Topic: Klondike - 16 chip ASIC Open Source Board - Preliminary - page 154. (Read 435369 times)

I don't think you can do this.  The pressure from the heatsink would probably lift the pads from the PCB and pop the ASIC off.  There are lots of excellent links here on thermal vias - seems like an industry standard for stuff like LEDs.

Not sure I like this idea.  Lots of cheap board places offer fixed sizes, once we exceed 10x10 the price will go up.

eg. http://imall.iteadstudio.com/open-pcb/pcb-prototyping/im120418009.html

Actually I think it would be smart to fund a run of two layer sample boards so that the assembly can be tested early.  This would allow us to build mechanical prototypes to check the heatsink attachement etc.  Bkk send me a PM if you want me to help with this.

http://www.mgchemicals.com/products/adhesives/electrically-conductive/silver-conductive-epoxy-8331/

Are there any tracks passing from the under side of asic chips placement? If not why don't we use my cooling idea?
http://mining-bitcoins.co/simple-cooling-idea-for-klondike-project/

Great progress, thanks!

Perhaps you should consider taking on a partner for the software work so you can focus on the electronics?

I haven't seen much on heatsink selection recently.  I'm wondering about the hole pattern and which heatsink it is designed for?

If possible, it might be worth putting 5 holes per quad (outside corners and center).  Those of us using heat pads will benefit from the compression. This would also allow people to use 4 heatsinks instead of 1.

You might also consider taking the design to a fab shop and getting DFM feedback on stuff like component clearances etc.

*edit* actually, thinking aloud, it is worth buying some other random chip in the same package as the avalon and having a run done to check on issues like this?

That is huge news... your boards if they work will have more functionality than the avalon itself. Just buy a few more chips reflow them on the board and volia more hash power.

Have you heard about these boards? http://blog.pnconline.com/?p=76

Would they help with heat issues?

I think I've figured out how to make the board auto-detect how many chips are installed. At this time I believe you'll be able to have anywhere from 1 to 16 chips on a board and it will mine with whatever is present. This kind of depends on the protocol specs coming soon, but if things go as I expect then this will be possible, and I'll write the code to support this functionality so that people can install more chips as they get them.

Also, the I2C daisy chain will auto-detect how many boards are connected. I've got that worked out regardless of Avalon specs.

( As an aside, I'd like to know if using silver epoxy solder available at some electronics shops is viable for on the fly upgrades. It may be easy to paint a little bit on the QFN tabs and then carefully place and hold until hardened. My question would be heat transfer - whether silver/epoxy will conduct enough heat? )

Have updated original post now.

I wouldnt take it, CC's are possible to chargeback. I had this in my shop a couple of times through paypal. PP didnt had a chance against the CC-Companies then.


+1

I want to see you rewarded. I think the ultimate test would be the sample miner anyway. But i think the sample chips will be a few only, so probably you need to think about how to test a board with only some chips on it.

Adding 5mm to all sides of a 100mm board would be 110mm.

I agree with others, Burnin'. I wouldn't post anything until you're sure it's actually ready. Personally, I think you should hang onto the prints for long enough to make sure that you get paid for your work, before somebody else swoops in and steals it, and starts selling boards from underneath you. I think you deserve to at least have a decent head start on selling boards before everybody else.

If I were in your position, I would finalize plans, run a group buy/order of assembled boards (minus chips) done by an assembly shop, sell them to anybody who wants them, and then release the prints to the public something like 2 weeks after you've shipped boards to those who ordered.

This gives you a window to recoup some money for your efforts, and gives people an incentive to order through you, while allowing people to chose other routes if they're willing to wait a little longer, in order to be cheapskates. In addition, if you don't feel like dealing with organizing the assembly and shipping, you could license your plans (with a non-disclosure agreement) to a few people who are willing to give you a cut of the profits while they produce and sell the boards, and again, hold back the plans for public release for a set window of time.

I'm all about the open source, but I believe you deserve the opportunity to make a return for all your hard work. Regardless of how you work it, I'll be sending donations to you as well (in addition to whatever I have to pay to get your boards), once I am up and mining Bitcoin.

Can I pay through Visa?

I don't have much BTC

Thanks!

1. Easy to add 2.5mm to the edges to get 105mm x 105mm around is best suggestion for those needing manufacturing clearance.
2. I think you should limit your designs as much as possible until you are good and ready and have a prototype you are happy to release.
3. Take your time on the update to the first post spend whatever time you need on the important stuff we can point to the posts in the thread if people need clarification for now.
4. NICE looking boards... compact... 4.5GH/s monster that will be 32W!!! Meow!

THANK - YOU!!!!!!!!!!!!!

There is almost that now but there are a few parts within 5mm of the edge. Not sure about that as I didn't really design it for automated assembly lines (something which I know nothing about). I suppose you could use a slightly bigger board size.

´thank you. One thing, can you have a small border around the board 5mm? with such a border it is much easier for assembly lines or something like a pulling border?

Thank you for doing this all... i hope you will make some bitcoins with your work...

No. I've made assumptions here that may require changes when the protocol docs come out. I wouldn't make the first boards without knowing the full specs.

I can use the waiting time to double check the design files and verify all the parts footprints match the datasheets. These kind of things can really mess up. Almost all the footprints are my own but it's very possible for mistakes to occur.

Dont overdo it... and when you decide to share it for feedback, simply write to it that its an untested design. If someone is foolish enough to build the pcb with it and it doesnt work afterwards you warned about it before.
Is the avalon reference already this far that you know exactly how the chips communicate? So that you can finish the board theoretically and only has to build it to check that it works?

I'll try to get that updated soon. Same with my github readme. Too much to do.

~~~~~~~

I'm in the process of tidying up the silkscreen layer now, and trying to make it more readable. That means deleting most of the junk on there. I haven't changed the thermal vias grid yet, or done ground stitching but other than that all tracks are run. The design rule checks are clear.

The parts list has been updated again. New parts added and double checked against the BOM spit out by Kicad. Also for first time the K1 Nano parts list is uploaded. There's a couple alternate parts listed now as well.

I'm really debating myself about when to upload the schematic and PCB design files. I want to share and get feedback, but at the same time I'm worried that someone will take them pre-maturely and think they're going to make a product before I even have a chance to finish the work. I really don't want to hear of 1000 boards being made and then be blamed when it doesn't work. This thing is still being developed, and I'm moving as fast as I can.

~~~~~~~
Now - here is a couple new images to soak up.

Just looks the same almost but the next one shows you what's new under the hood.



I think this one looks pretty cool - like a city of parts with the waterworks showing.

As long as i know we still do not have them. Bkk will build demo pcb first solder the chips test it with FW mininig and tweak it if something went wrong. After that we will have them.
Bkk can comment it also but that should be right way to go as long as he is not making just clone of Avalon PCB but designing it from scratch. Providing final schematics in the moment will be like "shooting in the dark"

Please forgive my ignorance but have we got the final schematics to work with? Or just the PCB board gerber file?

I used a soldering iron in the past, including soldering paste and tin solder... of course this looks like the first grade in soldering. I didnt use an oven yet for soldering.
When i see the videos im pretty sure i can handle the DIY.

I really would prefer if there would be stencil available too directly from BKKCoins. I mean there could be created one big stencil that serves all chips that has to be soldered right? And bkkcoins would be in the position to create a bunch of it for very low cost, while private persons would have a hard time to do the same. So if possible bkkcoins, please think about this. I would pay what it needs.

Im not sure how many chips i will have at the end... maybe i will check out if i can help somewhere. Because i think one of these plates will take some time too, to be created. But i think ill give it a try on my own too... and i think youre right... i will buy 2 sets on top in case im not as good as i thought and im crashing something. But normally i have good hands for diy in general. So i think the stencil is, with the oven, my biggest doubt. Everything else will need information on how to do, what to do and so on.