Topic: [ANN] Bitfury is looking for alpha-testers of first chips! FREE MONEY HERE! - page 6. (Read 176742 times)

Presuming that upon power-up the chip is not in "pass-through" mode your statement would be correct - as long as you do not ever put the chip in that mode it should work. However if the chip somehow starts in that mode ... or if you ever put it in it (even if just once) then there is no other way to get it out of it - you either have to power it down and hope that it will self-reset, or you have to have the ability to send the RESET sequence.

@vs3 just checked the chip documentation ... here you state that the main issue is the reset sequence, but for a single chip you shouldn't need to reset the chain, because there is no chain ... few NOPs and a brake should do same. In the worst case if the chip expects data you will need ~80 NOPs, so on start you may send 80, but once the communication is started it should not need resetting any more.

None built-in as far as I know. At least in Gen-1 chips. Hopefully in the next generation such thing is added ... or at the very least just a diode on the die that can accessed externally and could easily be turned into a thermometer (sort of, not the most accurate but good enough for our purposes).

My only guess as to why there isn't one in its current incarnation - those chips just LOVE heat! From what I've seen - all other chips will stop working, burn, melt, explode, fall apart, etc., you name it, before bitfury's stop working

Oh - and my favorite quote:

... any software accessible temperature sensors ......... ?

www.dropbox.com/s/zogwlms8b8h0wpo/1%20layer%20m-board%2016%20v2.3.jpg
www.dropbox.com/s/cvse0gp2opfdoa4/1%20layer%20m-board%208%20v2.3.jpg
www.dropbox.com/s/6t5sddm7trc7ja7/1%20layer%20m-board%204%20v2.3.jpg

Hello everybody, i know the beta tests are over on BitFury so please dont jump me on that.
Since i dont know exactly where to share this on the forum, i chose here.
First id like to thank intron and for the help provided and a bunch of h-cards for testing.
The schematic is taken from here:
www.github.com/nanofury/NanoFury/wiki/The-missing-bitfury-chip-documentation

What you'll find in those links are 3 m-boards, layout and schematic for 4, 8, 16 H-cards.
Since they are hard to find right now i thought id make this.

For techinical details there isnt much, the connectors are plain old PCI-ISA cut every 10+1 contacts... with a paper cutter.
Resistors are plain old 1/4W ones 18ohms, 56ohms ... i hope ill find some.
The only expensive parts are the level shifters TXB0104D (4 pcs... about 5 euros)
LM 7805 ...rather cheap
LM 2596 power module ... again 4 or 5 euros.
2x1m one layer pcb a ... about 30 euros (but you can make 40 or so m-boards from 1 pcs... so again, rather cheap).
The jump wires can go on the bottom of the card(no dangerous contacts there) or on top but they'll have to be isolated.

Critics, ideeas, parts, donations are welcome, im going to build this for real and ill update on the work.
The drawing is 1:1 scale, so if you print it, put it on a 1 layer pcb you can start cutting traces, i still use tape + feric chloride to do my pcb's

I have everything except the level shifters which im going to order and the H-cards which are on their way.

On a side note, i dont know if the schematic or layout are correct so if you decide to build, please check and tell if something is wrong.

Enjoy

I've heard February, but I'm still holding out hope that someone has a bunch they'll be willing to sell in the meantime.  It'd be nice to make up my own miner rather than paying these ridiculous markups.

Anyone have any word on when more chips might be available, or they gone for a while, I guess I am glad I did not wait any longer than I did to get my 4.

Thanks for the tip man.

I'm pretty sure it is essential! The rpi's I/O use 3.3v, the Bitfury chips will not like that (i.e. die horribly), as they're meant for 1.8v I/O...

I would also be very wary of power-sequencing issues if you are not powering your rpi from your rig - I had a try at using opto-isolators to separate the rpi and miner power domains, but my optos were way too slow for the SPI. Now my board supplies the pi with 5v directly.

Guys, is a level converter essential or is it optional when connecting to  a rPI? Thanks

Can you beat 4 dead chips in a row ?
And that too starting from Chip#1

https://bitcointalk.org/index.php?topic=228677.msg3012226;topicseen#msg3012226



But we never had the time to finish it completely.

intron

I'm loving that! 

You're also beating my 8-chip setup which seems to peak at 9GH/s and then rapidly dip to ~3 on chainminer. For some reason, I get 4,5 or 7 chips detected, depending on the direction the wind's blowing

If I thought I could get my hands on some more chips, I'd be tempted to respin my board with a way to take dodgy chips out of the SPI chain...

Did anyone ever get a QFN ZIF-socket and build a chip tester?

Cheers c-scape, that makes sense!

Just wanted to thank everyone for the help, this is what the rig looks like. 8.5 GH/s

Well I mounted 3 chips on my quad board, running cgminer it finds them but only the first chip generates and accepted units, they all are getting a bad HW errors.
   

I am pretty sure the chips are not bad , I have checked the connections on the board and slowed the spi down to try and eliminate that, it is drawing 4.3 A @ 1.3 Gh/s for 3 chips so I think there all drawing current.
Thanks for any thoughts


Edit:
May have figured it out setting MAXBANKS from 1 -> 2 might have been the issue since the !OE on my level shifter is tied low I guess it saw the chips twice. now setting it back seems to be good.
Thanks

My guess is that the MISO pin keeps showing the last bit that you read (or the one after that). In the case of normal miner operation, that would be the register that indicates the job it is working on, which toggles between 1 and 0 every 2^22 clock cycles.

Thanks - I downloaded the rpi image and had a play with spitest-D

Has anyone else seen their chips go into a mode where the in_miso pin (i.e. the output of the chain back to the rpi) starts oscillating very slowly - e.g. about 1Hz, even when the SPI clock isn't running?

Well in the midst of not knowing what I was doing I put the v2 image from this and it is the home directory , you could mount it and it is in the /home/pi/spic1 dir
https://bitcointalksearch.org/topic/guide-bitfury-miner-supporttuning-287590

I used kpartx -a image-file
mounts them to a loopback then you can mount /dev/mapper/loop0p2 to a dir and get it
of course this is all guessing your using linux

Where did you get that version of spitest from that takes command-line options? Can you post a link please?