Bitcoin Forum
December 14, 2024, 10:18:30 PM *
News: Latest Bitcoin Core release: 28.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [26] 27 28 29 30 31 32 »
  Print  
Author Topic: GRIDSEED G-BLADE Overclocking 7Mh/s, improvements and repair  (Read 74143 times)
racebyu
Sr. Member
****
Offline Offline

Activity: 346
Merit: 260


View Profile
July 16, 2014, 06:56:31 PM
 #501

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.
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 16, 2014, 11:12:07 PM
Last edit: July 16, 2014, 11:32:14 PM by Freeman2000yes
 #502

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.
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 16, 2014, 11:36:24 PM
 #503

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?
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 16, 2014, 11:43:22 PM
Last edit: July 17, 2014, 04:32:01 AM by Freeman2000yes
 #504

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.
racebyu
Sr. Member
****
Offline Offline

Activity: 346
Merit: 260


View Profile
July 17, 2014, 02:27:37 AM
 #505

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.
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 03:07:40 AM
Last edit: July 17, 2014, 03:38:14 AM by Freeman2000yes
 #506

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.
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 04:09:56 AM
Last edit: July 17, 2014, 04:29:05 AM by Freeman2000yes
 #507

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.
zhinkk
Sr. Member
****
Offline Offline

Activity: 266
Merit: 250


View Profile
July 17, 2014, 04:11:28 AM
 #508

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?
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 04:26:21 AM
 #509

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.
racebyu
Sr. Member
****
Offline Offline

Activity: 346
Merit: 260


View Profile
July 17, 2014, 04:30:47 AM
 #510

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.
ZiG
Sr. Member
****
Offline Offline

Activity: 406
Merit: 250


View Profile
July 17, 2014, 04:40:52 AM
 #511

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.

Excellent explanation + table, Freeman2000yes ... Smiley

You seem to be pretty knowledgeable...what is your education / profession, buddy... Huh

Thanks,

ZiG
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 04:44:04 AM
 #512

Quote
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.

And that´s the reason why only they can call themselves Mr. BIG. Wink
No warranty, no money loss.
But what about the promise of fully decentralized crypto currencies?
Only if you have the real big money to buy at first the big Hash machines, you can get some profit.
So don´t care about warranties , try to do the best you can to maximize the income , so maybe don´t loose to much money.
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 04:57:17 AM
 #513

Quote
Excellent explanation + table, Freeman2000yes ... Smiley

You seem to be pretty knowledgeable...what is your education / profession, buddy... Huh

Thanks,

ZiG
Thanks for  your appreciation! Roll Eyes
At the moment i try to do a study as an ingeneer in electronic communication. But i´m not so far and too lazy to learn. Roll Eyes
I´m proud of me that i can participate in a forum in english that is not my native language. So sorry about any misspelling.
I only try to give something back to the forum , because i get some informations ,too.
styson
Newbie
*
Offline Offline

Activity: 20
Merit: 0


View Profile
July 17, 2014, 05:14:42 AM
 #514

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?

Still waiting for the replacement STM32's, RS takes between 4-6 days to ship non stock items to South Africa.

I got the firmware from Gades, just hope it is for the 40 Chip Gridseed
styson
Newbie
*
Offline Offline

Activity: 20
Merit: 0


View Profile
July 17, 2014, 06:20:55 AM
 #515

Damn I ordered the LQPF 48 (only on the 5 chip schematic) STM32 instead of the LQPF 64 for the blade, at least I also ordered samples from ST Micro of the 64 last week Grin Hopefully it arrives asap

Don't you just love free samples ?  Grin
scryptr
Legendary
*
Offline Offline

Activity: 1797
Merit: 1028



View Profile WWW
July 17, 2014, 08:11:16 PM
 #516

Gridseed Moddermen-

 I have been following this thread for the last couple of weeks.  It is really full of info, and it has raised some questions in my mind.

For starters, I have a Gridseed blade that was produced in Jun '14 (it has zero Ohm resistors).  From reading, I see one preparatory step, and three main mod steps:

          1) In preparation, improve the cooling with extra heatsinks.

          2) Replace the 33 KOhm resistor with another (36-to-44.2 KOhm examples given in the thread).

          3) Short 1 or 2 junctions for the "voltmod=1" configuration.

          4) Replace the 220uF, 16 volt capacitor can with a 220uF, 35 volt model with the same fit.

There are other improvements possible, such as changing the power socket, etc.

The questions I have follow:

          1) Are the modifications #2 and #3 above compatible?  I believe that J4bberw0ck put both in place.

          2) If both modifications #2 and #3 are performed, is more heat produced at the same hashing frequency?

          3) If both modifications #2 and #3 are performed, is a greater hash rate obtainable overall?

          4) If both modifications #2 and #3 are performed, is a greater hash rate obtained at a single frequency? 

It is my understanding that modification #4 does not directly improve the hash rate, but does improve the heat capacity of the blade at higher hashing frequencies, as the capacitor will stay intact longer.

One last question:

          5) If you use an adjustable pot and run at a set hashing frequency, say 925 MHZ, is the heat generated greater if the pot is set to 37.5 KOhm compared to the pot being set at 36 KOhm?

I hope that someone might take the time to answer.   Thanks for the difficult work already done!  -Scryptr

SCRYPTR'S NOTEBOOK: https://bitcointalk.org/index.php?topic=5035515.msg46035530#msg46035530
GITHUB: "github.com/scryptr"  MERIT is appreciated, also.  Thanks!
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 11:42:17 PM
 #517

I try to explain... if i´m wrong , someone can correct me please.

Everything what means higher hashrate is like in overclocking any CPU. This means the frequency gives you the maximum number of calculations the chip can do per second.
You can surley try to set higher frequencies whithout any modifications like with the 5 Chip GSD they come with stock freq of 600 Mhz but you can set easily up to 850 Mhz.
Higher frequency means more calculations per second and more Hashrate. The problem is that with higher frequencies the chip becomes unstable and this result in
HW´s or miscalculations. It´s like someone try to do too many things at the same time , and at one point he can´t handle so many opperations and stuck.
So the goal is to get the chips stable at higher frequencies  an they need more Voltage to do this . And of course more voltage and frequency  generates more heat. The second problem.
The default Voltage of the GSD Chips is at 1,2 V With the 33K ohm resistor. If you replace this with a 39 K ohm resistor for example the Voltage rise to 1,4 V , so they
handle more stable frequency but also more heat. This you can easily measure at the big yellow caps.

Quote
1) Are the modifications #2 and #3 above compatible?  I believe that J4bberw0ck put both in place.
The bridge voltmod=1 does the same thing but in other way it affects the 2-bit VID convertor to select another output line, this means it selects another resistance. There are 4 Resistors to set 36k, 30k, 33k, and 27k. With the clasic vmode you choose the 36k resistor istead the 33 k , like an overclock jumper without the need to solder any parts.If you do the vmode and choose the 36k , but you replace the 33k resistor , i think the replace of the 33k is useless and do nothing . The convertor coose only one output.
And of course you need the modifed cgminer and set the voltage=1 parameter. So think this is more like an alternative to solder a resistor.

Maybe J4bberw0ck can tell us more if i forgot something.I only apply the bridge mod to my first gsd 5 chip miner. There are for sure more jumpers and you can influence the output voltage in more ways but for me i think is better to stay with one mod at same time its possible to have unespected effects or that both methods add the voltage change and you have to much output.
Quote
2) If both modifications #2 and #3 are performed, is more heat produced at the same hashing frequency?

For me this happens, at the same freq setting you have permanently change the chip voltage if you apply the resistor mod. Higher Voltage means more heat.

Quote
3) If both modifications #2 and #3 are performed, is a greater hash rate obtainable overall?

I don´t no any combination of both methods , but both results  in increase the chipvoltage to rise higher frequencies at very low Hardware errors. So if you have more Vcore and can set higher Frequency at very low HW this results in higher hash Rate. And have in mind that only the accepted shares counts for the effective hash rate.
There are for sure more "little" things to adjust this and get the best compromise between frequency and HW´s.
I don´t know if there are more benefits if you combine both methods for me more simple to replace just one resistor or to solder just one bridge.
Quote
4) If both modifications #2 and #3 are performed, is a greater hash rate obtained at a single frequency?

The effective Hashrate for sure i think, if you have less Hw at the same Frequency , maybe the mining software say the same Hash rate for the same Freq but you get more accepted shares, if there was some HW´s in default mode.

Quote
5) If you use an adjustable pot and run at a set hashing frequency, say 925 MHZ, is the heat generated greater if the pot is set to 37.5 KOhm compared to the pot being set at 36 KOhm?
Maybe the difference is not so big but if you set 925 Mhz @ 37.5 K the chips works with higher voltage than when you set 925 Mhz @ 36k. So the power consumption rise with the higher voltage and this produce more heat at the same freq.
So if you don´t want to overclock so much , its better to stay with 36K if this runs stable. Just an example : I have 2 blades , one with 36 K Mod ,one with 39 K mode, the one with 39 K Mode is a little bit faulty so i run them "only"  at 900 Mhz. The 36k one runs better so i set 950 Mhz , but even if the 39K one has a lower Freq it produce more or min. the same heat like the 36k one. That´s because this runs with higher voltage.

I hope this helps you a little bit , and maybe someone can complete or correct me if i´m wrong.Especially for the bridge mods you can read here more : ----> https://bitcointalk.org/index.php?topic=519112.0
Freeman2000yes
Newbie
*
Offline Offline

Activity: 28
Merit: 0


View Profile
July 17, 2014, 11:59:57 PM
 #518

But what happens to the profitability , guys  Shocked

Anyone see a chance that we have better times in the future? I can´t belive that this can be true , if this continue in this way i don´t  see more than 25% return of my investment , and the rest 75% are gone. Undecided
racebyu
Sr. Member
****
Offline Offline

Activity: 346
Merit: 260


View Profile
July 18, 2014, 03:12:04 AM
 #519

But what happens to the profitability , guys  Shocked

Anyone see a chance that we have better times in the future? I can´t belive that this can be true , if this continue in this way i don´t  see more than 25% return of my investment , and the rest 75% are gone. Undecided

I am sure profitability is heading south as the overall difficulty is increasing
and the need for more powerful miners are needed at a higher cost.
Kinda of a catch 22 situation, like anything the highest rewards are to those
that take the greatest risk and fork out money for the newest and fastest
miners before release. Only risk what you can afford to loose in mining
hardware and new coin purchases. With the alt coins no limitation in new
coins being released the value of the coins may never be able to generate
profits like previous bitcoin. Also with so many different coins on the market
the hash power is being spread so thin that there might never be enough
prolonged hash power kept on any specific coin to push it up and keep it
there for great profits to be made. The pump and dump in the alt coin market
maybe the only way to make some money. The alt coins of the future might
be no more then old penny stocks. I hope this is not true as my last year
of mining might only be for a few thousand dollars, time will tell and without
adoption of alt coins by the marketplace, countries and businesses the future
might only be very profitable for the coin owners/programmers and miner
hardware manufactures. These are just my own thoughts and conclusions.
Blisk
Sr. Member
****
Offline Offline

Activity: 412
Merit: 250


View Profile
July 18, 2014, 01:43:37 PM
 #520

I have calculated, no mather what you buy, you mostly won't get payed off a long time or maybe even never.
I have never payed off my miner and I bought them about 3 months ago and I am not even at half.
Isn't that logical? Why do you think they sell those devices, if minig is more profitable then they will use it by them self not selling it.
But they have calculated and most profitable now is selling that devices and not mining on them.
Because when you buy 500Mhash for alot for alot of money after 3 months they will sell 2THash for the same money and people will buy it and dificulty will raise fast and after 3 months they will sell for the same money 5THash and people will buy it and dificulty will raise again and you didn't payed half of your device after 6 months and dificulty raise more and more. Smiley

Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 [26] 27 28 29 30 31 32 »
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!