Author

Topic: Klondike - 16 chip ASIC Open Source Board - Preliminary - page 168. (Read 435369 times)

full member
Activity: 378
Merit: 100
I've found a rather bigger difference in pricing between Mouser and Digikey, with Mouser always being better priced. I found a couple of alternates in the BOM with a better price. Ordering 500 to several thousand for a large quantity build can make a big difference. I will forward what I have to you.
hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
It also just occurred to me that the chaining connectors don't need to be right at the corner - they could be in a bit as long as the same on all sides. This gets them away from the corner posts which is a bit tight now.

Can use this for FAN connector horizontally.


And this for I2C along with a female header with pins tied, or a tiny jumper pcb.


This allows also for a tiny pcb with I2C repeater.
member
Activity: 114
Merit: 10
interested in assembled boards, however I don't have chips.

I'm in Australia
legendary
Activity: 2126
Merit: 1001
Thank you, both of you! :-)

Ente
newbie
Activity: 35
Merit: 0

Does anyone know if there are SMT "three pin headers" available?
These (and the PCIe-power connector) are the only through-hole parts, which I will try to avoid (to put the board on one big heatsink). If not, I'll use wires to daisy-chain, or "SMT-bent" pins or something.

Ente

Yes of course:
http://www.proto-advantage.com/store/product_info.php?products_id=1900013
The problem is only that they cannot be put so close on the edge that you can jumper between the boards like designed.

But then again, I think some of the pcb manufacturers does not allow you to get copper that far out, due to panelizing problems? (Vscoring destroys the copper or something)
I guess some very short wires for chaining would be better?
hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
Does anyone know if there are SMT "three pin headers" available?
These (and the PCIe-power connector) are the only through-hole parts, which I will try to avoid (to put the board on one big heatsink). If not, I'll use wires to daisy-chain, or "SMT-bent" pins or something.

Ente
Still thinking on that issue. I may put rectangular pads that allow soldering the pin headers horizontally, sticking outwards similar to the USB. There is SMT pin headers but they tend usually to be finer pitch. This 3 pin one left on the board is for the fan. So you could cut fan leads and solder them on but I'm still preferring some better approach.

The PCIe is more of a problem. I'm not aware of any SMT connector for that. But I will likely add pads behind it for direct soldering or whatever else I can find that is SMT. For this it may need a hole drilled in the heat sink, not all the way thru even, but enough for clearance.
legendary
Activity: 2126
Merit: 1001


Does anyone know if there are SMT "three pin headers" available?
These (and the PCIe-power connector) are the only through-hole parts, which I will try to avoid (to put the board on one big heatsink). If not, I'll use wires to daisy-chain, or "SMT-bent" pins or something.

Ente
hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
Is it just a simple standard value standard package SMT resistor?  if so you should be able to find another manufacturer version that digikey or mouser have digi-reels or mouser reels (i forget what they're called) where they sell you a partial size reel of the part for 7$ fixed part + higher per unit cost - this makes it orders of magnitude easier to pick & place.
Yes, standard. Some of them need to be 1% but I tagged them all as 1% since it looked like they were all the same price. IDK, used to be 1% cost more. I think all are 0603.

I have a Project Manager project on mouser.com with all the parts in it. Apparently this can be shared but when I checked they require an email to send it to who you will share with. If someone wants that shared too them then PM me your email and I'll plug it in. and it can be accessed at this url:

http://www.mouser.com/ProjectManager/ProjectDetail.aspx?AccessID=b645e5b812

It may be inaccurate now as I haven't updated it in at least a few days. I find mouser's web site much easier to work with. Digikey is ok but looks like late 90s style. They have one nice thing missing from mouser - you can sort based on lowest price at a given qty rather than just qty 1. That's really useful because who cares what qty 1 costs when you want to price qty 100 etc.

What makes the project manager useful is that you can enter qty for your project and then tell it you want to make 500 qty and it will extend them out with price breaks.

@roybitcoin

Just checked my power supply test board and realized I haven't fixed the layout. I copied the big board and deleted all the unrelated stuff. Now need to put the pours ("zones" in kicad talk) back in. So let me know - either I can throw it up now, or wait for me to fix it up and put up.

I've been on the software project for a couple days as I wanted to make sure everything is in order there before ordering chips to work with. eg. no issues that would prevent this PIC model from working out.




sr. member
Activity: 322
Merit: 250
Mouser just changed their resistor pricing. When I started a couple weeks back they had small qty available but now they'll only sell 10k rolls. Blah.
Is it just a simple standard value standard package SMT resistor?  if so you should be able to find another manufacturer version that digikey or mouser have digi-reels or mouser reels (i forget what they're called) where they sell you a partial size reel of the part for 7$ fixed part + higher per unit cost - this makes it orders of magnitude easier to pick & place.
hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
For your enjoyment.

Still missing some parts - power inductors, PCIe conn. I've put in substitute inductors for now but they don't really look like these ones.

hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
Any idea of lead time for parts? Is it time to order parts for power supply?

I´m looking at mouser and digikey in order to purchase some of these parts. They deliver really fast when they have stock.

Bkk do you have a preliminary PCB design for the power supply? I´m interested to make some reliability testing of this DC-DC converter  (full load and temperature).
I did already copy this onto another small board that I was going to have made right away, as 2 layer for testing. I'll put it on my github so you can get it if you like. It's not the best because it has 4 layers really and a GND pad under the feedback/analog area on inner layer. Also, I think it still needs some tweaking because the GND/AGND joint should be close to the chip and only once, and I haven't checked that carefully yet.

I'll be placing an order very soon as well, along with a new PICKit 3 (as mine is PICKit 2 and doesn't support this device). I'm stuck on converting bitcoin atm. Either mouser or digikey are quite close pricewise with mouser usually being a bit better. Most parts are typically in stock at both. But the PIC16LF1459 is not at most yet as it's still new, though it's on order. It can be got directly from MicrochipDirect. I changed the oscillator to an Abracon ASMDB and Mouser doesn't stock those but digikey does.

Mouser just changed their resistor pricing. When I started a couple weeks back they had small qty available but now they'll only sell 10k rolls. Blah.

newbie
Activity: 33
Merit: 0
Any idea of lead time for parts? Is it time to order parts for power supply?

I´m looking at mouser and digikey in order to purchase some of these parts. They deliver really fast when they have stock.

Bkk do you have a preliminary PCB design for the power supply? I´m interested to make some reliability testing of this DC-DC converter  (full load and temperature).
full member
Activity: 378
Merit: 100
You can also get adapters for about $1 a piece.

The right angle 6-pin molex comes in a bit cheaper, part number 1586041-6.

You have to be very careful with the PCIe connector sourcing. It is not a standard MiniFit/MicroFit/ValuLok connector. That one you specify above will not work (check the drawing).

The standard connectors have a spacing of 4.2mm x 5.5mm, non square arrangement of pins, both in connector and on board. The proper correct PCIe connector has a square arrangement of 4.2mm both in connector and on board.

Correct part#s are Molex 45718-0002 for RA and 45558-0003 for straight.

I'd be very interested if anyone locates any Asian second source for these as 4UCON does not ahve them listed and they would be my goto for connector clones.


It looks like molex makes 10 different right angle connector to this specification. Gold, tin, tin over nickel, white, black, etc.
sr. member
Activity: 378
Merit: 250
Any idea of lead time for parts? Is it time to order parts for power supply?
full member
Activity: 378
Merit: 100
I have 4 or 5 local companies I will be contacting for assembly quotes this week. It may be that they will not give me a number without gerbers and cad files though.

Where are you located? Would you consider pooling for assembly?

I am in Detroit, Michigan.
full member
Activity: 378
Merit: 100
You can also get adapters for about $1 a piece.

The right angle 6-pin molex comes in a bit cheaper, part number 1586041-6.

You have to be very careful with the PCIe connector sourcing. It is not a standard MiniFit/MicroFit/ValuLok connector. That one you specify above will not work (check the drawing).

The standard connectors have a spacing of 4.2mm x 5.5mm, non square arrangement of pins, both in connector and on board. The proper correct PCIe connector has a square arrangement of 4.2mm both in connector and on board.

Correct part#s are Molex 45718-0002 for RA and 45558-0003 for straight.

I'd be very interested if anyone locates any Asian second source for these as 4UCON does not ahve them listed and they would be my goto for connector clones.


I've been told that the drawing for the connector is incorrect and that it is supposed to be the correct one. I will look further in to it.
full member
Activity: 309
Merit: 100
Out of curiosity BKK did you see this?

https://bitcointalksearch.org/topic/bfl-6-may-2013-asic-update-197675

Curious if any of that makes sense to you as I am no EE by any means ;p

kosta

What exactly is there to see?

Ente

Ente ... I was wondering about that info that BFL release as far as the Board Picture and something to do with molex and other Board specifics that again I have no clue about.


Was wondering if any of that made sense to BKK or if anything could be gained by that info for his Design.


Kosta
sr. member
Activity: 336
Merit: 250
♫ the AM bear who cares ♫
I have 4 or 5 local companies I will be contacting for assembly quotes this week. It may be that they will not give me a number without gerbers and cad files though.

Where are you located? Would you consider pooling for assembly?
full member
Activity: 250
Merit: 100
RockStable Token Inc
Yes, at this point we're all guessing...
hero member
Activity: 784
Merit: 1009
firstbits:1MinerQ
Thanks Bkk. I don't think we disagree on how it works. Each chip needs only a message to get started for every work item. Each work item can be several nonces, maybe millions of nonces, right? The things is, the less work you give to each chip, the faster it gets done. Each chip does not have to report the hashes it completes (unless it satisfies the current difficulty), yes, but still, the PIC has to send at least the start message (and you're right, maybe there's only one start message for all chips). The point still is that it's the PICs responsibility to keep the chips all busy. Why would ngzhang choose a Spartan-6 for this job? (This part is probably the most expensive part in the Avalon system: at least it's more expensive than each of the ASICs.)
I believe he chose a Spartan 6 because it has lots of pins, and he has great past experience with them, plus stock on hand no doubt. He had up to 32 modules in one case and saw the FPGA as a simple way to direct traffic from the FTDI to each module. My impression is that he didn't want to get into programming firmware for this. But I've never talked to him, and I'm guessing as I seem to keep fully occupied with my own work. Judging by Icarus and Avalon his goals haven't focused on finding the cheapest design solutions.

If he put an FPGA there to act as a nonce counter and feed them at high speed to the hashing chip then I would lose a lot of respect for his design. A counter is very cheap silicon space wise and it makes way more sense to put it in the ASIC and load it once per work unit.
Jump to: