Topic: DIY PCB with AVALON: "The Quarter Stick" - Needs Help! - page 25. (Read 89497 times)

Just waiting for some basic Chip specifications from Avalon to make sure the design approach is adequate.
As soon as this is available. It is full steam ahead!

Dude,
all of us are out of patience. But there is no design yet. Cool down...
We have to wait ......

Im in for the groupbuy as well...

I wanted to ask if it will be possible to overclock the chip with the design. Currntly the avalons allow to change the clockrate.

Allten,

Depending on the power requirements of the other components, it appears that a single chip could fit within the allowance given by USB 2.0.  Here's a quote from earlier in the thread.


All,

The Avalon Wiki page has more pictures now: https://en.bitcoin.it/wiki/Avalon

Hey Burnin,
     Sounds like you are making some good progress. That's great!

I don't have any experience with the design or firmware/software for the FPGAs or ASICs.
Do you have an AVALON? You wouldn't happen to know the clock fed into the Avalon ASIC would you?

One advantage of starting small with these simple USB ASICs, is it would allow us to prove the concept
and we would be much more likely to get 20 to 30 "samples" without the need to order 10K. Once the
design is proven, getting crowdfunding for 10K chips should be easy.

thanks

Well, for some reason I thought USB was capable of 1 Amp. Now I see what others are saying
with regards to using 2 ports for this device. Looks like it's time to redesign a little.
One objective of this DIY project is to keep it ultra simple. Once it is proven and works,
no doubt it will be followed by other designs.

http://en.wikipedia.org/wiki/Universal_Serial_Bus#Power

max Current is 500 mA with a minimum Voltage of 4.75V.
500mA * 4.75V = 2.375 W.

It would probably be a good idea to design it so the current
being used can be measured.

Also, the design should target for about 80% of what the
maximum current since it will be continuous draw and it's
a good design practice to leave some wiggle room.

Ok, Tell me if I'm calculating this correctly:

There are 620 Watts used for 240 Chips. Correct?

That is 620W/240 = 2.583 Watts per Chip; Therefore, we should only need
1 chip for this first design to prove the concept and lead the way to other
Open Source Designs.

This should be around ~280MH/s.
 

+1

Definitely have some dollars to help move this along. Very interested in using these modules with our Middle School / High School students as part of the curriculum.
[email protected]

I would commit funds to this.

Edit: BTC funds specifically, but I have dollars too.

Actual results: expect some from me.  

I made significant progress with my design.
Currently i plan with a PIC32 as controller.
Regarding the Host interface, i will use the Avalon host protocol.
Any host software that supports it will work. (currently cgminer)
It seems to be flexible enough.

The design will be close to the boards in the Avalon, but with on-board mining controller and stacking connectors.
I'll do 3d renderings soon so you know what to expect.

I don't have the BTC to buy 10k chips.
We will need to arrange some sort of group buy.
(Yifu will most likely insist on BTC.)

I can do large scale manufacturing. (also for other designs)
I'm based in Europe so its going to be more expensive then China.  

what NOT to expect: pretty websites or (pre)ordering systems, i'm an engineer.

They are in QFN48 packages.

Well hopefully if something actually results from this discussion then there'd be an alternative to Avalons and BFL. Sure, the chips from from Avalon but there would at least be a third hardware vendor. If the chips are so expensive that it requires a group of people to purchase the chips and pay for assembly then I suppose you could consider it a group buy.

I wonder how much it would cost to have someone like Enterpoint assemble a device based of Avalon chips. Community provides the PCBs and controller firmware, Avalon provides the chips, Enterpoint assembles them.

so, one asic uses what, 240 chips? Is that correct?
You could build 41x60,000/hash asics with that. THat would be quite difficult to cool off. maybe people will do group buy? Like batch 3
Would you, guys, do group buy? DIY asics will be interesting IMHO

My gut feel is $5 per chip (hopefully less )

And I also hope the pricing its in BTC, let's forget about fiat

My WAG was $25k-$50k for 10,000 chips.  I based that number partly on the sale price of the batch 2 machines.

However now that they've moved to BTC-denominated pricing with batch 3, who knows?  Maybe the chips will only be available to purchase for BTC...

I wonder what those "chips" taste like...

fried asics. yumm

The price is completely open yet.
Submerging in DIY and then using it with a USB would be a surefire way for accidents. Plus you would need to create a device that is fully sealed. And you would need much oil because the strenght only lies in the distribution into all the oil. If the oil isnt cooled then after a time the temperature will rise constantly.

so... does anybody know how much those 10,000 chips will cost?
I guess for cooling these DIY asics could be submerged into mineral oil or thermal liquid like intel uses

I'd really like to see a device hit the market that you could expand in a modular fashion as you become more confident in mining. With each chip providing 275 MH/s, use 8 chips like one of the sub-module's in the Avalon for 2.2 GH/s. Each module would consume ~25w so you could power 3 from a 6-pin PCIe connector. This could be an interesting product to bridge the gap between BFL's Jalapeno and the 66 GH/s Avalon and the 25 GH/s BFL SC. I fear if you aim too small then it won't be that interesting. $25 is a neat price point but if anyone does the math the monthly income would be depressing. Right now a 6.6 GH/s device @ 75w @ 0.11USD/kWh would produce $60 USD/day. That's a significant amount. With 10x difficulty it is still making $42/week. At 100x difficulty it has a 4 year ROI @ $250 USD price point but by the time we hit that difficulty we'll hopefully have more efficient chips. This all however becomes useless if Avalon prices their chips in a similar fashion to their rigs.

Something stackable like Enterpoint's Cairnsmore boards should be something to consider. A parallel data bus using a cheap cable so you can stack a couple units and save cable clutter. Power stacking such that something like a 6-pin PCIe power connector can power three boards would also be great. Learn from Enterpoint's mistakes and include simple things like an on/off switch.

This is great to see.....Seeing what BFL just pulled....Nearly double price and less hash. I understand Supply /Demand but come on I thought we were in this for the good of the Bitcoin?...

If us simple folk can help in any way please ask!