I'm leaning more towards buying S4's than buying any more S3's. My farm is downstairs in the recreation room against the wall adjacent to the garage. It's nothing to change out the wiring to 10/3 AWG; install 30 AMP NEMA L-14-30-R outlets [circular]; install 30 AMP 2 POLE BREAKERS and plug in a 3 ft. Generator 30 Amp 4-prong to 15-20 Amp (x2) Y-Adapter which will power two S4's.

BY THE WAY: DOES ANYONE KNOW IF THE INTERNAL PSU GOES BAD IN AN S4, IS IT EASY TO CONNECT AN EXTERNAL PSU TO KEEP IT GOING?I'm in the process of installing a 125 AMP SUB PANEL [QO SQUARE-D]. That is enough for four (4) 30 AMP outlets to power eight S4's [16 TH/s].

8 S4's x 1380 watts at the wall = 11,040 watts for 16 TH/s TOTAL without over clocking at 2 TH/s EACH connected to four 30 AMP outlets.

OR

30 S3's x 368 watts at the wall [with Corsair AX860 PLATINUM PSU's] = 11,040 watts for 13.65 TH/s [without over clocking] at average of 455 GH/s EACH connected to five 20 AMP outlets.

To me it's a NO BRAINER to invest in S4's for more hashing power with lower costs in power consumption. I'm keeping my 10 S3's with four more on the way for a total of 14 S3's. However, those four S3's will be the last S3's I purchase. I believe it is S4's from here on out till I have 8 of them.

That means [when completed] my house farm will be 6.37 TH/s of S3's [5,152 watts total] at average of 455 GH/s each [without over clocking] and 16 TH/s of S4's [11,040 watts total] at 2 TH/s each.

TOTAL = 22.37 TH/s Hashing power with 16,192 watts [Divided by 120 watts per amp = 134.93 AMPS] at the wall.

That leaves me 65.07 AMPS for the rest of the house. The AC runs at 19 AMPS when it is on. It starts with 30 AMP surge then settles down to 19 AMPS. ALL LIGHT BULBS in the house are LED. So I believe I'm good to go with the 200 AMP MAIN BREAKER PANEL and everything else running in the house, such as AC, stove and oven, fridge, freezer, LED TV (only two - LED TV's - use less power than any other type of TV out there) and LED lights.

Power cost 1 month: 1774.24 USD @ 0.15 per kWH

Revenue per 1 month: 3290.69 USD @ $340.00 for 1.0 BTC / 34,661,425,924 Difficulty

Less power costs 1 month: 1516.46 USD @ $340.00 for 1.0 BTC

We need BTC to go up to $440.00 US and the difficulty to stay around 35,579,431,281:

Power cost 1 month 1774.24 USD @ 0.15 per kWH

Revenue per 1 month 4258.54 USD @ $440.00 for 1.0 BTC / Estimated Next Difficulty: 35,579,431,281 (+2.65%) Source: BitcoinWisdom

Less power cost: 2484.31 USD @ $440.00 for 1.0 BTC / Estimated Next Difficulty: 35,579,431,281 (+2.65%) Source: BitcoinWisdom

IF BTC went up to $540.00 US and the difficulty went up to around 42,579,431,281 the profitability would be almost the exact same:

Power cost 1 month 1774.24 USD @ 0.15 per kWH

Revenue per 1 month 4254.50 USD @ $440.00 for 1.0 BTC

Less power cost: 2480.26 USD @ $440.00 for 1.0 BTC

AT PRESENT ON SLUSH WITH 23.7 TH/s HOUSE FARM WOULD BE THE FOLLOWING:

8230601 GH/s (8.230601 PH/s OR 8231 TH/s) is the TOTAL HASHING POWER OF SLUSH ON AVERAGE AT THE MOMENT (estimated)

Block found in 5 hours 7 minutes on average (luck plays a part in it as well).

0.07013480 PER BLOCK FOUND with 23.7 TH/s (estimated) x 5 blocks each day = 0.35067402 each day

THAT WOULD TOTAL 10.5202207 BTC EVERY 30 DAYS

10.5202207 x $340.00 BTC = $3,576.88 US - $1,774.24 for power consumption; with present difficulty

10.5202207 x $440.00 BTC = $4,628.90 US - $1,774.24 for power consumption; with Estimated Next Difficulty: 35,579,431,281 (+2.65%) Source: BitcoinWisdom

So my math compared to the calculator at http://www.bitcoinx.com/profit/ isn't too far off to my first estimate by that calculator.IF that link does not load the calculator, HIT REFRESH and it should come up. Also, I believe the "Profitability Decline Per Year" at 0.61 will be much less as difficulty goes up without the price of BTC going up with it. That profitability calculator was made when the Difficulty was just over 19 Billion.NOT BAD AT ALL FOR A HOUSE FARM.