf

It could be replaced with the following pin to pin compatible, but not tested (it took some time to dig them out)
some of them are available from mouser

AP7215-33YG-13
AP1115BY33
AX1113BF
AX6647-33-E
AX6647-33-F

When ordering, check that these are SOT89 package with the following pinout:

pin 1 is ground
pin 2 is the input (center tab) +5V
pin 3 is the output +3.3v

Did you test those TO220 does it work?
newbie
Activity: 20
Merit: 0
Definitely the AME8808 that has blown on my board, going to order some AP1115BY33-13 this week and see if they work, have checked the datasheet and definitely are pin for pin compatible.
hero member
Activity: 840
Merit: 1000

here is the picture with what you should measure.

You simply need to plug the USB, nothing else


you can use any ground point on the board. the easier one to use is this one. square pad in the connector.


Thanks J4bberwock! I tested the chip as described, and on the non-working blade I do get 5V on the input, but no 3.3V on the output. Just a 0v read. Cry

The working blade shows the correct 5V input and 3.3V output.

I'm assuming this means that the AME8805 is fried. I'm certainly no good at SMD soldering, but do you think I could have this replaced by a pro? Could this failure have caused damage to any of the other components?

Thanks for any other info. Hopefully I can get half my money back from the seller.

This is a "big" smd part, so even with a fine tip iron, you can easily replace it.
The harder thing will probably be to remoove the failed one.

AME8805AEFT (the AEFT thing tells the output voltage and package, so it needs to be AEFT) can be ordered from aliexpress/alibaba, hard to find for me in usual shops.
datasheet:
http://pdf.datasheetcatalog.com/datasheet/ame/ame8805.pdf

It could be replaced with the following pin to pin compatible, but not tested (it took some time to dig them out)
some of them are available from mouser

AP7215-33YG-13
AP1115BY33
AX1113BF
AX6647-33-E
AX6647-33-F

When ordering, check that these are SOT89 package with the following pinout:

pin 1 is ground
pin 2 is the input (center tab) +5V
pin 3 is the output +3.3v
member
Activity: 88
Merit: 10

here is the picture with what you should measure.

You simply need to plug the USB, nothing else


you can use any ground point on the board. the easier one to use is this one. square pad in the connector.


Thanks J4bberwock! I tested the chip as described, and on the non-working blade I do get 5V on the input, but no 3.3V on the output. Just a 0v read. Cry

The working blade shows the correct 5V input and 3.3V output.

I'm assuming this means that the AME8805 is fried. I'm certainly no good at SMD soldering, but do you think I could have this replaced by a pro? Could this failure have caused damage to any of the other components?

Thanks for any other info. Hopefully I can get half my money back from the seller.
hero member
Activity: 840
Merit: 1000
J4bberwock,

Can you please confirm if the REGULATED voltage rails on the blades are just 5V and 3.3V ONLY?

I have not seen any other mentioned voltages; other than 5V and 3.3V.

What I like to do is to jumper in an ATX PSU and be done with all that volt-modding.

1. remove any on-board regulation, i.e. remove any in-series components.

2. Solder in screw-in terminals after the components


Do you think that would work?

ASIC chips voltage is 1.2V at stock...60-80A ...

Voltmod is rising it to 1.35 to 1.75V ...

Check the schematics first...then re-evaluate your "idea(s)"... Grin

ZiG
Couldn't have said it better.
You could bypass the 3.3 and 5v regulation, but you still need something for the 1.2-1.75v with high amps.
newbie
Activity: 20
Merit: 0
My one blade died, thought it was the upper fet which I replaced, but now when I connect the power supply I hear a buzzing noise from the power connector and the power supply goes off, have checked the boards for shorts and nothing I can spot, any idea's ?

maybe the 16v 220uf Capacitor.
I suppose you also replaced the ferrites/0 ohm resistors that burned with the upper gate mosfet.
There are very few components between the power plug and the mosfets, so they are easy to track:
Plug -> ferrites -> capacitors -> mosfets

You can check the 8 chips schematics on github for some help. It's almost the same as what we have on the blade.

I posted a few test points you can check in this thread. It can help finding the origin of the trouble.

Was a dogdy FET I replaced with, power now works, but looks like the Micro controller is fried now, does anyone have the firmware for the STM32F103RCT6 ?

Could also be that regulator that you guys are talking about, will have to check that tonight.
Pages:
Jump to: