Topic: GRIDSEED G-BLADE Overclocking 7Mh/s, improvements and repair - page 13. (Read 74041 times)
Q9 is added but it is not there
As Requested: 
I just got a new blade in today, and the board design has changed again. They added solder pads in for a 4th mosfet, but no chip, and added a 2nd red LED that seems to just come on with 12v. Maybe some headroom can be had by adding in the missing mosfet. Seems odd that they would redesign the board with that pad, but then not actually solder one in.
post some picturesI just got a new blade in today, and the board design has changed again. They added solder pads in for a 4th mosfet, but no chip, and added a 2nd red LED that seems to just come on with 12v. Maybe some headroom can be had by adding in the missing mosfet. Seems odd that they would redesign the board with that pad, but then not actually solder one in.
post some pictures
I just got a new blade in today, and the board design has changed again. They added solder pads in for a 4th mosfet, but no chip, and added a 2nd red LED that seems to just come on with 12v. Maybe some headroom can be had by adding in the missing mosfet. Seems odd that they would redesign the board with that pad, but then not actually solder one in.
The stock fan is rated at 60 CFM on the blades.
Will anyone possibly link some of the parts you purchased to volt mod your g-blades in this thread?
It was 43k stable for weeks, main fan failed during the night.
The ferrites aren't the direct culprit. the upper gate mosfet fails, shorting gate to drain, causing the ferrite to blow, if I'm not mistaken.
It also happens on the pods from time to time, so it shouldn't be the load.
upper gate mosfets should have been doubled as we can see on the initial schematic, but it saves a few $$ when manufacturing.
I suspect that with summer temperature, we will see much more blown mosfets on pods especially those overclocked with 5v fan mod.
When T° rises, the efficiency of the mosfets drops, causing more heat to be dissipated, causing the mosfet to short itself.
Do you have any CFM specs for the fan? Seems prudent to replace it with a better model before it fails. How has the choke held out on your board? On my blades, the choke is getting pretty warm alongside the mosfets.
I managed to give a blade artificial breathing . pictures of Frankenblade soon.
anxious await pics of frankenblade!
and any others of your creations, would be very helpful.
Actually, wires from the pods to the blade are quite hot when using dual pods with 56k resistors to power the blade. Only 1.4v stable instead of 2v on the pods. Blade pcb hashing 3.1MH/s 70/80 watt used
Single pod 49.9k was stable around 1v and not hot, blade pcb hashing 2MH/s.
I managed to give a blade artificial breathing . pictures of Frankenblade soon.
anxious await pics of frankenblade!
and any others of your creations, would be very helpful.
More fun facts the upper fet is (AQ6) 16321C, lowers (AQ7 & AQ8) are CSD17556.
Just a question. I have 2 blades (4 boards) running at 838mhz. It's a good problem to have but they are only drawing about 200 watts total (~50w per board) at the wall... doesn't that seem low? The hashrate at the pool is ~11.2mh.
my blade draw about that with seasonic psu .. measure at wall on killerwatt probably morelso 110 for the blade before killed it
Just a question. I have 2 blades (4 boards) running at 838mhz. It's a good problem to have but they are only drawing about 200 watts total (~50w per board) at the wall... doesn't that seem low? The hashrate at the pool is ~11.2mh.
I don't know how it's possible to draw 50w for 80 chips at this speed, but don't touch anything, let's say you are lucky.edit... you meant 50w per 40 chips right?
It doesn't make any sense.... I have an undervolted antminer s1 using about ~215w and 2 (4 boards) blades on one psu... total usage 411 watts as we speak via my Kill-A-Watt... 11.4mh shown via cgminer 11.5mh showing at pool, so the hashrate is about the same locally and poolside...
edit I'll investigate further before I overvolt/overclock these...
yes, 40 chips.
edited.
Just a question. I have 2 blades (4 boards) running at 838mhz. It's a good problem to have but they are only drawing about 200 watts total (~50w per board) at the wall... doesn't that seem low? The hashrate at the pool is ~11.2mh.
I don't know how it's possible to draw 50w for 80 chips at this speed, but don't touch anything, let's say you are lucky.edit... you meant 50w per 40 chips right?
It doesn't make any sense.... I have an undervolted antminer s1 using about ~215w and 2 (4 boards) blades on one psu... total usage 411 watts as we speak via my Kill-A-Watt... 11.4mh shown via cgminer 11.5mh showing at pool, so the hashrate is about the same locally and poolside...
edit I'll investigate further before I overvolt/overclock these...
Just a question. I have 2 blades (4 boards) running at 838mhz. It's a good problem to have but they are only drawing about 200 watts total (~50w per board) at the wall... doesn't that seem low? The hashrate at the pool is ~11.2mh.
I don't know how it's possible to draw 50w for 40 chips at this speed, but don't touch anything, let's say you are lucky.
Just a question. I have 2 blades (4 boards) running at 838mhz. It's a good problem to have but they are only drawing about 200 watts total (~50w per board) at the wall... doesn't that seem low? The hashrate at the pool is ~11.2mh.
that's why I will build in thermal switch if fan fails it will turn off mining
95 oC max
95 oC max
I'll do it too.
I managed to give a blade artificial breathing . pictures of Frankenblade soon.
A pod is actually providing the power to the blade and it's hashing, but I need to fix the voltage drop. From 1.3 it's dropping to 0.95 when hashing starts. 49.9k resistor user on the pods. 33k gives a voltage drop to 0,75.
BAUAR, can you check that you have the 5v output from the up1707 and 3.3v output from the ame8805?
My 5v was around 2,4v on one pcb. And output at 8805 was 0.1v.
There are a few pin to pin compatible replacement available for 8805 if it's fried. I haven't checked for up1707 yet,
If you can't track the defect, you can still use external power. Output from up1707 on a pods goes to large tab on any of the 5 8805 user for each row of chips. Disconnect the c42 capacitor from the pod and from the blade. Connect both "+" together and both "-" together.
I also removed up0111 from the pods, it gives avdd and dvdd pll, so I expect the chips from the pods won't draw any power.
49k resistor will give 1v to the blade chips. I'll try 56k tomorrow.
If helps revive your blade, you can donate btc/ltc
If it doesn't, you can still send the blade to me for free
that's why I will build in thermal switch if fan fails it will turn off mining
95 oC max
95 oC max
Last update, replacing ferrite with resistors might not be such a good idea.
On one of my blades, the power mosfet has shorted (as usual), but there were resistors.
The 3 ones for the +12V are blown, as is my 470uF 25V capacitor, and probably other components, since even after replacing the fet, resistors and capacitor, the blade isn't hashing. I'm currently tracking the damages and will report if it happens to someone else.
SMD Ferrite permanently damages the PCB when blowing because they are simply melting and they also melt the PCB board/tracks under them.
What voltage/resistor combo were you running when this happened? Did you have active cooling on both sides of the ferrite bead? The 470uF 25v replaced the 220uF 16v?
Using a larger size cap causes somewhat higher current draw during it's charge phase, and may exacerbate the problem. If the charge phase is the same length of time and the capacity is double, it will pull twice as much current to charge it in the same period of time. You might perhaps be better off running (2) 110uF caps in parallel. Same capacitance, half current draw per cap & twice as much surface area for heat dissipation. That should cool them down considerably.
It was 43k stable for weeks, main fan failed during the night.
The ferrites aren't the direct culprit. the upper gate mosfet fails, shorting gate to drain, causing the ferrite to blow, if I'm not mistaken.
It also happens on the pods from time to time, so it shouldn't be the load.
upper gate mosfets should have been doubled as we can see on the initial schematic, but it saves a few $$ when manufacturing.
I suspect that with summer temperature, we will see much more blown mosfets on pods especially those overclocked with 5v fan mod.
When T° rises, the efficiency of the mosfets drops, causing more heat to be dissipated, causing the mosfet to short itself.
better is 220uF and 35V or 2x 100uF
Jump to: