Here's the latest figures, now running a 20Mhz crystal on ICA 7 which is stable at 406.2 MHash with ASIC powered externally with 1.7V @ 3.96A. ICA 6 is the stock reference running at 333MHash.
bfgminer version 3.1.3 - Started: [2013-08-15 20:58:51] - [ 0 days 03:00:24]
-------------------------------------------------------------------------------
5s: 2.66 avg: 3.65 u: 4.37 Gh/s | A:2752 R:30+0(1.1%) HW:92/.82%
ST: 4 GF: 0 NB: 21 AS: 1 RF: 0 E: 9.66 U:15.3/m BS:36.6k
Block: ...5e0f8f39 #252279 Diff:50.8M (363.7Th/s) Started: [23:45:10]
-------------------------------------------------------------------------------
ICA 0: | 960.4/364.6/422.2Mh/s | A:266 R:4+0(1.5%) HW:10/.88%
ICA 1: | 543.2/363.7/393.6Mh/s | A:248 R:1+0(.40%) HW:13/1.2%
ICA 2: | 245.2/364.5/477.8Mh/s | A:301 R:1+0(.33%) HW:10/.85%
ICA 3: | 423.3/364.2/455.5Mh/s | A:287 R:3+0(1.0%) HW:10/.89%
ICA 4: | 581.1/364.4/447.6Mh/s | A:282 R:6+0(2.1%) HW:10/.91%
ICA 5: | 348.0/361.7/441.3Mh/s | A:278 R:1+0(.36%) HW: 9/.84%
ICA 6: | 394.9/333.8/296.8Mh/s | A:187 R:3+0(1.6%) HW: 7/.85%
ICA 7: | 294.3/406.2/563.5Mh/s | A:355 R:4+0(1.1%) HW:11/.79%
ICA 8: | 532.0/363.8/460.3Mh/s | A:290 R:5+0(1.7%) HW: 5/.44%
ICA 9: | 239.9/365.9/411.1Mh/s | A:259 R:2+0(.77%) HW: 7/.61%
-------------------------------------------------------------------------------
I've bought a few crystals and heatsinks too many so if anyone wants me to upgrade their units to either 16 or 20 mhz speeds I'd be happy to do it.
2 day maximum working turnaround and high quality soldering done with an SMD rework station with all components, heatsinks and external power supplies (if necessary) included. Please PM me if interested.
That is Beautiful. how exactly are you running extra power to it? I understand the amount but is it from a extra power brick you had laying around?
When I did the math I say that its not possible to run it with on board power past 16. at 20 you need extra power. Please update after 24 hours, I am quite curious to see the results.

That's what I got when I plugged some numbers in.