Topic: Swedish ASIC miner company kncminer.com - page 77. (Read 3049514 times)

Please attempt installing the new released v.99a firmware .bin file here:
https://github.com/GenTarkin/Titan/releases/tag/v.99a

It wont attempt to install wget anymore, should have no problems with this release.
Upgrade should finish in a few mins.

Hey vegas and gentarkin ,you have a lm75 on the controller board too!!!just take a look between the lcd display and the Rpi...

Announcing new firmware v.99a:
Quick, incremental fix for failed installs when hanging on wget installation(apt-get syncing etc...)
No longer relys on wget being installed to download files when using STATUS->upgrade.

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

A schematic might help.

HAHA! Figures, well it must have a remote temp sensor somewhere. Would have to trace the traces... which being a multilayer is impossible.
I bet the sensor is somewhere or in the actual ASIC package.

It would be super retarded if KNC just put that there and the sensor was close but all it did was measure ambient, lol Id wanna burn their farm down then, oh wait I already want to...

ok location of LM75:



close up of LM75 (you can hardly read it, I had to use a magnifying glass) :

Alright, found a way to actually download the file using curl despite the obscure URL which is why I had problems w/ curl previously. I will code up a new upgrade method and package and make a new incrimental release. Will do away w/ the questionable reliability of sync'ng apt and trying to install wget ... yay!

BTW, if more and more people experience issues w/ wget installing, then for the next release Im gonna go ahead and remove the upgarde button option located beneath my firmware version on status page. People will have to upgrade using .bin file method instead.

Ok I can run it with plate only and give a readout of what my temp gun says and what "temp" says on status screen. Ill let you know. If these are within 1C +/- I would like to see it reworded "average die temps" on status screen. There is no chips mounted under the pcb, Ill check the layout of the board, and post a screen shot too and see if we can locate it.

Vegas

Just to add my 2 satoshis here. From what I've seen tinkering with these things is that the dies pull more power when they get hot. I had a couple of cubes damaged with the screw fixing at the back coming loose. Dies 1 and 2 would come on then the amps and watts of the DC/DC / VRM's would shoot up to 50 then the dies would go down/offline when watching the advanced page.

I took the heatsink off replaced the thermal paste then secured it back in place and the dies came back on as normal but no longer shot up in power or temperature. managed to get all but 1 die back and working, now its running stable with decent temps.

The pages are already correctly labeled LOL! DCDC = VRM's ... loosely interchanged to refer to the same thing.
The temperature column on the status page grabs those temps directly from a LM75 sensor... I dont know where this sensor is on the board, could u inspect the underside of the board perhaps? maybe there is 1 LM75 sensor mounted beneath the ASIC package. Its reading temps through the board, or maybe somehow the LM75 is integrated within the ASIC package.
But there has to be one LM75 sensor somewhere on these damn boards.
Either way, I suspect its an external sensor for the ASIC package. It functions to readout external package temps which are usually a good 10-30C less than actual die temps.

Out of curiousity how much did your temps change by when u slid the HSF around?
I have 2 theories,
1. when sliding HSF around this could potentially change airflow caused by your hand etc... and therefore the DCDC's were being cooled more or not depending on that airflow aversion.
2. If slid the HSF caused temperatures to skyrocket for a short period the dies may have went into a limp mode to avoid being blown up.

I suspect there is no on die temp sensor, otherwise we would see it being used. We arent.
Hey vegas, w/ ur thermal sensor thing, can you read temps from atop the base of the HSF while its hashing? Actually, if u could aim the laser at the package underneath the heatsink as its hashing, that would be really telling. If its close to the readout of TEMPERATURE on status page then the ASIC is using a single external package temp monitor. (thats already my suspicioun anyways, but it would prove the theory)
What I wish we knew was the TDELTA between the dies & the package temp sensor. If I knew what that DELTA was on average, I could code it into the firmware to reflect the correct average temp of the dies.

Bummer, do you only have 1 titan or more? If more then have you tested it on the others? Also, what method is the Pi hooked up to internet, wifi or eth?

This screen stays for an infinite time
http://static1.keep4u.ru/2015/08/31/Asic-44.th.png
Waited 3hrs then 5hrs..
internet is 6Mbit and SD 10class NEW one!

Just dont make eye contact

Vegas

Ok found this and Tarkin looks like you are correct the 8 temps in advanced page are temps for VRMS. https://bitcointa.lk/threads/swedish-asic-miner-company-kncminer-com.100083/page-1804
So It would be nice to label that properly as VRM Tempretures. I also agree 2 vrms per 1 die. But if its possible I would like to see the temps of the 4 dies displayed indvidually if possible on next release as seen on earlier link. Its probably from a jupier or neptune, so I hope we can do it on the titans.

Vegas

I am not sure, everything is running fine at 325 MHz, so I don't want to think about it too much lest they get mad ...

agreed, and I think you might be right. I found this. Its old but might have some helpful information. https://bitcointalksearch.org/topic/knc-asic-users-thread-faq-307055. So are you saying in the advanced page , the temps to the far right are the vrm temps? If thats true, and I think it might be, where are the temps of the 4 cores found, or do they average it on the status page? I think the next update might need clear labels on these numbers , as KNCminer gave us temps, but for what!!? Everything is not labeled in the miner, and its needs to be to clear up all of this confusion.

Vegas

Lets not confuse things, dies = physical seperated pieces of silicon.
cores = logical/functional seperation of logic inside dies.

Ok so lets discuss the advanced page. To the far right, these temps are the temps for the  8 dies. The reason I say this is because I had my heatsink loose when hashing, when I slid the heatsink slightly forward the temps in advanced would change. Ive also removed the heatsink I had attatched to the vrms and the temps did not change in advanced page. I beleive the dc/dc temp of the vrms is in status page. Thats why I believe its not the temps of the vrms on the far right in advanced, its the temp of the individual dies. I know there are 4 cores as Ive had the cap off. I believe there are 2 dies to each core, and 1 vrm per die. Ill do more research on this.

Vegas

Well thats odd. Either way, in the source code, the 2 temp fields are derived from reading the DCDC's internal temperature register.
The temperature shown in status page under "Temperature" column is derived from LM75's temperature register. Is LM75 the FPGA?... Im not sure but after seeing this: http://datasheets.maximintegrated.com/en/ds/LM75.pdf
If that is identical to what our LM75 is ... then Im inclined to think the LM75 is some sort of external temp sensor that maybe sits as close as they can to their respective places around the dies?

I dont see anything labeled LM75 on the picture u showed, are there any surface mount pieces on the bottom side of the PCB? Maybe LM75 is integrated into the package somehow, doubt it tho. Im just guessing tho =)