Swedish ASIC miner company kncminer.com

[vegasguy]

did you try it with 1000uf??

google says: sometimes a higher uf is acceptable, but not always ... depends on the circuit   :|   isn't 1000uf to 1500uf a pretty big jump?

Google is correct. We can make circuits for noise reduction and for another purposes. So if this is noise reduction purpose then there will be the lower limit for capacitor NOT LESS, but if it is some oscillating circuit then higher capacitor will change the frequency or amplitude.. this was our hand book http://users.physics.harvard.edu/~horowitz/aoe/

I'd try a 1000uf, afterall we are troubleshooting, have to see what works and doesn't  

here's the pic of the spreader removed

http://s18.postimg.org/py0qzeio7/1_kch2zw_Ak_GWOPBBu2o_Sg_OEg.png

looks like he used that little precision screwdriver to pry on the upper left corner, I am not saying to try it, just showing you the picture  -- I've wanted to look under there myself but don't want to risk losing a good board

and I guess that's not a Titan board, maybe a Neptune??

Ok this is a Neptune board. The BTC logo tells me that. Only Neptunes have the BTC logo on them.  I was trying to find which cores include die 0 and 1, as there are 2 dies per core. Die 0 and die 1 are by far the most troublesome, and seem to always fail. I THINK its the core on the top left corner. Can anyone confirm this? Anyone else notice in this pic that the top left core is the only one that does not have thermal grease on it? I suppose the tech could have cleaned it, but then why just clean one? Is it possible there is a fault in the design where the plate does not meet the core on the top left , which I think is contains die 0 and 1? I think the tech in this pic is showing a major fault in design, that is the reason for this pic (im guessing) .This would explain a WHOLE lot! Guys im not suggesting to remove your plates, and there is a strong possibility it will damage your cores.

Vegas

[TXSteve]

did you try it with 1000uf??

google says: sometimes a higher uf is acceptable, but not always ... depends on the circuit   :|   isn't 1000uf to 1500uf a pretty big jump?

Google is correct. We can make circuits for noise reduction and for another purposes. So if this is noise reduction purpose then there will be the lower limit for capacitor NOT LESS, but if it is some oscillating circuit then higher capacitor will change the frequency or amplitude.. this was our hand book http://users.physics.harvard.edu/~horowitz/aoe/

I'd try a 1000uf, afterall we are troubleshooting, have to see what works and doesn't  

here's the pic of the spreader removed

http://s18.postimg.org/py0qzeio7/1_kch2zw_Ak_GWOPBBu2o_Sg_OEg.png

looks like he used that little precision screwdriver to pry on the upper left corner, I am not saying to try it, just showing you the picture  -- I've wanted to look under there myself but don't want to risk losing a good board

and I guess that's not a Titan board, maybe a Neptune??

Ok this is a Neptune board. The BTC logo tells me that. Only Neptunes have the BTC logo on them.  I was trying to find which cores include die 0 and 1, as there are 2 dies per core. Die 0 and die 1 are by far the most troublesome, and seem to always fail. I THINK its the core on the top left corner. Can anyone confirm this? Anyone else notice in this pic that the top left core is the only one that does not have thermal grease on it? I suppose the tech could have cleaned it, but then why just clean one? Is it possible there is a fault in the design where the plate does not meet the core on the top left , which I think is contains die 0 and 1? I think the tech in this pic is showing a major fault in design, that is the reason for this pic (im guessing) .This would explain a WHOLE lot! Guys im not suggesting to remove your plates, and there is a strong possibility it will damage your cores.

Vegas

Looks like they are using thermal pads to fill the gap between the chips and the plate, could also be using thermal paste ...thermal pads come in different thicknesses and are pretty expensive for the good ones with high heat transfer, which still aren't as good as thermal paste

actually dies 0,1,2,3 are the problem dies for me. It looks like the Neptune has 4 mounting holes for the heat sink, the titan has only 3. Maybe the 2 chips for dies 0,1,2,3 are at the top, where there is only one mounting hole in the middle, so maybe those 2 chips are not getting enough pressure on the pads??

I solved all the vrm heat problems by using teslaman's mods in this thread and using high speed fans, I didn't do it exactly the way he did. I only fixed the ones that were running hot and I didn't epoxy everything. If they only needed to come down a few degrees I used the thermal pad that came with the heatsink. The epoxy will knock it down 10c or more if you use it.  I also found some .25 in heatsinks that fit under the eplate, but they don't work as well as the bigger ones where you need to modify the eplate

the funny thing is except for the high speed fans, some cubes just don't need any mods, they just run cool ... others you need to mod everything in sight

[GenTarkin]

ANNOUNCING v.99 firmware:
https://github.com/GenTarkin/Titan/releases/tag/v.99

Release notes:
Changes for v.99:
After a failed original v.99 release, this is the new version with fixes for OTA & .bin upgrades!!

NOTE: If upgrading from v.98 via OTA, it "WILL" say theres an error in red letters and show error: as error result. But, if everything above the red letters doesnt state an error then its safe to reboot - there was no error. If upgrading using .bin file, there should be no issues.

NOTE: if anyone upgraded to v.99 prior to this new release, you will need to upgrade again using the .bin file listed on the page linked above!

Implementation of new graphics. Removal of KNC branding. Rewording of warning message on SYSTEM page.
Implemented auto increase of die clocks which have been underclocked due to DCDC overheat.
Created upgrade packages to allow installing my firmware via SYSTEM -> UPGRADE(knc's firmware upgarde method).
Various code fixes for monitoring script, donations & webgui.
Creating Backup now saves voltages, die clocks & minername. Minername now saved as part of backup filename.
Branding now enabled for all builds.
Groundwork started for enabling / disabling lighttpd access logging(disabled saves SD card life).
NOTES:
.bin file can be used to upgarde from STOCK KNC 2.00 firmware, the process takes a good few minutes.
.img file will be uploaded on request but is no longer needed really due to .bin being able to upgrade stock firmware to my modified firmware.
If upgrading from previous versions of my firmware then using the upgrade button on status page should upgrade to v.99.

Reboot after upgrading!
Note: if upgrade happens to fail repeatedly then redownload full v.98 img and attempt upgrade again.

[vegasguy]

So im am replacing some surface mount resistors on the ericson. I bought some new off ebay a few months ago, and Im  out. I cant remember the size or specifications or that resistor. Does anyone know? I tried to go to Ericsson's website and look up the specs on the 764-BMR4641002 to find out the surface mount resitor as there are no markings on it.

Vegas

[mindtrip]

Here is a Titan controller board if anyone needs one: 


http://www.ebay.com/itm/KNC-TITAN-CONTROLLER-BOARD-/221867035009?hash=item33a84e4981

this may be a dumb question but could someone just 'homebrew' clone such a board....? Not saying it is practical but for folk on here with the skills and such
is such a thing possible ..even in a mad scientist 'breadboard' type of manner?

Just curious is all with all the electronics talk on Titan fixes how far someone could go with all this ...expertise wise on this thread

Sure you can all you need is an old Neptune controller and a Pi only tricky part is the card to interface the pi tot he old board

[TXSteve]

So im am replacing some surface mount resistors on the ericson. I bought some new off ebay a few months ago, and Im  out. I cant remember the size or specifications or that resistor. Does anyone know? I tried to go to Ericsson's website and look up the specs on the 764-BMR4641002 to find out the surface mount resitor as there are no markings on it.

Vegas

good luck non of that stuff has markings, maybe try to find an old cube for parts, maybe a Neptune would work

[GenTarkin]

UPDATE, small error found in previous v.99 upgrade. For anyone that has updated so far, please reapply the .bin found here:
https://github.com/GenTarkin/Titan/releases/tag/v.99

BTW, as of now the enable / disable lighttpd option under SYSTEM tab works =)

[vegasguy]

So im am replacing some surface mount resistors on the ericson. I bought some new off ebay a few months ago, and Im  out. I cant remember the size or specifications or that resistor. Does anyone know? I tried to go to Ericsson's website and look up the specs on the 764-BMR4641002 to find out the surface mount resitor as there are no markings on it.

Vegas

good luck non of that stuff has markings, maybe try to find an old cube for parts, maybe a Neptune would work

Steve thanks for your response. I wanted your opinion on something:

Ive carefully looked at these vrms and it appears that they could be desoldered with a professional desoldering tool like the Hakko FR-300

Take a look at the pic of our vrms :

(TOP):

Ive marked the holes in red that I believe a desolder tool with a built in vacuum could remove all the solder out, releasing the VRMs from the board for repair, and MAYBE upgade.


 (The underside):

What are your thoughts Steve? Anyone else is welcome to chime in with info.

Vegas

[TXSteve]

it might work, I'm not sure how that vrm is mounted if you can get everything from the top, you might need to preheat the board too

if you have trouble because of the thickness, if you get as much out with the hakko as you can, fill the joint with chip quik to dilute that no lead solder, and hit it with the hakko again, bet that would do it

https://youtu.be/H0m_d52m1DU

https://youtu.be/gZxqpH23ZEM

another cool hakko tool, this one's for removing smd's:    https://youtu.be/nKxS6yNqess

[frenchy]

Hello vegas i was searching about dcdc modules and found this page where you can download the datasheet of BMR m46X series.Don't know if it will help...this is the link:
http://www.ericsson.com/hu/ourportfolio/products/bmr464-series?nav=fgb_101_378|fgb_101_379

[vegasguy]

Thanks frenchy. I saw that. It doesnt say anything about the surface mount resistor size. Ive ordered that professional desoldering tool , the Hakko FR-300 because I feel strongly that I can remove the entire vrm and replace. So that tool should be in in the next few days. IF I can remove the vrms, this will help fix alot of our problems, AND we might be able to upgrade to the 50A version of the vrms THAT would solve LOADS of problems!

Vegas

[vegasguy]

Ok so ive mapped the die locations. see pic. The readings to the far right of the screen are the temps for the dies and NOT the vrms. There are no temps for the vrms.

I turned off all the dies except 2 dies at a time. I then use my thermal laser gun to find the only hot spot on the plate, that told me where the dies where. Anyway here you go:

[Searing]

Thanks frenchy. I saw that. It doesnt say anything about the surface mount resistor size. Ive ordered that professional desoldering tool , the Hakko FR-300 because I feel strongly that I can remove the entire vrm and replace. So that tool should be in in the next few days. IF I can remove the vrms, this will help fix alot of our problems, AND we might be able to upgrade to the 50A version of the vrms THAT would solve LOADS of problems!

Vegas

Just so it is on this thread someplace I have to say the following : Ahem

"My god man do you have BIG BRASS ONES IMHO...sheesh........"

Ok just figured someone had to say it in that ..if you guys are like me .. I rarely make direct eye contact with my Titan(s) in case they become offended and brick
themselves into door stop status...I mean MAJOR SURGERY the mind boggles

did you try it with 1000uf??

google says: sometimes a higher uf is acceptable, but not always ... depends on the circuit   :|   isn't 1000uf to 1500uf a pretty big jump?

Google is correct. We can make circuits for noise reduction and for another purposes. So if this is noise reduction purpose then there will be the lower limit for capacitor NOT LESS, but if it is some oscillating circuit then higher capacitor will change the frequency or amplitude.. this was our hand book http://users.physics.harvard.edu/~horowitz/aoe/

I'd try a 1000uf, afterall we are troubleshooting, have to see what works and doesn't  

here's the pic of the spreader removed

http://s18.postimg.org/py0qzeio7/1_kch2zw_Ak_GWOPBBu2o_Sg_OEg.png

looks like he used that little precision screwdriver to pry on the upper left corner, I am not saying to try it, just showing you the picture  -- I've wanted to look under there myself but don't want to risk losing a good board

and I guess that's not a Titan board, maybe a Neptune??

Ok this is a Neptune board. The BTC logo tells me that. Only Neptunes have the BTC logo on them.  I was trying to find which cores include die 0 and 1, as there are 2 dies per core. Die 0 and die 1 are by far the most troublesome, and seem to always fail. I THINK its the core on the top left corner. Can anyone confirm this? Anyone else notice in this pic that the top left core is the only one that does not have thermal grease on it? I suppose the tech could have cleaned it, but then why just clean one? Is it possible there is a fault in the design where the plate does not meet the core on the top left , which I think is contains die 0 and 1? I think the tech in this pic is showing a major fault in design, that is the reason for this pic (im guessing) .This would explain a WHOLE lot! Guys im not suggesting to remove your plates, and there is a strong possibility it will damage your cores.

Vegas

OK don't laugh but on this thread someplace someone said that they could not get an RMA on a cube of some kind (think it was a Titan but maybe not) from KNC because they
simply told him to TIGHTEN THE SCREWS down on the plate thingy (hey be careful this is 20/20 hindsight remembering here) he was like wtf? i just remember because it was such
a lame way not to do his RMA

but anyway that would go along your idea above I guess? (don't try this just tossing it out as a clue to how this stuff hangs together in case it is important for another solution)

[LongAndShort]

Thanks frenchy. I saw that. It doesnt say anything about the surface mount resistor size. Ive ordered that professional desoldering tool , the Hakko FR-300 because I feel strongly that I can remove the entire vrm and replace. So that tool should be in in the next few days. IF I can remove the vrms, this will help fix alot of our problems, AND we might be able to upgrade to the 50A version of the vrms THAT would solve LOADS of problems!

Vegas

Just so it is on this thread someplace I have to say the following : Ahem

"My god man do you have BIG BRASS ONES IMHO...sheesh........"

Ok just figured someone had to say it in that ..if you guys are like me .. I rarely make direct eye contact with my Titan(s) in case they become offended and brick
themselves into door stop status...I mean MAJOR SURGERY the mind boggles

did you try it with 1000uf??

google says: sometimes a higher uf is acceptable, but not always ... depends on the circuit   :|   isn't 1000uf to 1500uf a pretty big jump?

Google is correct. We can make circuits for noise reduction and for another purposes. So if this is noise reduction purpose then there will be the lower limit for capacitor NOT LESS, but if it is some oscillating circuit then higher capacitor will change the frequency or amplitude.. this was our hand book http://users.physics.harvard.edu/~horowitz/aoe/

I'd try a 1000uf, afterall we are troubleshooting, have to see what works and doesn't  

here's the pic of the spreader removed

http://s18.postimg.org/py0qzeio7/1_kch2zw_Ak_GWOPBBu2o_Sg_OEg.png

looks like he used that little precision screwdriver to pry on the upper left corner, I am not saying to try it, just showing you the picture  -- I've wanted to look under there myself but don't want to risk losing a good board

and I guess that's not a Titan board, maybe a Neptune??

Ok this is a Neptune board. The BTC logo tells me that. Only Neptunes have the BTC logo on them.  I was trying to find which cores include die 0 and 1, as there are 2 dies per core. Die 0 and die 1 are by far the most troublesome, and seem to always fail. I THINK its the core on the top left corner. Can anyone confirm this? Anyone else notice in this pic that the top left core is the only one that does not have thermal grease on it? I suppose the tech could have cleaned it, but then why just clean one? Is it possible there is a fault in the design where the plate does not meet the core on the top left , which I think is contains die 0 and 1? I think the tech in this pic is showing a major fault in design, that is the reason for this pic (im guessing) .This would explain a WHOLE lot! Guys im not suggesting to remove your plates, and there is a strong possibility it will damage your cores.

Vegas

OK don't laugh but on this thread someplace someone said that they could not get an RMA on a cube of some kind (think it was a Titan but maybe not) from KNC because they
simply told him to TIGHTEN THE SCREWS down on the plate thingy (hey be careful this is 20/20 hindsight remembering here) he was like wtf? i just remember because it was such
a lame way not to do his RMA

but anyway that would go along your idea above I guess? (don't try this just tossing it out as a clue to how this stuff hangs together in case it is important for another solution)

Yeah i remember that person a few months back. If one is not keen to rma; due to costs. Generous repasting is the way to fix some issues of dies shutting down. The copper pipes are machined but what they have done is leave a small plate on there and machined it also. Thus resulting in an uneven finnish in some/most cases.

Also, i found taking off the foam from the sides because even with the screws tightened down the whole heatsink shifts extremely easy. So putting the cover back on puts pressure on that foam and moves the heatsink. The suggestion was to tighten the screws and possibly add a washer to get more leverage. (that does not void the warranty) Repasting does i think but its $24 for paste vs $200+ to RMA.

I have burnt vrm moduals on some of mine so i am not sure what to do there i do not have a de-solderer.

[frenchy]

Woow vegas thank you good work!!!Do you think there is a way to do the same with dcdc modules to assign them with each dies?

[zybc]

UPDATE, small error found in previous v.99 upgrade. For anyone that has updated so far, please reapply the .bin found here:
https://github.com/GenTarkin/Titan/releases/tag/v.99

BTW, as of now the enable / disable lighttpd option under SYSTEM tab works =)

how long should it take when upgrading 2.00 ver??
I am looking for 2 hours at this line

System upgrade in progress
Copying files... files copied No wget found, syncing apt & installing wget Syncing apt-get...

Can I refresh a page or smth?

[Westerstar]

https://www.youtube.com/watch?v=O-RBI4am-wY

A guide on how to easily get the temperature down on a KNC Miner Titan's Cube in under 30 minutes.

The guy talking swedish, of according to him, The Cube will run 15C degrees cooler
Interestingly, with the modification that he does on the dc / dc converter

Maybe this isn't news for you guys, but I want to share it with you

Sorry for my bad english (Im from Sweden)

[vegasguy]

Just so it is on this thread someplace I have to say the following : Ahem Smiley

"My god man do you have BIG BRASS ONES IMHO...sheesh........"

Ok just figured someone had to say it in that ..if you guys are like me .. I rarely make direct eye contact with my Titan(s) in case they become offended and brick
themselves into door stop status...I mean MAJOR SURGERY the mind boggles Smiley

Ohh ya, in case I wasnt clear, I will be removing the center plate to examine the cores from a mostly working cube today. Has .8v on dies 0 and 1. Will any of this void my warranty?  LOL

Update: Center plate removed. This was from a from a cube that had .8v on die 0 and 1. As you can see what WAS thermal grease is now just dried dust. So the question on everyones mind is: What will happen, if I clean and regrease the cores? Give me some time and Ill let you know, today. Do NOT attempt this!!! WAIT for the video! There are MANY wrong ways to do this and 1 right way. Do it wrong and you have a dead cube!!



Vegas

[elrippo]

Hy Guys,
is there a chance that KNC comes up with a affordable miner for home miners since bitmain launched their Antminer S7?

Kind regards,
elrippo

[TheRealSteve]

is there a chance that KNC comes up with a affordable miner for home miners

There's always a chance.  It's just highly unlikely given how they've burned consumers, have issued news releases indicating that they've moved away from physical sales and are transitioning away even from cloud mining contracts, and the general headache they would be returning to if they did.