Pages:
Author

Topic: ZTEX USB-FPGA Modules 1.15x and 1.15y: 215 and 860 MH/s FPGA Boards - page 19. (Read 182443 times)

hero member
Activity: 784
Merit: 500
I tried nearly everything now and i think that this is the only solution Sad

Maybe u could find out whats wrong with it. It was mining happily for two month straight now.


How Should i send it in? With or without the heatsink i used? 
donator
Activity: 367
Merit: 250
ZTEX FPGA Boards
But i need some help on the second board. Could it be that i damaged it somehow. I can't see any damage on the parts (VRM Chip etc), maybe ESD (Elektrostatische Aufladung) killed it?

Maybe a bad cable, maybe a bad USB connector, maybe a bad USB port.

If changing the cable and USB port does not help it is the easiest if you return the board.
hero member
Activity: 784
Merit: 500
Hi, its me again .....

tried everything u told me via mail but i'm unable to get satisfying results.

Quote
You obviously installed the wrong firmware. 1.15x boards are not factory programmed. Because firmwares of one type cannot be overwritten by firmwares of other type you have to proceed as stated above (and you have to use the correct firmware files).
That could be it for the new single.

€: do you know the feeling when u realize that u didn't see your error (Thats what i feel now:P) Smiley New single works!. Sry for inconvenience and THx for hlp:)

Hers a Pic:


But i need some help on the second board. Could it be that i damaged it somehow. I can't see any damage on the parts (VRM Chip etc), maybe ESD (Elektrostatische Aufladung) killed it?
donator
Activity: 367
Merit: 250
ZTEX FPGA Boards
One of my older singles seems to be dead for no reason ..... It won't be recognized buy the mac (or any other PC i have in my household). It's listed as an unknown device under the Hardware Manager (Widows 7) and as i try to install the Ztex driver for this i always get the error: "There is no Driver ....blah blah blah".

It's probably a defect firmware: Power-on the board while JP1 is closed (see http://www.ztex.de/usb-fpga-1/usb-fpga-1.15y.e.html#jp1), open the jumper and install new firmware into EEPROM using the programming mode ("-m p") of BTCMIner.

Quote
€: The new Single
I can however program it with an old bitstream from 120221...jar an it'll mine right away. Somehow that isnt persitent and when i use the new jar it still belives to be a quad...

You obviously installed the wrong firmware. 1.15x boards are not factory programmed. Because firmwares of one type cannot be overwritten by firmwares of other type you have to proceed as stated above (and you have to use the correct firmware files).
full member
Activity: 158
Merit: 100
Now a test to prove my theory.

I've applied some new AS5 and scrachted it linear very slightly. Then I've put the heatsink and moved it around a little bit. With the same result: the heatsink touches only at the corners. Here are the photos:

http://25.media.tumblr.com/tumblr_m38gp7e1oF1ruij9wo1_1280.jpg
http://26.media.tumblr.com/tumblr_m38gpjH9L71ruij9wo1_1280.jpg

What can I do???

Use some low viscosity thermal grease, e.g. the one included in the cooler kit. Apply a blob of it into the center of the FPGA, install the heat sink, and squeeze out excess grease by pressing the heat sink down and moving it.

This will not work with high viscosity thermal grease. This stuff is designed for CPU's which dissipate 15x more power and for installations that press down the heat sink with much higher force.

The plastic packages are not always totally flag. But this is not problem if you apply thermal grease as described above.

Another issue: if PCB is bent for some reason, heat sink of FPGA 1 may touch the USB controller IC1 (which is thinner then the FPGA's). The simple solution is the shift the heat sink so that it does not cover IC1.


I will give it a try.

What about high end thermal pad's, like Phobya XT (http://www.cool4pc.de/Phobya_Waermeleitpad_XT_7Wmk_05mm_120x20mm_fuer_ramplex_innovatek_Mips_Koolance.html)? Is it worth a try?
full member
Activity: 158
Merit: 100
I am surprised that the contact is so poor even with the stock heatsink that should hold the heatsink down pretty good.  Do you know if the heatsink is bent or it is the FPGA?

I've tried another heatsink (Alpenföhn Ötzi). Result: same contact pattern, so the FPGA-4 is bent. Checked FPGA-3, bent too. I've put some extra grease in the middle. Frequency went up to 200, but with errors.

Stats without the "extra" amount of grease in the middle:
Code:
005-0: ztex_ufm1_15y1-04A3476278-1: f=200.00MHz,  errorRate=0.02%,  maxErrorRate=2.60%,  hashRate=200.0MH/s,  submitted 15 new nonces,  luckFactor=0.99
005-0: ztex_ufm1_15y1-04A3476278-2: f=196.00MHz,  errorRate=0.09%,  maxErrorRate=1.91%,  hashRate=195.8MH/s,  submitted 13 new nonces,  luckFactor=0.99
005-0: ztex_ufm1_15y1-04A3476278-3: f=192.00MHz,  errorRate=0.00%,  maxErrorRate=0.52%,  hashRate=192.0MH/s,  submitted 15 new nonces,  luckFactor=1.01
005-0: ztex_ufm1_15y1-04A3476278-4: f=196.00MHz,  errorRate=0.00%,  maxErrorRate=0.98%,  hashRate=196.0MH/s,  submitted 15 new nonces,  luckFactor=0.99

...and after:
Code:
005-0: ztex_ufm1_15y1-04A3476278-1: f=200.00MHz,  errorRate=0.24%,  maxErrorRate=1.22%,  hashRate=199.5MH/s,  submitted 14 new nonces,  luckFactor=0.92
005-0: ztex_ufm1_15y1-04A3476278-2: f=196.00MHz,  errorRate=0.51%,  maxErrorRate=0.88%,  hashRate=195.0MH/s,  submitted 12 new nonces,  luckFactor=1.09
005-0: ztex_ufm1_15y1-04A3476278-3: f=200.00MHz,  errorRate=0.00%,  maxErrorRate=0.50%,  hashRate=200.0MH/s,  submitted 12 new nonces,  luckFactor=1.00
005-0: ztex_ufm1_15y1-04A3476278-4: f=200.00MHz,  errorRate=0.87%,  maxErrorRate=1.78%,  hashRate=198.3MH/s,  submitted 16 new nonces,  luckFactor=1.02

Here the photo of the other heatsink:
donator
Activity: 367
Merit: 250
ZTEX FPGA Boards
Now a test to prove my theory.

I've applied some new AS5 and scrachted it linear very slightly. Then I've put the heatsink and moved it around a little bit. With the same result: the heatsink touches only at the corners. Here are the photos:

http://25.media.tumblr.com/tumblr_m38gp7e1oF1ruij9wo1_1280.jpg
http://26.media.tumblr.com/tumblr_m38gpjH9L71ruij9wo1_1280.jpg

What can I do???

Use some low viscosity thermal grease, e.g. the one included in the cooler kit. Apply a blob of it into the center of the FPGA, install the heat sink, and squeeze out excess grease by pressing the heat sink down and moving it.

This will not work with high viscosity thermal grease. This stuff is designed for CPU's which dissipate 15x more power and for installations that press down the heat sink with much higher force.

The plastic packages are not always totally flag. But this is not problem if you apply thermal grease as described above.

Another issue: if PCB is bent for some reason, heat sink of FPGA 1 may touch the USB controller IC1 (which is thinner than the FPGA's). The simple solution is the shift the heat sink so that it does not cover IC1.


donator
Activity: 367
Merit: 250
ZTEX FPGA Boards
can i have some values in mh/s about ztex 1.15x ?

(Measured) hash rates from different lots and and for both FPGA variants can be found in the initial posting.

The variance of the first 1.15y batch is very large because it was build from several FPGA lots.

hero member
Activity: 784
Merit: 500
thx Smiley

Ill try it and i hope that Ztex will reply to my posts and emails on monday.

Actualy it shows up on windows but only as an unknown device witch I'm unable to install drivers for (see posts above)
donator
Activity: 305
Merit: 250
That is weird that it doesn't work on any of the PCs.  Good luck with Ubuntu.  Hopefully it works for you.  Remember to use sudo before the command, ie sudo java -cp...
hero member
Activity: 784
Merit: 500
ill download ubuntu right now ... Smiley

But its not only limited to my mac. It doesn't show up in any windows PC i currently own....
legendary
Activity: 966
Merit: 1000
Now a test to prove my theory.

I've applied some new AS5 and scrachted it linear very slightly. Then I've put the heatsink and moved it around a little bit. With the same result: the heatsink touches only at the corners. Here are the photos:




What can I do???

try to clean both surfaces with isopropylen alcohol. they became clean from the old grease.
you should be able to read the letters on the fpga chip.
after that put carefully a thin film of grease and the heatsink on the chip, fix it with the clips and start the miner.

good luck!
donator
Activity: 305
Merit: 250
donne:
1-stock on screws and sommekind of grease
2-stock on screws and Thermal pad 1mm 7kw (the 0.5 was 5kw only?)
3-stock and AS5 grease
4-stock and sommekind of grease

testing under unfamilliar conditions - 34*C ambient - stay tuned

after 15 minutes - this same....
i`ll take it for a trip sowmere cooler - 25*C ambient.
this same...1 208 and the rest 196.... board is almost cold
Sorry for the multiple posts.  There seems to be multiple topics going on.

If you have a temperature probe, try measuring the temperature on the heatsink and directly under the LX150 to see if the heat is being conducted efficiently from the LX150 to your heatsink.  I had one board where the contact was not as good (no paste at the corners of the LX150).  The underbelly is hot while the heatsink was not as hot as the other boards.  Try to see if you get the same temps from the 4 LX150s.

For example, data I got testing a 1.15x, ambient 19C, bottom of LX150 = 30C, top of heatsink = 24C
donator
Activity: 305
Merit: 250
Now a test to prove my theory.

I've applied some new AS5 and scrachted it linear very slightly. Then I've put the heatsink and moved it around a little bit. With the same result: the heatsink touches only at the corners. Here are the photos:




What can I do???
I am surprised that the contact is so poor even with the stock heatsink that should hold the heatsink down pretty good.  Do you know if the heatsink is bent or it is the FPGA?
 
The most straight forward way if it is the heatsink is just to replace the heatsink.  The alternative (which I know will sound like heresy to some) is to put quite a bit of the paste in the middle instead of the nice and thin layer that you have.  My first few boards I applied a thick layer of AR5.  Some of it will spill over onto the board.  2 of the boards I cleaned off the AR5 and re-applied a nice thin layer.  It actually didn't make a difference in the performance.  From measuring the temperatures with a probe, a thin layer and a thicker one didn't make any difference in the heatsink or the temperature under the LX150.  So what I am saying is that if you have an air pocket between the LX150 and the heatsink, you might be able to get away with putting a little bit more paste to fill the air pocket.  The thermal conductivity of AR5 is not as good as aluminum, but it should be good enough for running the LX150 (and definitely better than air). 
donator
Activity: 305
Merit: 250
This unit seems to have died on me while i was playing arround with the Quad and the others.

I don't think so. I had this issue with both the 1.15x and 1.15d. It happened both times after removing and reconnecting them. I think the problem is on the micro USB side. Try there.


Same issue here, after repowering und reconnecting the boards (usb and power) several time, it worked again.
Ditto.  I had this happened to me on 2 separate occasions, both on windows XP (on a laptop I use to program the fpgas individually).  The first time with the single it resolved after reconnecting several times.  The second time (with the quad) windows wouldn't recognize it even after several tries.  I plug it into the production machine running Ubuntu and it recognized it just fine.
legendary
Activity: 1022
Merit: 1000
BitMinter
i got water block witch fits perfectly, but i cant se sense in doing it :/


I had this one board (single) I was never able to cool down the way i wanted. I ended up taking a clamp (Schraubzwinge) to pull the heatsink down to the chip. The clamp is in place to this day. Perhaps you can take off one of the fans from one of the 196 MHz chips and try with a little clamp and some LIGHT pressure. This way you can see if it's a contact problem.

I checked my heatsink then and it was perfect level... you can check it on a window glass. Floatglass is very even.
sr. member
Activity: 274
Merit: 250
i got water block witch fits perfectly, but i cant se sense in doing it :/
legendary
Activity: 1022
Merit: 1000
BitMinter
 Undecided only this one board ?
sr. member
Activity: 274
Merit: 250
donne:
1-stock on screws and sommekind of grease
2-stock on screws and Thermal pad 1mm 7kw (the 0.5 was 5kw only?)
3-stock and AS5 grease
4-stock and sommekind of grease

testing under unfamilliar conditions - 34*C ambient - stay tuned

after 15 minutes - this same....
i`ll take it for a trip sowmere cooler - 25*C ambient.
this same...1 208 and the rest 196.... board is almost cold
legendary
Activity: 1022
Merit: 1000
BitMinter
I have got 7W/mK by Phobya, i`ll use.
Heatsinks are not good polished, lots of space in there. I`ll cut a square of phobya pad and screw heatsink on it.
As i said before, i got 0.5, 1.0 and 1.5mm - witch should i use ?

I'd start with 0.5...
Pages:
Jump to: