Quote from: QuintLeo on July 12, 2015, 01:56:31 PM

Quote from: J4bberwock on July 01, 2015, 04:56:14 PM

I don't know for the G black (still haven't got one in hands), but the heatsinks on the G blade are designed to dissipate something like 200 watts.
From what I can remember, my first fully overclocked G-Blade was drawing 350-400w.

With SF3301, 200 watts would be 5-6 chips hashing at a total of 400-500GH SHA, and 10-16MH scrypt on each PCB.
The number of chips on board is probably not high enough to make it economically interesting for the end user, but it can be done.

My dev board is 150x100mm

How about 8 chips, running at .6 volt 450 Mhz (BTC about 570 GH/s) .8v 450Mhz (LTC bit over 9 MH/s) option? Should be easy to fit on a 4"x8" PCB (perhaps a bit of overhang on one end to fit taller caps if needed kinda like the original GBlade boards), and the common clock in theory should make the board design a bit simpler and less expensive? Might be better to have non-common clocks kick the LTC side up to the .9v 700Mhz (15ish MH/s) option.

Current pricing on the SF3301 chips might make this a bit hard to keep to a reasonable price though - $300 or so would be competative with anything currently available IF doable but the SF3301s alone would be $240ish....

What you are asking for is one panel of the SF100 miner.
You have an estimate of chips price, then, add the price of the voltage regulators for both sides (BTC + LTC), heatsinks, and your final price won't be too far away from the unofficial SF100 price.

I'm waiting for a few informations regarding the 2 chips boards before I can confirm that it'll be a viable miner that's worth manufacturing.
The design isn't fully fixed yet for the voltage regulator. The first board uses a converter with integrated MOSFETs to save some space on board, but the definitive one may use a standard pair of MOSFETs like the old gridseed pods/blades to reduce cost for the end user.