Bitcoin Forum
September 15, 2019, 01:19:43 PM *
News: Latest Bitcoin Core release: 0.18.1 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 ... 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 [78] 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 ... 160 »
  Print  
Author Topic: New Official AMT Thread  (Read 149167 times)
This is a self-moderated topic. If you do not want to be moderated by the person who started this topic, create a new topic.
mrpark
Sr. Member
****
Offline Offline

Activity: 364
Merit: 250


View Profile
May 22, 2014, 01:44:40 PM
 #1541

Photos as promised... boards were damaged in shipping..

broken off pin connector: (notice 3 of the boards have one kind of heat sink, the others have a different kind, not sure what's up with that?)



missing cap (not a missing cap in this photo, but there is a missing on one the edge of the board somewhere) and PCI-E connector missing.



no heat sinks on many of them... Different heatsink of different boards...



What might work instead, see the A1 with no grease or heat sink.... two 8x30x8mm Aluminum heatsink




BTC: 1JDjCGtxtxoZ46XgTqUoXBDxNFKwcsEmik
1568553583
Hero Member
*
Offline Offline

Posts: 1568553583

View Profile Personal Message (Offline)

Ignore
1568553583
Reply with quote  #2

1568553583
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
1568553583
Hero Member
*
Offline Offline

Posts: 1568553583

View Profile Personal Message (Offline)

Ignore
1568553583
Reply with quote  #2

1568553583
Report to moderator
1568553583
Hero Member
*
Offline Offline

Posts: 1568553583

View Profile Personal Message (Offline)

Ignore
1568553583
Reply with quote  #2

1568553583
Report to moderator
FrictionlessCoin
Legendary
*
Offline Offline

Activity: 854
Merit: 1000


★YoBit.Net★ 1400+ Coins Exchange


View Profile
May 22, 2014, 01:47:32 PM
 #1542



Opieum,

The AMT script is design to run with the UI only. So basically if you log in directly and using the:

killall cgminer

amt-setup

(what ever else)

and you run from the command line - you'll be using default bitmine values. The script itself starts with the automatically when the miner first starts, killing it or logging in directly via the pi null and void this script. You can refun this, but the best way to see if its working/having any change is to hook up the second monitor to the pi and also have your computer handing, work on the pi itself, use "reboot" to reboot the miner itself, then do not login to the pi but to the admin panel on the browser and run the miner that way. You can always check the activity in the settings section of the UI as well.

...

The A1 disables itself and the entire chain if it reaches the point of overheating (or when it thinks it's overheating), we've always had a gut feeling that chip parameters later set are somehow remembered physically and are not being told to "reactivate later" hence getting the "no a1 chip chain detected".



We all need a decent description of the programs that are on the system.

Can cgminer be run from the command line or do we have to run it via the webserver?


FrictionlessCoin
Legendary
*
Offline Offline

Activity: 854
Merit: 1000


★YoBit.Net★ 1400+ Coins Exchange


View Profile
May 22, 2014, 01:50:07 PM
 #1543

Photos as promised...

broken off pin connector: (notice 3 of the boards have one kind of heat sink, the others have a different kind, not sure what's up with that?)

.... (snip)


Why are the heatsinks here different from the heatsinks that we have?  Why so much inconsistency in the build?

BTW,  you may need to look carefully in the case for the missing parts.  They may have fallen off.

The heatsinks are known to fall of by themselves.

BTW,  did you have a back plane?

ISAWHIM
Hero Member
*****
Offline Offline

Activity: 504
Merit: 500



View Profile
May 22, 2014, 01:53:15 PM
 #1544

Ok, latest update... xD

The boards with only one of the four caps/resistors around the A1, with the alternate other caps...
ISAWHIM,

We appreciate your diagnostic efforts thus far, they have been well documented and thorough. Send back to the two non working boards and we'll you send you another 2 working boards plus an additional 2 non working with issues...

So far, there are three that need to be send back, the one which needs the A1 reworked or replaced. (That was one of the "good ones", the later designs. You would not have seen that chips was out of place, unless you got all personal with the chip like I had. It started running fine, until the heat got to it, I assume. So I am not sure if the inductors or other power components were also hurt. They were all super hot.)

The other two, were the boards which have just stopped responding to detection. (The ones with only the one cap at the A1.) Both have obvious screw-marks across that "COUT46" set of terminals, thus, that was "shorted" at one point in the components operation. Though they were not shorted at the moment I was using them, or running them... I was not sure if that had something to do with the issue those boards were having.

If you could tell me what to look for on those... as they were both once operating perfect. One at 208GHs, and one at 198GHs, prior to just stopping. Those, I believe, had the issue where the boards seemed to start mining faster and faster, just before failure. They eventually started hashing around 50-20GHs, which caused me to re-boot, which would normally seem to "reset" the operational state. But at the time of those cards failures, the "last time they did this", re-booting resulted in the cards just failing to reply for operation.

Are those going to need a component swap? (Thus turning them into the other boards?)

As for the other two boards with the blue heat-sinks... I don't recall any "markings" on those. So, those might work, but may eventually have an issue, as I do recall someone posting "Does anyone have any boards with the blue heat-sinks that is still working". So, that leads me to believe, as well as the fact that they were devoid of heat-sinks, that they will have something to challenge me. xD

Though, I will give those a once-over, to see if I can see anything wrong, before mounting to the heat-sinks and plugging them in.

That shorted board, I will pull and inspect tonight.

While the unit is apart again, I will play with the individual card-shrouds. Then too, I guess I will order some thermal-pads. To facilitate some desired cooling, where I feel it could be improved on those operating units. The pads I was looking at, are expensive, at a consumer level. (They are all overpriced. xD) However, these had the best thermal attributes I could find, without containing metal-foils/shims...

They accept BTC, however, at this low BTC value, it is better to hold the BTC and pay with USD. (BTC has started that rise I have been talking about for the last 2 months. The last thing you want to do now, is cash-out BTC. xD)
http://www.frozencpu.com/cat/l2/g8/c487/list/p1/Thermal_Interface-Thermal_Pads_Tape.html

I would love to use the "Ultra extreme" stuff, which is 17.0 W/mK, or the "Extreme" which is 11.0 W/mK, but not at those prices. xD
I will be getting the "Premium" stuff, with is more than acceptable at 6.0 W/mK. That is 6x to 4x better than the crap others sell as "normal". (Normal thermal pads, the kind used and sold without numbers, are usually 0.5 W/mK to 1.5 W/mK. Eg, they are just airless-foam/rubber/silicone that is heat resistant and impregnated with aluminum-oxide. xD.)

The pads are only 0.5mm thick, which is enough to raise the heat-sink off the surface of the PCB, however, I may look at getting the 1.0mm thick stuff, which I believe is similar to the stuff used on the small heat-sinks. The extra lift will break any capacitance happening between the heat-sink and the PCB itself, and even allow some air-flow under the heat-sink, adding another 30% surface-area for direct cooling. (The under side of the heat-sink, and the top exposed side of the PCB itself.)

I will be adding spacers under each screw, to "shim" the board to a matching height, enough to allow slight compression on the pads, without stressing the PCB. (This is not a solution that is as simple as it sounds. xD)

Some will be added to the tops of the inductors, as soon as I get, or make, a heat-sink for them.

AMT, a tip for applying thermal-pads, is to use a "flat roller" or just a bowed-surface...
1: Peel the backing off one side
2: Rest the pad on the curved surface
3: Using the edge of your nail, hold it in place, as you gently press the other edge to the surface to bond to... (Just the edge)
4: Roll the heat-pad down onto the surface while applying firm pressure. (That stops any large air-bubbles from forming.)

Here is the big trick to the next part...
1: Wait one day before applying this in a similar method, to the chip. (That allows the pad to "expand" back to normal.)
2: Rest one edge on the edge of a chip, and angle it downward, applying pressure towards the edge that is already in contact.
3: Use a heat-gun to heat the pad real hot... Below the SMT solder temps, obviously. (This blows-out any bubbles, by heat-evacuation/expansion.)
4: Apply slight pressure to the two taped/bonded items, and allow to cool. (That seals any "gaps", and when cooling, creates a vacuum that will pull the thermal-pad onto the surface for 100% microscopic contact.)

The alternative to that last step, is to have a vacuum chamber with a device that can apply pressure after the air has been evacuated and before the pressure has been returned to normal. That is an expensive unit. The heat trick works better anyways. Makes the bond stronger and pre-treats the unit for "operating conditions".

The issue with doing this by hand... You use your FINGER to press the pad onto the chip. The pad compresses in the center, creating a "dome". Once applied to the chip, that "dome" is now a sealed "bubble" of air and humidity. That decays the bond, stops the heat-pad from making contact where it is needed most, and ends-up in failure similar to the photo a few pages back.

I lied, there is another application method that works better, if the pads can be compressed. It involves pre-compressing them into a reverse-dome, or a bowl-like curve. However, that is best used for machines/robots to do the application, as they can hold the two surfaces flush, and lower them so the "now raised center", touches the chip first. That also stops them from having to wait 24-hours for the pad to "swell back to normal", between both applications. That swelling time is a big issue, after having been compressed.
ISAWHIM
Hero Member
*****
Offline Offline

Activity: 504
Merit: 500



View Profile
May 22, 2014, 01:58:37 PM
 #1545


YES... that is the correct mounting and heat-sink setup I was looking for... xD

Didn't know about that version!

Those should have at-least one mounting to a screw though, with a spacer-obviously. That small thermal-tape will not hold that weight, once heated. I can peel the heat-sinks off by hand. (Have done that to four of them, where they were not seated correctly on the chip. Just got to be real slow about it. xD I am a master of tape removal! That is my only skill.)

That heat-sink covers the "span of the air flow" more, forcing the air through it. (Unlike the copper ones, where 90% of the air flows around it, not through it.)

That is hard to apply, when they hang over the chip like that, and are in two separate pieces. I would have made a spacer with regular double-sided tape, at the least, to support the excess overhang, on both sides, secured to the PCB.

One long one, across both those chips would have been better. But for the other four, an offset mounting would have been preferred, with the under-side support.

I am sure that was just another manufactures "substitutions".
mrpark
Sr. Member
****
Offline Offline

Activity: 364
Merit: 250


View Profile
May 22, 2014, 02:11:23 PM
 #1546

It looks like some where just forgotten about. I cant say 100% for sure, they could have fallen out of the box. It did come with a backplane board, and that looks like it is not damaged.  I wonder if I can get the microchip looking connectors from anywhere? I will also look for the PCI-E connectors. The funny thing about this, those other two boards I think are just decoys, they never hashed from the begging. Again, to refresh you memory, this was supposed to be a full miner, and not a kit. It appears as though someone was trying to assassinate me with no thermal glue on these A1 chips that heat up to who knows how high. The screws on the fan were just way too long, even touching the board. I mean, this thing must have been put together by some homeless people. Just so you know, the case was so damaged, most of the screws popped off, and I mean almost ALL of them. I think these is very good evidence if anyone wanted to show carelessness on AMT's part.

Is there a wiring diagram for this thing? What are the major connections? 1) Power supply to Pi board 2) power to Fans 3) back plane data and power?

As I've said, I have worked as an electrical engineer in the past, but I work more in management now, so I am not as sharp as I used to be when I was troubleshooting boards and such.


Photos as promised...

broken off pin connector: (notice 3 of the boards have one kind of heat sink, the others have a different kind, not sure what's up with that?)

.... (snip)


Why are the heatsinks here different from the heatsinks that we have?  Why so much inconsistency in the build?

BTW,  you may need to look carefully in the case for the missing parts.  They may have fallen off.

The heatsinks are known to fall of by themselves.

BTW,  did you have a back plane?

BTC: 1JDjCGtxtxoZ46XgTqUoXBDxNFKwcsEmik
mrpark
Sr. Member
****
Offline Offline

Activity: 364
Merit: 250


View Profile
May 22, 2014, 02:15:03 PM
 #1547


YES... that is the correct mounting and heat-sink setup I was looking for... xD

Didn't know about that version!

Those should have at-least one mounting to a screw though, with a spacer-obviously. That small thermal-tape will not hold that weight, once heated. I can peel the heat-sinks off by hand. (Have done that to four of them, where they were not seated correctly on the chip. Just got to be real slow about it. xD I am a master of tape removal! That is my only skill.)

That heat-sink covers the "span of the air flow" more, forcing the air through it. (Unlike the copper ones, where 90% of the air flows around it, not through it.)

That is hard to apply, when they hang over the chip like that, and are in two separate pieces. I would have made a spacer with regular double-sided tape, at the least, to support the excess overhang, on both sides, secured to the PCB.

One long one, across both those chips would have been better. But for the other four, an offset mounting would have been preferred, with the under-side support.

I am sure that was just another manufactures "substitutions".

Hi ISAWHIM:

Thank you for the kind words: I was just showing them placed on the board, I have not put them down yet. I don't want to screw them down because I don't want to modify anything if I can avoid it. AMT may ask me to RMA this thing again?, and I will need to send it back in original condition.

BTC: 1JDjCGtxtxoZ46XgTqUoXBDxNFKwcsEmik
mrpark
Sr. Member
****
Offline Offline

Activity: 364
Merit: 250


View Profile
May 22, 2014, 02:27:08 PM
Last edit: May 22, 2014, 02:42:45 PM by mrpark
 #1548

I need some of these: Any ideas where to get them? Two of these were broken off, this is one of the pieces that fell out of the box. Also compositors fell out, sharp pieces of plastic. I guess its call a "Flat female header" ?

http://www.ebay.com/itm/Lots-of-10pcs-Pitch-2-54mm-2-x-10-Pin-Double-Rows-Female-Pin-Header-/160938983646?pt=LH_DefaultDomain_0&hash=item2578b600de



BTC: 1JDjCGtxtxoZ46XgTqUoXBDxNFKwcsEmik
FrictionlessCoin
Legendary
*
Offline Offline

Activity: 854
Merit: 1000


★YoBit.Net★ 1400+ Coins Exchange


View Profile
May 22, 2014, 02:39:41 PM
 #1549

I need some of these: Any ideas where to get them? Two of these were broken off, this is one of the pieces that fell out of the box. Also compositors fell out, sharp pieces of plastic. I guess its call a "Flat female header" ?

http://www.ebay.com/itm/14pcs-2-X-20-Pin-Double-Row-2-54-Break-Away-Female-Header-Wire-Connector-/371048802872?pt=LH_DefaultDomain_0&hash=item56643b6a38




See... I told you it was in the box somewhere.

Same thing happened to me... lots of busted parts all over the place.  You just had to look for them.

However,  till this day... they never shipped me a backplane.   

mrpark
Sr. Member
****
Offline Offline

Activity: 364
Merit: 250


View Profile
May 22, 2014, 02:45:47 PM
Last edit: May 22, 2014, 02:59:44 PM by mrpark
 #1550

actually that was a double side 20 pin, I posted the one for double side 10 pin. While some of the parts were in the box, some were not. Even though its in the box, its unusable. I bought some of these 10 pin, let me know if anyone needs them. They come from China, so might be a few... Now to find that female PCI-E board connector. Any ideas?

I need to find this one now... I guess it is called a "6-Pin Graphics Card PCIe Male Header Connector"


BTC: 1JDjCGtxtxoZ46XgTqUoXBDxNFKwcsEmik
NotFuzzyWarm
Legendary
*
Offline Offline

Activity: 1932
Merit: 1327


Evil beware: We have waffles!


View Profile
May 22, 2014, 03:01:54 PM
Last edit: May 22, 2014, 03:13:28 PM by NotFuzzyWarm
 #1551

I need some of these: Any ideas where to get them? Two of these were broken off, this is one of the pieces that fell out of the box. Also compositors fell out, sharp pieces of plastic. I guess its call a "Flat female header" ?

http://www.ebay.com/itm/Lots-of-10pcs-Pitch-2-54mm-2-x-10-Pin-Double-Rows-Female-Pin-Header-/160938983646?pt=LH_DefaultDomain_0&hash=item2578b600de



First, when it comes to electrical components, do NOT rely so much on e-bay. Yeah might be fine for generic connectors but being solidly in the grey-market may not be either. The assembly house AMT used getting parts from the grey market is part of what has caused so many problems. Unless one is a hobbyist, getting components from non-vetted distributors is verboten! You could be getting anything that comes in a similar looking package that has been re-labeled.

The PCie connector is Molex Mini-fit Jr. series Molex pn-45732-0001. Fora right angle one, straight should be close to that. Mouser pn http://www.mouser.com/ProductDetail/Molex/45732-0001/?qs=sGAEpiMZZMs%252bGHln7q6pm0yYNlAgQWB%2fw9pdNaIn3qc%3d

For the coms header just measure the pin-to-pin spacing on it and how many pins there are.

For bitcoin to succeed the community must police itself - Joshua Zipkin aka Joshua Alexander leaked AMT A1 miner skype chats
How a miner mfgr SHOULD operate: HaggsFIN trip to Canaan My info useful? Donations welcome! 1Fuzzyk398kDWVjuC5qPX5v6CjSkvbgAbd
-Support Sidehacks miner development. Donations to:   1BURGERAXHH6Yi6LRybRJK7ybEm5m5HwTr
opieum2
Hero Member
*****
Offline Offline

Activity: 532
Merit: 500


View Profile
May 22, 2014, 03:13:26 PM
 #1552

I don't get it,  why is the program called I2Cset when the chipset doesn't use I2C but rather SPI?

I believe there is an i2c to SPI bridge, which would explain the need for both and why they are both detected.
I am still understanding those aspects a bit more as I am finding I need the information to make potentially bring the boards back to life.

http://www.i2cchip.com/mix_spi_i2c.html is one resource (of many I found)

The good news is that it might be possible to bring the cards back provided I can find the right command to bring them back. The AMT-setup dpot command (in my previous post) do not appear to work at all as they are coming up with an invalid value. That just means it doesn't work. I am reading through the AMT script to understand the issue.

As you can see in the above I get [OK] Error: Data value invalid! is a sign that the code might not be correct. OR something else is wrong. Hopefully AMT can clear this particular piece up.
The card on the chain is found so that's positive. BUT the command to set it is invalid (bad). I know at this point I have 7 cards that are potentially zombiefiable (new word to add to webster dictionary), I have 3 working cards (not messing with that) and 1 totally dead card that will prevent any sort of power up (need to check it for shorts)

On the hardware side I tried something based on ISAWHIM's mention of the potential board short from the PCB touching the metal cage. I taped up the edges where the cards contact the metal I had hoped on a long shot that might actually bring up a card or two, but no dice Sad

That said least now that is a potential issue that is mitigated. I think the boards COULD be brought back with software. Maybe not all but some of them. The fact that the i2cset commands was not actually taking because the value was invalid would be a good reason why this happened. I am currently testing now with AMT's default firmware before I move on to the porting process (which I have partially done). I figure it might be best doing it with what is already available before reinventing the wheel. If I get the cards back up then I will proceed with the next steps of porting over to the new firmware.

I think there is progress tho seeing as the cards are detected but nonfucntional at this point. Hopefully not beyond the point of recovery.

EDIT: To add I am working on the i2cset commands to do all this manually if I have to (not something I recommend to anyone not experienced in this sort of thing), and seeing how the script interacts with that. If I can manually do this bypassing the script I will modify the existing scripts to issue the correct commands and add comments in the scripts for proper use. Its fairly trivial to add the usage part.  That might actually get the hardware working. While the progress is slow (in my opinion) its progress that's happening. My time is limited so I really only put in a couple hours a day on this at most.

Opieum,

The AMT script is design to run with the UI only. So basically if you log in directly and using the:

killall cgminer

amt-setup

(what ever else)

and you run from the command line - you'll be using default bitmine values. The script itself starts with the automatically when the miner first starts, killing it or logging in directly via the pi null and void this script. You can refun this, but the best way to see if its working/having any change is to hook up the second monitor to the pi and also have your computer handing, work on the pi itself, use "reboot" to reboot the miner itself, then do not login to the pi but to the admin panel on the browser and run the miner that way. You can always check the activity in the settings section of the UI as well.

We can ask Zefir to help us with this one. He is a nice man who we know face to face, he's a bit down/sad about the industry. Although he is from the squad of optimists turned pessimist, he may still be willing to help. Zefir is we looked up to since before we knew zefir, and when meeting him in person for the first time in swiss, it was kind of like a celebrity shock type of thing. He's someone that really wanted to bitcoin to work as intended rather than how it exists today. We'll see if he'd be willing to come over to our thread and assist thus far.

Here is a thought that we've had for sometime but could never verify it because as brilliant as our software guys are, they are just arrogant. And although they've confirmed this isn't a problem on several occasions, we still feel that it might be a good place to look. Our software guys like to create rather than verify someone's else work.

The A1 disables itself and the entire chain if it reaches the point of overheating (or when it thinks it's overheating), we've always had a gut feeling that chip parameters later set are somehow remembered physically and are not being told to "reactivate later" hence getting the "no a1 chip chain detected".




The "feeling" you guys have on that appears to be spot on. The boards or the backplane are not being told its ok to power back up. Something is causing it to stay in a failed state. Due to the fact that I would see cards drop and then later come back up after a reboot it made me wonder a bit about that. As I am now down to 2 working cards, I am focusing my work on the totally disabled miner to see if I can get that working. But I am lacking alot of info mostly due to my own lack of knowledge of the driver and hardware as a whole. Its been a fair bit of guesswork and research. There is some initialization process that is not properly being executed. If the scan time could be lowered OR executed in a way that checks each link in the chain individually (which means slower BUT more stable power up of the cards/chips) it would be the place to start.

"amtminers scam joshua zipkin scammer"
-Joshua Zipkin leaked skype chats http://bit.ly/1s7U2Yb
-For bitcoin to succeed the community must police itself.
NotFuzzyWarm
Legendary
*
Offline Offline

Activity: 1932
Merit: 1327


Evil beware: We have waffles!


View Profile
May 22, 2014, 03:28:37 PM
 #1553

For designers, Aligent has a new measurement Fundamentals webcast & Apps notes DVD out. https://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=2381549&id=2381549&cmpid=MA44870AM&MKCID=18254238

For bitcoin to succeed the community must police itself - Joshua Zipkin aka Joshua Alexander leaked AMT A1 miner skype chats
How a miner mfgr SHOULD operate: HaggsFIN trip to Canaan My info useful? Donations welcome! 1Fuzzyk398kDWVjuC5qPX5v6CjSkvbgAbd
-Support Sidehacks miner development. Donations to:   1BURGERAXHH6Yi6LRybRJK7ybEm5m5HwTr
mrpark
Sr. Member
****
Offline Offline

Activity: 364
Merit: 250


View Profile
May 22, 2014, 03:37:09 PM
 #1554

Thanks NotFuzzywarm, I ordered two of them. I guess they did not used angled ones to begin with because it would hit the case.  On the 10 pin double row, it's going to be a challenge to plug that board back in and have it all line up.

BTC: 1JDjCGtxtxoZ46XgTqUoXBDxNFKwcsEmik
freddyfarnsworth
Full Member
***
Offline Offline

Activity: 238
Merit: 100


View Profile
May 22, 2014, 07:43:37 PM
 #1555

Here is some info on signalling issues you guys working on these are up against.

https://bitcointalk.org/index.php?topic=294197.msg6873065#msg6873065   

Quote from: blackarrow on May 21, 2014, 08:01:38 PM
"The issue is the SPI communication fails on the current backplane as soon as you attach more than 1 hashboard."

When problem solving, sometimes it helps to start with the basics:
http://www.embedded.com/electronics-blogs/beginner-s-corner/4023908/Introduction-to-Serial-Peripheral-Interface

Quote from: tempestb on Today at 02:08:55 PM
I wonder if by adding the second connection to the SPI if it is losing power?  It should be in third state I imagine.  Difficult to make any guesses without having any idea of what is going on. 

To put a finer point on it, from the article above (beginner's guide to SPI lol):

Quote
SPI's full duplex communication capability and data rates (ranging up to several megabits per second) make it, in most cases, extremely simple and efficient for single master, single slave applications. On the other hand, it can be troublesome to implement for more than one slave, due to its lack of built-in addressing; and the complexity only grows as the number of slaves increases.

Quote from: fivejonnyfive on Today at 02:54:28 AM

I'll say it again:
Replace The Backplane with Wires
As in 1 (one) SPI channel PER HASHBOARD going direct to a controller. I hope to god you didn't have all the hashboards sharing a common bus on the backplane before - because I can't imagine trying to get that to work.

Quote from: blackarrow on Today at 07:07:46 AM
Quote from: fivejonnyfive on Today at 02:54:28 AM

I'll say it again:
Replace The Backplane with Wires
As in 1 (one) SPI channel PER HASHBOARD going direct to a controller. I hope to god you didn't have all the hashboards sharing a common bus on the backplane before - because I can't imagine trying to get that to work.

Dear Jonny,

The PCIE connector has 98 pins and there are 5 connectors in each case.

If you still believe that this can be done even on one X3 (not to talk about mass production) you are welcome to come to our office and try to do this yourself.

Regards!

Quote from: Vcore:
Surely most of those 96 pins are power and the backplane doesn't have any problem with power does it?
So why not cutting the spi traces and running wires only for the spi?
It might not be elegant but depending on the pcb it might be workable (ie, desoldering something and putting an spi only connector glued to the board) or at least it works to test it.

....

BTC: 1F1X9dN2PRortYaDkq89YJDbQ72i3F5N3h MEOW: KAbvy9jrrajvN5WLo7RWBsYqYfJKyN9WLf DOGE: DAyKSrTiVeRZaReTu1Cyf5Je6qPdKTuKKE
opieum2
Hero Member
*****
Offline Offline

Activity: 532
Merit: 500


View Profile
May 22, 2014, 08:12:46 PM
 #1556

Freddy this is a great post and highly informative!!!

Alternately, since the hashboards are already using SPI, COULD it be possible to run them along from the USB port using a USB to SPI bridge? This seems like a possibility. And might solve a number of engineering problems. That said it would require some adapter for existing boards to be able to run. This would be a MUCH cheaper solution than reingineering the boards, altho that is already necessary due to the PCB related issues and other faults on the boards themselves. But from the electrical/signal and software standpoint a USB setup not unlike the way technobit does things might be an excellent way to address this. 

Just spitballing on this. There could be alot of things wrong with my suggestion. I am just putting it out there. I will study up more as I would be interested in a quick retrofit to get this working with USB vs doing all this extra work.

"amtminers scam joshua zipkin scammer"
-Joshua Zipkin leaked skype chats http://bit.ly/1s7U2Yb
-For bitcoin to succeed the community must police itself.
freddyfarnsworth
Full Member
***
Offline Offline

Activity: 238
Merit: 100


View Profile
May 22, 2014, 08:58:56 PM
 #1557

Thanks, I am not spi usb programmer, Just want to help out, and keep you barking up the right tree Smiley
Seems most industry devs already know most of this, It just hit the hobby level and I found it relevant to some of you guys diagnostic posts. AMT must also know this.

Seems all three have similar issues HF BA AMT. Would that boil down to Bitmine original design ?

1 SPI for each master slave.... Is correct way.

BTC: 1F1X9dN2PRortYaDkq89YJDbQ72i3F5N3h MEOW: KAbvy9jrrajvN5WLo7RWBsYqYfJKyN9WLf DOGE: DAyKSrTiVeRZaReTu1Cyf5Je6qPdKTuKKE
AMT_miners
Sr. Member
****
Offline Offline

Activity: 392
Merit: 250


View Profile
May 22, 2014, 09:50:50 PM
 #1558

Thanks, I am not spi usb programmer, Just want to help out, and keep you barking up the right tree Smiley
Seems most industry devs already know most of this, It just hit the hobby level and I found it relevant to some of you guys diagnostic posts. AMT must also know this.

Seems all three have similar issues HF BA AMT. Would that boil down to Bitmine original design ?

1 SPI for each master slave.... Is correct way.

Yes its a good point and most of have gone through that one, but in this we don't think it's the SPI. Zefir outlined this in his thread:

https://bitcointalk.org/index.php?topic=294235.0

The chip resets itself when:

A. when supply voltage is unstable,
B. chip gets too hot,
C. there is noise on SPI bus and communication gets messed up.

Once that happens, we have not found a way to get the chain back to life other then possibly issuing a HW reset which is covered in zefir's thread.
mdiggy
Member
**
Offline Offline

Activity: 102
Merit: 10


View Profile
May 22, 2014, 10:12:11 PM
 #1559

actually that was a double side 20 pin, I posted the one for double side 10 pin. While some of the parts were in the box, some were not. Even though its in the box, its unusable. I bought some of these 10 pin, let me know if anyone needs them. They come from China, so might be a few... Now to find that female PCI-E board connector. Any ideas?

I need to find this one now... I guess it is called a "6-Pin Graphics Card PCIe Male Header Connector"


If youre looking for a 6 to 8 pin connector, I have some laying around from GPUs I bought (for mining of course). I could either send you one if thats what you're looking for, or id imagine they're easy to find at a local electronics DIY place.
Let me know!
NotFuzzyWarm
Legendary
*
Offline Offline

Activity: 1932
Merit: 1327


Evil beware: We have waffles!


View Profile
May 22, 2014, 11:14:49 PM
Last edit: May 23, 2014, 12:20:42 AM by NotFuzzyWarm
 #1560

Thanks, I am not spi usb programmer, Just want to help out, and keep you barking up the right tree Smiley
Seems most industry devs already know most of this, It just hit the hobby level and I found it relevant to some of you guys diagnostic posts. AMT must also know this.

Seems all three have similar issues HF BA AMT. Would that boil down to Bitmine original design ?

1 SPI for each master slave.... Is correct way.

Yes its a good point and most of have gone through that one, but in this we don't think it's the SPI. Zefir outlined this in his thread:

https://bitcointalk.org/index.php?topic=294235.0

The chip resets itself when:

A. when supply voltage is unstable,
B. chip gets too hot,
C. there is noise on SPI bus and communication gets messed up.

Once that happens, we have not found a way to get the chain back to life other then possibly issuing a HW reset which is covered in zefir's thread.


Along the SPI com lines... I've been wondering how that is being handled: The the SPI link that ISA posted a while back It says it *does* kinda allow addressing. 4 lines are available for that as I recall. Now as to if each addressed chip can be a master to more in a serial chain from it - dunna know. Gotta dig more.

Now even more relevant, from Zefir's thread: "I understood that the eval board used in China (the one you saw in the pictures) for testing has a level shifter for input and output signals, while Bitmine's boards use resistors to lower the input signals and a level shifter for the output signal (MISO) - seem to work both."

2 things on that: Just where is that level shifting done? and, Is it still being done with a resistor divider network?

Far far better to use a level shifter chip for both data out AND data in! By its nature using a resistor network makes the output signal from the divider highly dependent on what is fed in on the hi side and sensitive to the load presented by the A1. If either changes for any reason so does that signal level...

A level shifter chip not only changes the voltage but also acts as a Schmidt trigger (very tolerant to input signal level changes) giving very predictable switching as well as being a decent current source that will always deliver the specified output voltage regardless of load (within chip specs of course). End result is very clean data signals.

And also per that thread (and should always be done in ANY logic design until proven unneeded) Is the clock signal to each A1 buffered?

For bitcoin to succeed the community must police itself - Joshua Zipkin aka Joshua Alexander leaked AMT A1 miner skype chats
How a miner mfgr SHOULD operate: HaggsFIN trip to Canaan My info useful? Donations welcome! 1Fuzzyk398kDWVjuC5qPX5v6CjSkvbgAbd
-Support Sidehacks miner development. Donations to:   1BURGERAXHH6Yi6LRybRJK7ybEm5m5HwTr
Pages: « 1 ... 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 [78] 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 ... 160 »
  Print  
 
Jump to:  

Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!