[ANN] Bitfury is looking for alpha-testers of first chips! FREE MONEY HERE! - page 28.

ssi

member

Activity: 70

Merit: 10

Quote from: needbmw on August 25, 2013, 11:59:07 PM

Quote from: ssi on August 25, 2013, 11:02:29 PM

and changing the clock doesn't seem to change the average hashrate much. My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

Interesting, thanks.

So, just to make sure I understand you, when I download the legkodymov cgminer fork, it's set like so by default:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x7F, 0x00, 0x00 };

So you're saying that one setting faster is:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x00 };

and one setting faster than that is:

Code:

unsigned char osc6[8] = { 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x01, 0x00 };

Is that correct?

needbmw

legendary

Activity: 1302

Merit: 1008

Quote from: ssi on August 25, 2013, 11:02:29 PM

and changing the clock doesn't seem to change the average hashrate much. My clock is currently at ... 0xFF 0x10

If using cgminer please note reverse clock bits order, so you have to set 0xff 0x01, 0xff 0x03, ... , 0xff 0x3f, 0xff 0x7f, otherwise your clock will be 8-bits aligned.

ssi

member

Activity: 70

Merit: 10

I'm up and running... seeing 6GH on four chips. Running .77V VDD, and changing the clock doesn't seem to change the average hashrate much. My clock is currently at ... 0xFF 0x10, and I'm bumping it up slowly. Haven't seen any hardware errors yet. Not sure when/if I should try to move up to .9V.

LaserHorse

full member

Activity: 140

Merit: 100

Quote from: zulunation on August 23, 2013, 06:15:57 PM

Hello everyone,
I'm developing a board for 16 chips and having some problems with power supply.
It is the same like cscape used in his board.
Here is the front view of the power supply:

https://dl.dropboxusercontent.com/u/26351037/bitfury_ps_pinheight.JPG

I see that the pin height here is 1.57mm. Is it enough to solder it on the board?
As is understand there must be pads on the board and the ps is soldered into them and pin goes through the whole board and soldered at the other side of the board.
It seems to me that 1.57mm is not sufficient.

These are surface-mount - pins are not intended to go through the board.

In a rush to get my chips hashing, I soldered one of these to a through-hole board. You need to be careful, but it's possible.

I used 2 pieces of solid core wire wrapped around on each 'pin', routed them under the board and soldered them to massive tracks on the underside. I considered soldering directly to pads on an upside-down protoboard, but the spacing does not match up, etc.

some notes:
• Per the datasheet - solder the on/off pin to ground if you won't be using it.
• The smaller 'pins' on the board easily come loose while soldering extension wire to them
• Flux is your friend
• Needless to say - triple-check for bridges *everywhere* before applying power

I've got the board powering 8 chips currently with moderate airflow - temp is good, power is good.

zulunation

sr. member

Activity: 335

Merit: 250

Hello everyone,
I'm developing a board for 16 chips and having some problems with power supply.
It is the same like cscape used in his board.
Here is the front view of the power supply:

I see that the pin height here is 1.57mm. Is it enough to solder it on the board?
As is understand there must be pads on the board and the ps is soldered into them and pin goes through the whole board and soldered at the other side of the board.
It seems to me that 1.57mm is not sufficient.

klondike_bar

legendary

Activity: 2128

Merit: 1005

ASIC Wannabe

Quote from: cscape on August 23, 2013, 03:23:47 PM

Quote from: Gomeler on August 23, 2013, 03:03:25 PM

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

Yes, but the voltages are 3.3 and 1.8.

if you are feeling creative, i bet you could grab the 3.3V line directly from the PSU. 1.8V is another challenge though

Gomeler

hero member

Activity: 697

Merit: 500

Quote from: cscape on August 23, 2013, 03:23:47 PM

Quote from: Gomeler on August 23, 2013, 03:03:25 PM

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

Yes, but the voltages are 3.3 and 1.8.

Thanks for the confirmation. Maybe one day I'll get around reading up on PCB design and tinker with this.

cscape

sr. member

Activity: 251

Merit: 250

Quote from: Gomeler on August 23, 2013, 03:03:25 PM

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

Yes, but the voltages are 3.3 and 1.8.

Gomeler

hero member

Activity: 697

Merit: 500

Is the level-shifter required to transform the voltage coming out of the GPIO pins of the raspberry pi(5v?) down to levels that the bitfury chips operate on(3.3v?)?

ssi

member

Activity: 70

Merit: 10

Quote from: LaserHorse on August 23, 2013, 12:04:59 AM

Quote from: ssi on August 22, 2013, 10:25:02 PM

https://pbs.twimg.com/media/BSUpPk0CIAA2YGU.jpg

I lack a level shifter. Should be here tomorrow.

what you lack in level shifters, you make up for in density - nice work!

how do foresee mounting it?

Honestly I didn't give a lot of thought to mounting when trying to get these prototypes spun quickly

Next rev will have more options. For this one, I can just 3d print something with slots.

However, what I DID think about:

it supports RPi:

AND Beaglebone:

LaserHorse

full member

Activity: 140

Merit: 100

Quote from: ssi on August 22, 2013, 10:25:02 PM

https://pbs.twimg.com/media/BSUpPk0CIAA2YGU.jpg

I lack a level shifter. Should be here tomorrow.

what you lack in level shifters, you make up for in density - nice work!

how do foresee mounting it?

ssi

member

Activity: 70

Merit: 10

I lack a level shifter. Should be here tomorrow.

ssi

member

Activity: 70

Merit: 10

Quote from: kevinm on August 21, 2013, 09:08:40 PM

Great work fellas,
Am looking forward to seeing Introns test results and ssi's completed prototype.
Looks like that baby will need a heatsink, you got one in mind ssi?

cheers,
kev

I don't expect it to need a heatsink... the whole board should be ~25W or so, and there's four layers of copper pours to suck the heat out of the chips.

kevinm

full member

Activity: 228

Merit: 100

This is not good for my Chi... Yifu

Great work fellas,
Am looking forward to seeing Introns test results and ssi's completed prototype.
Looks like that baby will need a heatsink, you got one in mind ssi?

cheers,
kev

ssi

member

Activity: 70

Merit: 10

Quote from: intron on August 21, 2013, 03:59:18 PM

Ok. Think you can do without most of these:)

intron

I figured that would be the case, but with the timeline being so critical, I wanted to leave myself some "outs".

intron

sr. member

Activity: 427

Merit: 251

- electronics design|embedded software|verilog -

Quote from: ssi on August 21, 2013, 03:13:11 PM

Quote from: intron on August 21, 2013, 03:12:12 PM

Quote from: ssi on August 21, 2013, 03:04:59 PM

Quote from: intron on August 21, 2013, 03:03:21 PM

Using "strings" here?

intron

You mean the SPI chaining? Yes it's one long string of ten chips.

I mean that you can power the ASICs with a voltage that
is N times Vcore of a single ASIC. Whith all bitfurys in series:
ground tab of one ASIC connected to the Vcore pins of the
next, etc.

intron

Oh... no, we have a regulator that'll do .6 to .9V at up to 40A, and the chips are all paralleled.

The reason there's so many passives is I couldn't get any clarity on what we might need for pullup
or pulldown, so the option is there for either on all the SPI lines. Plus there are series terminators.

Ok. Think you can do without most of these:)

intron

ssi

member

Activity: 70

Merit: 10

Quote from: intron on August 21, 2013, 03:12:12 PM

Quote from: ssi on August 21, 2013, 03:04:59 PM

Quote from: intron on August 21, 2013, 03:03:21 PM

Using "strings" here?

intron

You mean the SPI chaining? Yes it's one long string of ten chips.

I mean that you can power the ASICs with a voltage that
is N times Vcore of a single ASIC. Whith all bitfurys in series:
ground tab of one ASIC connected to the Vcore pins of the
next, etc.

intron

Oh... no, we have a regulator that'll do .6 to .9V at up to 40A, and the chips are all paralleled.

The reason there's so many passives is I couldn't get any clarity on what we might need for pullup or pulldown, so the option is there for either on all the SPI lines. Plus there are series terminators.

intron

sr. member

Activity: 427

Merit: 251

- electronics design|embedded software|verilog -

Quote from: ssi on August 21, 2013, 03:04:59 PM

Quote from: intron on August 21, 2013, 03:03:21 PM

Using "strings" here?

intron

You mean the SPI chaining? Yes it's one long string of ten chips.

I mean that you can power the ASICs with a voltage that
is N times Vcore of a single ASIC. Whith all bitfurys in series:
ground tab of one ASIC connected to the Vcore pins of the
next, etc.

I'm asking because there seems a lot of passives between
the ASICs.

intron

ssi

member

Activity: 70

Merit: 10

Quote from: intron on August 21, 2013, 03:03:21 PM

Using "strings" here?

intron

You mean the SPI chaining? Yes it's one long string of ten chips.

intron

sr. member

Activity: 427

Merit: 251

- electronics design|embedded software|verilog -

Quote from: ssi on August 21, 2013, 02:58:43 PM

Using "strings" here?

intron

Topic: [ANN] Bitfury is looking for alpha-testers of first chips! FREE MONEY HERE! - page 28. (Read 176728 times)