I too like the look of the BFL unit mainly because of Mhash/$ this makes payback assuming constant difficulty and my power cost with 6$ per BTC 4.63 months, at 3$ per BTC it is 9.7 months. I realize that difficulty is dynamic and lower prices cause a drop in GPU because of profitibility and higher prices will undoubtedly push payback beyond 5 months.
Your numbers are a bit off. At 832Mhash/s and a difficulty of 1250757 (current) and a BTC price of $6USD/BTC, you will generate exactly $4.01USD per day. At $0.10/kWh, electricity costs will eat $0.12USD per day, making your profit $3.89. That makes payback
at current levels in 154 days (5.1 months).
In 9 days when difficulty increases by 10%, you will generate $3.49 per day, for a payback of 172 days (5.7 months).
Keep in mind that the above numbers are overly optimistic; the calculated payback periods assume difficulty and BTC price will remain relatively constant. More realistic, however, is that if the BTC price stays at the $6-$7 level, you can expect difficulty to rise further over the next few months. It is overly optimistic to expect/ssume difficulty will remain constant.
Back in September 2011 when BTC prices were hovering in the $5-$8 range (as they are now), difficulty was 1750000. That's 40% higher than today. It may be wise to assume that if today's BTC price holds for any length of time, difficulty is likely to reach 1750000 again.
And in that case, any payback period will be extended by 40%. So instead of 5-6 months, you are looking at 7-9 months. And that's *if* BTC price stays at $6/BTC during that period.