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

[nemercry]

Did anyone check if these companies which sell those scrypt machines use the same chips or not?
I think they do. And if they can buy it, we can do it also.
 How hard is to made miner like that when you get chips, using directly raspberry pi on it.

you still need the firmware to communicate with the chips.
The electronic part can be done, but might not worth it if you only plan to do a few.

http://www.innosilicon.com/A2.htm

200$ each chip if you order 50 pieces
roughly 10000$ = the actual price for the complete miner with 48 chips

these chips are too expencive and I am soure you can get cheapper one.
And if more people takes it we get discount also.

We need to search cheapper chips

I do not think Gridseed will make these chips anymore as they did sign a
deal with innosilicon

Gridseed GC3355 chips would cost nothing to buy now. Could worth the try, plus the firmware and the design to be improved already exists.
It all depends if they sell the chips and if they sell for really cheap.
They aren't that power hungry to move 28nm now. simply marketing.

I do not think that Gridseed will continue to manufacture the 3355 chips as they
could be signing a deal with Innosilicon to manufacture the A2 chips themselves now.
Good move on both parties as if they can ramp up production at both ends they
will take out most of the late to market competition.

If you can get 4Ghs out of 40 chips but draw approx 300w of power you would
only be able to obtain 20Ghs max on a 115vac@15amp service outlet. That is
getting the max (1750w) of a standard wall outlet, you are putting out 1500 watts.
The volt mod with higher freq settings pushing the chips really drives up the power
consumption.

These units are the first version on the ASIC Script miners and will draw the most
power thus making them not efficient enough as higher power miners. That being
said I have seen some company putting 6 stock freq units into a custom rack and
selling as a 33Ghs unit drawing 840 watts but price was way to high and around the  
same price of a newer 28nm 90Mhs miners using the A2 chips.

Good work to everyone inputing into this thread, I am going to mod my Blade soon
with the knowledge in this thread. Next round of blades coming out next week are in
the $900 range now!!

your power usage looks way off.

My blade running overvolted at 7.2MH/s uses around 300w from the wall
5.8MH uses 170 watt from the wall (30 watt just for the psu at idle nothing plugged)

we should be able to reach around 56MH with 1500 watt
price for this have to be no more than 800$ for 14MH to be competitive,
flexible design by multiples of 14MH (200 chips) up to 42MH 1200 watt to easily power it with server PSU like the DL580G3 1300 watt (noisy but efficient)

Does sound reasonable, all in all.

[1715194262]

1715194262

[1715194262]

1715194262

[1715194262]

1715194262

[Blisk]

Does anybody think about adding chips to gridseeds with 5 chips?
I think that is cheappest way to make more hashrate

[wolfey2014]

Does anybody think about adding chips to gridseeds with 5 chips?
I think that is cheappest way to make more hashrate

I haven't looked at the pcb drawings yet but I have been wondering the same thing. Can chips simply be stacked on top of existing chips on the 5 chip USB miner. I would think that running Scrypt would not cause them to cook each other but I could be wrong. That would be something eh? Turn it into a 1GHs+ overclocked little monster. I know it would run hotter and pull twice the current but I don't think it would hit 60W. Doing this with SHA would definitely be self defeating if not a Muth Buster's style destructive test
I don't know if the pins on them would be long enough to lay over and connect on top of the pcb soldered pins. Probably not!

Woof Woof!

[J4bberwock]

Does anybody think about adding chips to gridseeds with 5 chips?
I think that is cheappest way to make more hashrate

not enough space to connect new chips even if they are quite easy to daisy chain and you'll have to find a way to cool them.
At least, not a good choice except as a proof of concept.
a new design will be better and better components/design can be used for the powerstage
Even the casing of a blade could have been more efficient.

[J4bberwock]

Does anybody think about adding chips to gridseeds with 5 chips?
I think that is cheappest way to make more hashrate

I haven't looked at the pcb drawings yet but I have been wondering the same thing. Can chips simply be stacked on top of existing chips on the 5 chip USB miner. I would think that running Scrypt would not cause them to cook each other but I could be wrong. That would be something eh? Turn it into a 1GHs+ overclocked little monster. I know it would run hotter and pull twice the current but I don't think it would hit 60W. Doing this with SHA would definitely be self defeating if not a Muth Buster's style destructive test
I don't know if the pins on them would be long enough to lay over and connect on top of the pcb soldered pins. Probably not!

Woof Woof!

Don't forget that most of the heat is dissipated by the bottom of the chips.

And you can't simply stack them. they need individual lines to be accessed

[Blisk]

Does anybody think about adding chips to gridseeds with 5 chips?
I think that is cheappest way to make more hashrate

not enough space to connect new chips even if they are quite easy to daisy chain and you'll have to find a way to cool them.
At least, not a good choice except as a proof of concept.
a new design will be better and better components/design can be used for the powerstage
Even the casing of a blade could have been more efficient.

there is enough power for 30 more chips I think and what if you connect it with wires and another heatsink for those 30 chips?

[racebyu]

Did anyone check if these companies which sell those scrypt machines use the same chips or not?
I think they do. And if they can buy it, we can do it also.
 How hard is to made miner like that when you get chips, using directly raspberry pi on it.

you still need the firmware to communicate with the chips.
The electronic part can be done, but might not worth it if you only plan to do a few.

http://www.innosilicon.com/A2.htm

200$ each chip if you order 50 pieces
roughly 10000$ = the actual price for the complete miner with 48 chips

these chips are too expencive and I am soure you can get cheapper one.
And if more people takes it we get discount also.

We need to search cheapper chips

I do not think Gridseed will make these chips anymore as they did sign a
deal with innosilicon

Gridseed GC3355 chips would cost nothing to buy now. Could worth the try, plus the firmware and the design to be improved already exists.
It all depends if they sell the chips and if they sell for really cheap.
They aren't that power hungry to move 28nm now. simply marketing.

I do not think that Gridseed will continue to manufacture the 3355 chips as they
could be signing a deal with Innosilicon to manufacture the A2 chips themselves now.
Good move on both parties as if they can ramp up production at both ends they
will take out most of the late to market competition.

If you can get 4Ghs out of 40 chips but draw approx 300w of power you would
only be able to obtain 20Ghs max on a 115vac@15amp service outlet. That is
getting the max (1750w) of a standard wall outlet, you are putting out 1500 watts.
The volt mod with higher freq settings pushing the chips really drives up the power
consumption.

These units are the first version on the ASIC Script miners and will draw the most
power thus making them not efficient enough as higher power miners. That being
said I have seen some company putting 6 stock freq units into a custom rack and
selling as a 33Ghs unit drawing 840 watts but price was way to high and around the  
same price of a newer 28nm 90Mhs miners using the A2 chips.

Good work to everyone inputing into this thread, I am going to mod my Blade soon
with the knowledge in this thread. Next round of blades coming out next week are in
the $900 range now!!

your power usage looks way off.

My blade running overvolted at 7.2MH/s uses around 300w from the wall
5.8MH uses 170 watt from the wall (30 watt just for the psu at idle nothing plugged)

we should be able to reach around 56MH with 1500 watt
price for this have to be no more than 800$ for 14MH to be competitive,
flexible design by multiples of 14MH (200 chips) up to 42MH 1200 watt to easily power it with server PSU like the DL580G3 1300 watt (noisy but efficient)

Made a mistake in my calculation as for one board only.
I estimated that a overclocked Blade mining at 8Mhs (2 boards) might draw around 300 watts.
At that power draw you could only mine at 40Mhs and 1500 watts.
As per your spec of 7.2Mhs per Blade at 300 watt your only going to get max 36Mhs per 15A outlet
if you overclocked the boards.

The problem is once you overclock them the rise in power draw is greater then the performance increase
that you are getting out of the chip.

What is the max number of chips that you can address with the current logic on the board?
Did Gridsed limit to 40 per board due to bandwidth?

[J4bberwock]

What is the max number of chips that you can address with the current logic on the board?
Did Gridsed limit to 40 per board due to bandwidth?

I haven't found the data yet. looks like the chips do have an address assigned and then, the data is broadcasted either for one address (work), either for all chips (generic command)
But they are all connected to the same line, no individual lines, unlike what I expected.
way easier to build a board

My guess is Gridseed decided to go 40 chips because of the power draw.
The powerboard is weak and can't hold much more.
I guess they have re-used a powerboard made for the 8chips dual miner without any change but LTC only mode to keep lower power draw.

[Blisk]

As far as I checked I think also power si a botleneck of the board. But still can USB port take more traffic with more than 40 chips on USB 2.0 if there is 2.0 on board.
so doyou think there is possibillity to add a board with chips to a 5 chip gridseed?

[GenTarkin]

Im curious, can the bridge mod be done w/ a pencil? ... the bridge doesnt care about resistance right? just as long as long as its closed?
Thanks

[J4bberwock]

Im curious, can the bridge mod be done w/ a pencil? ... the bridge doesnt care about resistance right? just as long as long as its closed?
Thanks

I've seen people reporting that it wasn't working on the pod or it works only for some time.

[Xer0]

Im curious, can the bridge mod be done w/ a pencil? ... the bridge doesnt care about resistance right? just as long as long as its closed?

the vid=1 bridge not, but the one rerouting to the 36k resistor does

[J4bberwock]

Zoomhash is selling Voltmodded blades.

http://zoomhash.com/collections/top-sellers/products/new-7-5mhs-zoom-blade

looks like they are doing it the right way, with extra cooling and tips and tricks.

BTW, they should have called it J4b-Blades instead of Zoomblades 
Do you think I can ask for royalties?

For those of you who modded the blade to extreme with a 43k resistor, if you have to shut it down, here is what I'm doing just to let it cool a little.

-unplug the USB
-shut down the PSU for 5 seconds
-power on the PSU again so the fan spins for a few minutes

Doing this, the power draw is way less than when you simply close cgminer or cpuminer the components are gently cooling before I shut it down completely.

[Blisk]

They also made their ovn blades so they bought chips and created their ovn board with fully overclocked blade.

[J4bberwock]

They also made their ovn blades so they bought chips and created their ovn board with fully overclocked blade.

Where have you seen this?
I'm interested to see the device.

[majeis]

I'm interested to see the device.

[NellyWHads]

Hey Wolfey,

The post wasn't there a few hours ago, but they are facing the exact same limitations as me, but they haven't noticed the ferrite beads overheating. Thermal picture wold be nice to identify where are the hot/weak spots.
I managed to get the blade stable for 25 minutes at 1125Mhz, giving 3800Kh/s for the panel with 47K resistor, but after that, it's overheating, even with a fan blowing directly on it and stops sending shares
It's actually running with very few hw errors at 1063Mhz with 42k resistor.

TO-220 package might be a good idea to replace the 09N03. Datasheet says 25V 50A, but power dissipation might be too low I'll have to check if I have an acceptable replacement when back home next week.

Hey J4bberwock ,

I'm thinking of volt modding my Blade with your configuration but just need some advice - later in this forum you go on to say you've got this running stable at 1125Mhz and 7.3Mh/s, how are you actively cooling this setup? What is the temperature you try and stay below? And lastly, how are you powering it? (Direct wiring, and if so, where? Or through the notmal power inlet with the ferrite beads removed?)

Thanks again - and I'm quite impressed with your hash speed, tho I'm unsire if anyone else will come close to it as most of the miners don't run 6.5 Mh/s at 875Mhz stock

[Blisk]

They also made their ovn blades so they bought chips and created their ovn board with fully overclocked blade.

Where have you seen this?
I'm interested to see the device.

me too, but i think the only way is to buy it.
http://zoomhash.com/collections/top-sellers/products/the-lion-64-chips-1-blade-13-5mh-s-450w

[J4bberwock]

They also made their ovn blades so they bought chips and created their ovn board with fully overclocked blade.

Where have you seen this?
I'm interested to see the device.

me too, but i think the only way is to buy it.
http://zoomhash.com/collections/top-sellers/products/the-lion-64-chips-1-blade-13-5mh-s-450w

I can't believe it's made of gridseeds, but it would explain why the supplier replied that the 5000 chips they had was ordered by someone else.
even oveclocked, I'm barely getting 7.5MH/s with 40 80 chips and 300 watt.

Big edit, 7.5MH/S with 80 chips.

so 13MH/s with only 64 chips is far beyond the limits for GC3355.

They were supposed to be 70Kh/s per chip, we are pushing them at 105Kh/s
200+Kh/s looks way too high for such a low power draw.
It'll be 12MH/s with 64 chips, but they will have to run higher clock, so higher power draw. I guess the extra 10% hashrate could be acheived with 30% extra power.

Maybe with 47k or 49.9k resistor and upgrading the VRM you can reach 8MH/s, but I expect more power draw.
The only unknown limit for now with standard powerboard is the up1509 that we don't have the datasheet. How many amps can it stand?

[Blisk]

I think too, they pushed to the limit and use more powerfull power chips or they did their ovn power supply of 1.4 or 1.5V.
I think they use their ovn boards with their ovn power supply
have just sended an email for datashet to UPI semiconductors.
but found only this
http://wenku.baidu.com/view/5f088e176edb6f1aff001f3a.html

UPI semiconductors made chips it is question what is lowest quantity for production.