Pages:
Author

Topic: GridSeed 5-chip USB miner voltage mod - page 24. (Read 157001 times)

sr. member
Activity: 378
Merit: 250
April 26, 2014, 10:15:52 PM
Hey Guys Cheesy

I just updated my guide. It now includes:
- ZiG's rVMOD3 (Reversible VMOD3) in simple language and beautiful high resolution pictures.
- G-Blade VMOD1 Information & Pictures

Check it out, let me know what you think Smiley



What about the 2 or more pods that were ruined by trying to do 'Zigs' mod?
Have they been addressed and handled or not?
I am afraid of it happening to others who apply it.
And some others have already passed on it to in favor of the R52 aka 'the great R139 discovery' axial or SMD 49.9k   Grin
I may try it while modding this next batch coming in.
I have serious reservations that something has been overlooked and may be causing damage to crystal oscillators.
2 so far that I've come across said their miner started 'clicking' which is really weird but totally possible ergo the above.
The only parts that could even possibly click would be the crystal 'most probable' - or L1 the inductor. If somehow and it's highly doubtful some of its windings are lose, the click could perhaps 'longshot' come from there.
Only other possibility is the sound of pulsed Arcing hence clicking but I don't see how that's even possible. There ain't no HV running in there anywhere. And oh, possibly a capacitor that has blown internally, that might click too. Nothing else on the card can or will, unless it releases its magic black smoke Wink
I'll be reporting my results IF I even decide to try it.
Updates comin!
BAMM! BAMM!
full member
Activity: 140
Merit: 100
April 26, 2014, 10:12:28 PM
ZiG's rmod is adding a resistor in series and can be done at least 2 ways.

1. Cut the bridge at R46 and add a resistor between those two points.  To undue just re-bridge R46.  Removing the resistor is optional as it would be bypassed by the bridge.

OR

2. Cut the bridge at R46 and add a resistor between the outer point of R46 and CON3 pin 3 (DGND). To undue remove the resistor.

method 2 is a bit easier because you have lots of room to work at both points
method 1 is easier to undue

---------------------------------------------------------------------------------

I think that ZiG's discovery has additional uses. (I'll try to confirm tomorrow)
VMOD1 = 36k + ZiG's idea but using a 13k resistor (for example) would give you 49k. (not tested yet)
VMOD3 = 47K + ZiG's idea but using a 2k resistor (for example) would give you 49k. (not tested yet)
It looks useful for anyone who has done other mods but now want to try a higher resistor value.  It's possible you won't have to undue the other mod if this works.

nemercry or ZiG would know if the second part of what I posted is correct or not.  
For now, stick to the first part of my post (above the dotted line) as being ZiG's rmod.

(edit - I see ZiG posted as I was typing)
ZiG
sr. member
Activity: 406
Merit: 250
April 26, 2014, 10:05:07 PM
..................
- Reversible VMOD3 added (Courtesy of ZiG)

Step B) If you want to revert to VMOD1 simply solder together both of the two point sets shown [in Figure 7].
....................


Jess85 - Step B needs clarification as it has merged two separate mods.  ZiG's rmod requires that you cut the bridge at R46 then one method is to solder the resistor as described.  To reverse ZiG's rmod you would remove the resistor and re-solder the bridge at R46. Then you could do VMOD1 if you want but vmod1 has nothing to do with ZiG's rmod.  


I think I may have missed the R46 mention!
But as I understand it, re-soldering the bridge at R46 should be enough to reverse the mod no? having to remove the axial resistor would then defeat the claim that it is any more reversible then the original VMOD3 I would think. At least that's what was personally understood from reading ZiG's earlier posts.

Can you please clarify on this?

Yep...Happydaze is right...R46 needs to be cut first...

Reversal is as easy as shorting back R46...or the Rmod...the additional resistor on R46 pads...you can do this with a wire on the Rmod terminals...without even removing it...

So no other Mods are involved / needed...

If you have questions...Let me know...I like your guide and will be happy to help you improving it...

ZiG
ZiG
sr. member
Activity: 406
Merit: 250
April 26, 2014, 09:58:56 PM
Hey Guys Cheesy

I just updated my guide. It now includes:
- ZiG's rVMOD3 (Reversible VMOD3) in simple language and beautiful high resolution pictures.
- G-Blade VMOD1 Information & Pictures

Check it out, let me know what you think Smiley



Good job, Jess85...!

Thank you for the reference to my mod...Excellent guide... Smiley

ZiG

newbie
Activity: 17
Merit: 0
April 26, 2014, 09:45:48 PM
..................
- Reversible VMOD3 added (Courtesy of ZiG)

Step B) If you want to revert to VMOD1 simply solder together both of the two point sets shown [in Figure 7].
....................


Jess85 - Step B needs clarification as it has merged two separate mods.  ZiG's rmod requires that you cut the bridge at R46 then one method is to solder the resistor as described.  To reverse ZiG's rmod you would remove the resistor and re-solder the bridge at R46. Then you could do VMOD1 if you want but vmod1 has nothing to do with ZiG's rmod.  


I think I may have missed the R46 mention!
But as I understand it, re-soldering the bridge at R46 should be enough to reverse the mod no? having to remove the axial resistor would then defeat the claim that it is any more reversible then the original VMOD3 I would think. At least that's what was personally understood from reading ZiG's earlier posts.

Can you please clarify on this?
full member
Activity: 140
Merit: 100
April 26, 2014, 09:33:59 PM
..................
- Reversible VMOD3 added (Courtesy of ZiG)

Step B) If you want to revert to VMOD1 simply solder together both of the two point sets shown [in Figure 7].
....................


Jess85 - Step B needs clarification as it has merged two separate mods.  ZiG's rmod requires that you cut the bridge at R46 then one method is to solder the resistor as described.  To reverse ZiG's rmod you would remove the resistor and re-solder the bridge at R46. Then you could do VMOD1 if you want but vmod1 has nothing to do with ZiG's rmod.  
ZiG
sr. member
Activity: 406
Merit: 250
April 26, 2014, 09:20:11 PM
for those in the USA, I just cam across a very good price for a groupbuy

https://bitcointalksearch.org/topic/gb-for-gridseed-5-chip-0193btc-blades-299btc-583755

not really a great deal if you ask me...

GAW sells them for $83, right now!

I'm sure once the gen 2 chips come out, these will be DIRT CHEAP





Hey buddy,

Where did you see these Gridseeds for $83 on GAWMiners...?

I have couple beers too...but still couldn't find it... Grin

Cheers,

ZiG
sr. member
Activity: 308
Merit: 250
April 26, 2014, 09:10:04 PM
Hey Guys Cheesy

I just updated my guide. It now includes:
- ZiG's rVMOD3 (Reversible VMOD3) in simple language and beautiful high resolution pictures.
- G-Blade VMOD1 Information & Pictures

Check it out, let me know what you think Smiley



nice guide indeed.
newbie
Activity: 17
Merit: 0
April 26, 2014, 08:55:13 PM
Hey Guys Cheesy

I just updated my guide. It now includes:
- ZiG's rVMOD3 (Reversible VMOD3) in simple language and beautiful high resolution pictures.
- G-Blade VMOD1 Information & Pictures

Check it out, let me know what you think Smiley

sr. member
Activity: 308
Merit: 250
April 26, 2014, 08:47:52 PM
for those in the USA, I just cam across a very good price for a groupbuy

https://bitcointalksearch.org/topic/gb-for-gridseed-5-chip-0193btc-blades-299btc-583755

not really a great deal if you ask me...

GAW sells them for $83, right now!

I'm sure once the gen 2 chips come out, these will be DIRT CHEAP



Didn't see the GAW prices yet. went to see their website, they are selling for 130 dollars on GAW

http://www.gawminers.com/300-kh-s-single-gridseed-asic-miner/

yes no doubt that prices will fall real quick, I've got some pre orders at Fibonacci. 8.6Mh for 320$ that's going to kill Gridseed prices fast when those come out.

but here in Europe the gridseeds remain expensive it's hard to get one below 120/130 euros!
hero member
Activity: 672
Merit: 500
April 26, 2014, 08:23:52 PM
for those in the USA, I just cam across a very good price for a groupbuy

https://bitcointalksearch.org/topic/gb-for-gridseed-5-chip-0193btc-blades-299btc-583755

not really a great deal if you ask me...

GAW sells them for $83, right now!

I'm sure once the gen 2 chips come out, these will be DIRT CHEAP



sr. member
Activity: 308
Merit: 250
April 26, 2014, 08:21:07 PM
for those in the USA, I just cam across a very good price for a groupbuy

https://bitcointalksearch.org/topic/gb-for-gridseed-5-chip-0193btc-blades-299btc-583755
hero member
Activity: 826
Merit: 1000
°^°
April 26, 2014, 06:37:00 PM
amix, are you using a permanent tip?
if yes you can kill it with scraping

try a tip “clean&tin“ or “reactivator“
not me it was  gtraah. i was just thinking that he scrapped the tip, probably killed it like you say.
ah, ok, sry
i cant quote long texts when on mobile
newbie
Activity: 17
Merit: 0
April 26, 2014, 04:26:27 PM
I'm running 20 Gridseeds off a Raspberry Pi and did the 49.9k mod and 5V fan mod to each of them. When I use bfgminer with a clock speed of 1200 MHz, even though the app reports that I'm hashing around 10 MH/s, my pool only reports about 5.5 MH/s.

When I use sandor's cpuminer with autotune on and at 1200 MHz, the results are better. My pool reports 8.5 MH/s, but this isn't quite up to 10 MH/s. I'd expect my hashrate to be in the mid 9 MH/s range; my rejected share percentage is only around 1-3%.

Tomorrow I'm gonna try running two Raspberry Pi's with 10 Gridseeds each to see if the combined speed is higher than running one Raspberry Pi with 20 Gridseeds, but does anybody have any ideas on how I can increase the hashrate to match what I expect? I have 80 Gridseeds in all, and if I'm losing 1.5 MH/s on 20 Gridseeds, that's 6 MH/s across all 80. That's equivalent to buying another 16 unmodified Gridseeds, which is not an insignificant amount.

Are you getting HW errors?  I was only able to run the Gridseeds at 1200Mhz after I did the 49.9K mod and I put removed the stock thermal pad from the bottom of the board and applied non-conductive diamond thermal compound.  Still not all of them will run at 1200.  I run them at 1175Mhz with no HW errors.
Only a few hardware errors. The autotune takes care of optimizing the clock speed.
sr. member
Activity: 378
Merit: 250
April 26, 2014, 02:53:52 PM
There are some thermal image. GS is running at 1200MHz, R52=54.9kOhm, fan@USB5V. Ambient temperature 27 degC.

Top view:
snip - see above

12V DC/DC MosFETs:
snip - see above

12V Inductor:
snip - see above

DC Power connector:
snip - see above

Microcontroller:
snip - see above

Side view of 12V DC/DC MosFETs:
snip - see above

Side view of connectors:
snip - see above


Hah! Cool! I mean, HOT!  Grin
member
Activity: 107
Merit: 13
April 26, 2014, 02:44:27 PM
There are some thermal image. GS is running at 1200MHz, R52=54.9kOhm, fan@USB5V. Ambient temperature 27 degC.

Top view:


12V DC/DC MosFETs:


12V Inductor:


DC Power connector:


Microcontroller:


Side view of 12V DC/DC MosFETs:


Side view of connectors:
sr. member
Activity: 308
Merit: 250
April 26, 2014, 10:22:01 AM
Hello

I would like to share with you some impressions about Sandor_Cpuminer vs Bfgminer.
This is my setup graph, it's a bunch of Gridseed running from a raspberry:



Pods are still un-modded so original ones by various sellers (but they look like the same inside), as mod they only have the fan attached to 5v external supply. (via usb)

I tried first bfgminer with a frequency of @850 (the best to not get HW in every pods) you can see the red circle, it was not a great hashrate I was expected slightly more and they also had a drop down during the night.

I was very unsatisfied so I gave a try to cpuminer (I never tried it) and I discovered the Sandor's fork with the auto-tune feature and BOOM!
Hashrate went up immediately and thanks to the auto-tune feature it can go slightly more upper the extreme-limit for a limited period, but in general hashrate now is more than I was expected to obtain and it seems to be definitely stable (no drops down).

Also please notice the stale columns, with bfgminer they were a lot while with cpuminer they are almost 0.

So, many compliments to Sandor, he did a really good job.

Now I think the (small) problem is that cpuminer hasn't neither a summary (but I saw Sandor is working on it) neither an RPC to get stats/devs info, and so it's not possible to fork a scripta like web interface to work with cpuminer (sad while it's not indispensable).

I think a bfgminer with the sandor-cpuminer features (like auto-tune and per device/proc overclock) could be the best option possible, but I don't know the work needed to integrate the sandor's work on bfgminer (neither if it's possible).

Anyway thanks for excellent job.


very nice, I've made the same conclusion and the new cpu miner of sandor is just amazing.

So I ran 24 with 5 modified GS's and sandors CPUminer and this is a screenshot of pool side after 24H


Taken with Webpage Screenshot

before I did the same test when the pods arrived ( forgot to take a screenshot sorry ) but results then were barely at 2000kh/s 24h
member
Activity: 84
Merit: 10
April 26, 2014, 09:55:58 AM
amix, are you using a permanent tip?
if yes you can kill it with scraping

try a tip “clean&tin“ or “reactivator“
not me it was  gtraah. i was just thinking that he scrapped the tip, probably killed it like you say.
donator
Activity: 1057
Merit: 1021
April 26, 2014, 08:18:20 AM
I'm running 20 Gridseeds off a Raspberry Pi and did the 49.9k mod and 5V fan mod to each of them. When I use bfgminer with a clock speed of 1200 MHz, even though the app reports that I'm hashing around 10 MH/s, my pool only reports about 5.5 MH/s.

When I use sandor's cpuminer with autotune on and at 1200 MHz, the results are better. My pool reports 8.5 MH/s, but this isn't quite up to 10 MH/s. I'd expect my hashrate to be in the mid 9 MH/s range; my rejected share percentage is only around 1-3%.

Tomorrow I'm gonna try running two Raspberry Pi's with 10 Gridseeds each to see if the combined speed is higher than running one Raspberry Pi with 20 Gridseeds, but does anybody have any ideas on how I can increase the hashrate to match what I expect? I have 80 Gridseeds in all, and if I'm losing 1.5 MH/s on 20 Gridseeds, that's 6 MH/s across all 80. That's equivalent to buying another 16 unmodified Gridseeds, which is not an insignificant amount.

Are you getting HW errors?  I was only able to run the Gridseeds at 1200Mhz after I did the 49.9K mod and I put removed the stock thermal pad from the bottom of the board and applied non-conductive diamond thermal compound.  Still not all of them will run at 1200.  I run them at 1175Mhz with no HW errors.
newbie
Activity: 17
Merit: 0
April 26, 2014, 07:39:16 AM
I'm running 20 Gridseeds off a Raspberry Pi and did the 49.9k mod and 5V fan mod to each of them. When I use bfgminer with a clock speed of 1200 MHz, even though the app reports that I'm hashing around 10 MH/s, my pool only reports about 5.5 MH/s.

When I use sandor's cpuminer with autotune on and at 1200 MHz, the results are better. My pool reports 8.5 MH/s, but this isn't quite up to 10 MH/s. I'd expect my hashrate to be in the mid 9 MH/s range; my rejected share percentage is only around 1-3%.

Tomorrow I'm gonna try running two Raspberry Pi's with 10 Gridseeds each to see if the combined speed is higher than running one Raspberry Pi with 20 Gridseeds, but does anybody have any ideas on how I can increase the hashrate to match what I expect? I have 80 Gridseeds in all, and if I'm losing 1.5 MH/s on 20 Gridseeds, that's 6 MH/s across all 80. That's equivalent to buying another 16 unmodified Gridseeds, which is not an insignificant amount.
Pages:
Jump to: