Hacking KNC Titan / Jupiter / Neptune miners back to life. Why not?

[lightfoot]

And in the meantime I've been working hard on a way to isolate chips. To that end I finally took the top off a chip to see what's up.

Here it is.



Note removing it was a bitch on wheels, however it told me a few things:

1) There are only four small dies on a chip, location of each is where the heat is. Focus your heat sinking on that.
2) If the board tacos and the dies lose contact with the lid you're fucked (what I just fixed by de-tacoing)
3) Whatever that glue crap is required almost 500c of top heat after soaking the chip at 360c. Even then was not enough to keep the fabric mat from delaminating.
4) The mat is interesting. I can't see the possibility of shorts inside the chip but outside die so shorts must be hard inside the die. Maybe I could blow them clear with enough low voltage current.
5) Cheap way to do it, the power bus underneath must have been a bitch. Maybe the chips needed the subframe support.
6) Need to find a solvent to dissolve the glue without damaging the dies. Then I could just pull the top, rip off the bad die, and go to town. Any thoughts?

[lightfoot]

And the most interesting thing under a loupe: See those little balls around the dies? Those are solder balls. Probably from under the dies, extruded either due to overheating while pulling the top or overheating while running. That could short out a die very quickly, and could be the source of a control line short to ground.

Hm.

[lightfoot]

So pulled one of the dies off this evening. Not easy, I had to get under it with a knife after cutting around it, that ribbon/mask is of course glued to the side of the chip....

The dies themselves are a micro-sized BGA, never seen anything quite like it. So small you can only see the grid with a loupe. But yes, that solder in there can melt and when it does it smears. Most times it just shorts out the chip, sometimes it hits the one signal to spi (mumbles), and sometimes it hits the supply for the die (which shorts the die). Reballing would be interesting, less than a .2mm pitch.

In further notes, heating just that corner of a chip is not enough to clear a short. Damn.

[smacker]

Im back again. This night the miner stopped working again.

Two short white flashes from the led, and no cubes showing up on the status-page. Is the controller-board broken for real now?

[lightfoot]

Im back again. This night the miner stopped working again.

Two short white flashes from the led, and no cubes showing up on the status-page. Is the controller-board broken for real now?

Yep. Interesting. Is the FPGA hot?

Did it not work with just the one cube by chance?

[smacker]

Im back again. This night the miner stopped working again.

Two short white flashes from the led, and no cubes showing up on the status-page. Is the controller-board broken for real now?

Yep. Interesting. Is the FPGA hot?

Did it not work with just the one cube by chance?

Is FPGA the big black chip? That one is cold.
How does this happen? This was the second board in one week of mining. They are used before of course. Im running 4 cubes on 2x 750w PSU, can that have some impact?
It does not work with one cube either.

[lightfoot]

Im back again. This night the miner stopped working again.

Two short white flashes from the led, and no cubes showing up on the status-page. Is the controller-board broken for real now?

Yep. Interesting. Is the FPGA hot?

Did it not work with just the one cube by chance?

Is FPGA the big black chip? That one is cold.
How does this happen? This was the second board in one week of mining. They are used before of course. Im running 4 cubes on 2x 750w PSU, can that have some impact?
It does not work with one cube either.

Hm. So you blew out two boards in a week? Odd. One of your miners must have a short or something that's taking out the FPGA signal lines, they are *not* buffered.

Before we blow up repairs, can you use an ohm-meter to test resistance from ground to each of the 10 pins on the 10 pin plug on each miner? One of them will not be like the others....

C

[smacker]

Im back again. This night the miner stopped working again.

Two short white flashes from the led, and no cubes showing up on the status-page. Is the controller-board broken for real now?

Yep. Interesting. Is the FPGA hot?

Did it not work with just the one cube by chance?

Is FPGA the big black chip? That one is cold.
How does this happen? This was the second board in one week of mining. They are used before of course. Im running 4 cubes on 2x 750w PSU, can that have some impact?
It does not work with one cube either.

Hm. So you blew out two boards in a week? Odd. One of your miners must have a short or something that's taking out the FPGA signal lines, they are *not* buffered.

Before we blow up repairs, can you use an ohm-meter to test resistance from ground to each of the 10 pins on the 10 pin plug on each miner? One of them will not be like the others....

C

Havent got an Ohm-meter, but I am pretty sure that one of the cubes was damaged. Is there spare parts to order? Which one is the FPGA?

[lightfoot]

The FPGA is the big chip. Do you know how to remove BGA chips from boards, clean up and reflow? I think I put the basic steps in the thread here somewhere.

If you would like you can send one of them in and I can test it with some stuff here. PM me and I'll take a look.

[lightfoot]

Ok, this is weird. Checked in on the Titan last night after a warm day. Titans were hashing fine but the light on the control board was bright GREEN. Rebooted, little green light came on, hour later it was BRIGHT BLUE. I have seen bright RED, GREEN, WHITE...

Is this normal?

[Searing]

Ok, this is weird. Checked in on the Titan last night after a warm day. Titans were hashing fine but the light on the control board was bright GREEN. Rebooted, little green light came on, hour later it was BRIGHT BLUE. I have seen bright RED, GREEN, WHITE...

Is this normal?

er ...was it hashing? if so ...unsure....if that is the only symptom...not sure I've ever paid attention. For all I know mine does so too.

by the by had some issues with my Titans in last month 2x showing DISCONNECT and same on reboot (nothing like that to get the old blood pumping wth)

happened 3rd time tonite...s

Anyway, I used my head for once and rebooted my usb hub and also power to the charter cable (I have a static IP) and viola as per the last 2x back they came....

went into the hub of mine using Admin and looked at the manual settings... and double checked that the firmware was Newest (big whoop 2014 or some such) but latest

But as I hoped it was a simple problem ....it was as I suspected on a DIR-615 D-Link Hub ...there is a setting to SET your IP to a device with the MAC ID

and SAVE that IP as such ...so now 'where the hell did it go' when your hub 'farts'

so putting them in the table as such and matching the mac ID and IP to the Titan 1 and Titan 2 again  (previously showing up but un-named themselves off the table)



everything else was kosher and still set for my other stuff and IP's ..so it ..might have been something I did moving stuff around on the Titan Upgrade to Gen Tarkin's 2.0.0

(likely) ..then again the units are 'evil'


Anyway.....just something to look at first before making a NEW SD card or some other silliness if you get a big DISCONNECT message on continued Reboots and power downs.

and a reset of power off of your HUB and/or the cable modem brings it all back ..again I'm on static IP ....if that matters to anyone with same issues



(Every day ..every day....these Evil miners find new ways to give me a heart attack...never really die....but they are like the bi-polar girlfriend ....with the threats

too cut your manhood while you sleep.....in the titan's case .....ASIC suicide via bricking themselves...or in the Titans case..'show me some attention' ...always the 'threat' never the follow through..(yet) )

Drama Queens

sheesh

[lightfoot]

Hashing fine, now it's orange. Eh, I'll leave it alone for now but it is odd. Cranked the units down below 60mh and of course it's cold outside now....

[Searing]

Hashing fine, now it's orange. Eh, I'll leave it alone for now but it is odd. Cranked the units down below 60mh and of course it's cold outside now....

I know ONE of the incarnations of knc firmware the light I think was blindingly white and stayed on if I remember correctly ..till they made a patch and next firmware upgrade
not sure how/what or if that could be related ..but only light related issue I've ever heard of with these...never heard of the rainbow you have

[lightfoot]

Hashing fine, now it's orange. Eh, I'll leave it alone for now but it is odd. Cranked the units down below 60mh and of course it's cold outside now....

I know ONE of the incarnations of knc firmware the light I think was blindingly white and stayed on if I remember correctly ..till they made a patch and next firmware upgrade
not sure how/what or if that could be related ..but only light related issue I've ever heard of with these...never heard of the rainbow you have

Think I figured it out, it was cold the past day or so and I had the door to the mining stuff closed. Small wireless bridge in there to the main network, it was buried under stuff and the network connection was unstable. Put on table, seems happeir now.

Might have been signaling a intermittent network connection or something. No biggie.
C

[joef]

Disappeared DCDC on advanced page
I have noticed, a DCDC converter is not listed on the advanced web page if the DCDC PMbus of the missing device does not answer.
/home/pi/knc-asic/waas/waas -g all-asic-info
reports

i2c_smbus_read_word_data failed: addr 0xD0
i2c_smbus_read_byte_data failed: addr 0x98

In that case a power cycle is required to get the DCDC working again.
This may be a reason for a hanging DCDC:
On specification page 5:
Each DCDC converter shall be connected to

External CIN = 470 μF/10 mΩ, COUT = 470 μF/10 mΩ. See Operating Information section for selection of capacitor types.

Operating instructions are on Page 41:

A minimum capacitance of 300 μF with low ESR is
recommended. ..... For high-performance/transient applications or
wherever the input source performance is degraded, additional
low ESR ceramic type capacitors at the input is recommended.

The ceramic capacitors are on the board, near to each DCDC converter as recommended. But instead of having a capacitor for each DCDC converter, they have inserted a single capacitor with higher capacity for all DCDC together.
But the way from the capacitor to the DCDC converter creates additional inductance.
While the capacitor is getting older, the stable frequency is changing.
That may be a reason for these islands of stability on Titans for some frequency ranges.

The Ericson DCDC specification is available for download at
http://archive.ericsson.net/service/internet/picov/get?DocNo=28701-EN/LZT146435&Lang=EN&HighestFree=Y

[ph111]

Hi i have a pile of titans i need fixing 4 or 5 would like to post 1 to start to you how to go about it please which country are you in sorry haven't read the whole thread but if you could help might as well get this pile of shit back working if you can 
Thanks P

[mmfiore]

Hey Lightfoot!

Just cranked up the board that you repaired. It is working great. I have now implemented a voltage check on the cable before connecting in just to make sure wiring is correct. All is good thanks again...
 

[lightfoot]

Hi i have a pile of titans i need fixing 4 or 5 would like to post 1 to start to you how to go about it please which country are you in sorry haven't read the whole thread but if you could help might as well get this pile of shit back working if you can  
Thanks P

Sure, what seems to be the problems with them? Any of them short out the power supply when plugged in?

I'm in the US, not too difficult to ship where are you and can you do stuff like take the cubes apart to just send the boards?

[Searing]

Better late then never. I got around finally to putting in an electrical sub panel for my 100 amp service (you know you drank the kool aid when you put in an 8 slot sub panel

on only a 100 amp service that already HAD 32 slots..sheesh) Talk about frigging overkill!



anyone needs this done ..it is easy ...100 amp service ...sub panel to 50 amp breaker on main service...directions on YouTube easy peazy



Anyway got enough to know run that 220v line across the basement and FINALLY got around to testing Qberty's clone Titan Bridge.

Anyway .....take Neptune 6 port card...remove (carefully..wiggle slowly .may take 5 min) the beagle bone black ....put the model Raspberry PI 512mn B+ into the

clone bridge.....and again very carefully work the combo PI/bridge into the 6 port Neptune card. By the By left the LCD in works just like the Titan one then I

tossed up KNC IMG on a Sandisk 8gb Class 10 and VIOLA my creation WORKS!




I then tossed up my 1 cube with only 1 working die (as a potential sacrifice to the ASIC gods ..ie boom) but all was well came right up ...I dared to dream

and put the 1 die at 325 adv setting and -0.0366 .......brave I was ....over clocking this 1 working die and all Will put the tarky titan software on (3rd license)

when I get time....but anyway chugging along nicely




Still think a guy could just get a Raspberry Pi protoboard for like 25 bucks and just wire this up by hand and copy the PI to daughter board pin outs...BUT there is a small
chip on the daughterboard ...someone said it was a clock chip...but again....if someone knew electronics and could find the equiv clock chip (not so small hopefully) and just
show what leads this would be soldered to.....it would be a 50 buck hack  me thinks.

Qberty's clone bridge is the best option ..but can't see the numbers to get it lower then 0.5 BTC or about 210 bucks. In that there is I think a buy it now on ebay for
at Titan Board for 250 or so ...well these titans are not getting any younger....2 bad i have a Saturn 4 port and Jupiter 4 port sitting about here as well


unless the mythical BBB to Titan cube project comes to pass (knock wood) such sacrilege in knc's view thou shall not 'suggest' such (again told this was impossible
by knc techs...also told it was impossible to just snap in the 4 and 6 port fixes).......wish I could find that frigging email ...someone had a broke controller he had to send
it back rather then just use his Jupiter board...i miss knc forums ...lots of knc FUD Lies etc .but some other good tech info lost imho


anyway KNC being cheap ass no need to re-design such for Titans ...gives me a 3rd controller on the cheap ...so what the hell..now I can move and dork with 1 cube at a time w/o the whole shebang having to go up/down for testing a cube say..heh well gen tarkin's 3rd license is now gonna be used!

my complete setup

lostgonzo.imgur.com


Also STILL looking for KNC Y Adapters NEW. If you have any laying about from when KNC sent them to you 3 months late and never used them. PM me what you need for 1 or more.





(The ONLY thing I ever got from KNC at no cost by the by)

 
Damn..again with the "Quest for Cubes" adventure to make it look less...well lame with only 1 cube

As always you don't like my rants feel free to plop me on /ignore on here...I don't mind





edit: FML the 1 die working cube has been up on its lonesome on the titan/jupiter hack for  6.6 hours and found a frigging Block..unfortunately I'm on www.litecoinpool.org

(the ASIC gods always mucking with me)

[smacker]

Ive got 4  Neptune cubes that won´t start. I run the standard -0.0366 and I have tested with different MHz.
Opened them up and cleaned them out of dust and changed the heatsink-paste.
One of them seem to have those brown rings under the asic-chip. Not completely burned, just a hint of brown.
Is there anything to do to make them work?