So I did a little testing and want to post the results I got (if anyone wants to put it into an easy to read graph that would be great).
So I measured the voltage out of the wall socket at 118v which goes to my 3,000 watt step up voltage converter/transformer. Out of the step up converter/transformer it measures 230v. I've got my APC Smart-UPS 1500 (1,000 watt) sine wave (230v version) connected to that (which I measured only uses 1 to 2 watts when the battery is full). Also have a 5 port switch that measures about 1 watt of consumption at idle connected. The ThorTech Thunderbolt Plus 1200 watt power supply is connected to that and the pair of AntMiner S3+ are connected to that (the display is showing that the power supply is running at a 92.77% efficiency). So the measurements on efficiency and power are for the pair of (batch eight) S3+'s working in pair and mining for at least 5 minutes at each speed before the measurements were taken. Note that the case the AntMiners are in has 6 additional 120mm case fans as well as a fan controller. The wattage measurements fluctuate so I took the closest average.
Speed: #1 (GH/S avg) #2 Wattage at the wall Wattage after the Step-Up C/T(230v) Wattage after the power supply 12v (for pair of S3+'s and 6 additional case fans and cont.)
250M 486.62 499.42 928 877 825
237.5M 481.25 479.21 870 820 778
225M(S3+) 450.55 448.46 818 774 725
218.75M(S3) 438.89 438.95 790 749 700
212.5M 428.44 414.86 765 725 676
200M 407.21 411.31 717 679 630
193M 390.32 387.29 697 658 610
175M 342.16 356.92 630 595 548
Looks like I'm going to underclock a little 3 S3+'s down to S3 default speed of 218.75M on a EVGA 1300 watt G2 power supply for my next rig for about 1,050 watts at the power supply.
Here's a little odd fact I found out this weekend, when I set up a pair of S3+'s (at 237.5M for the pair) with a Corsair AX1500i Titanium power supply on the same power set up as above and only 4 smaller case fans, the wattage was about 20 watts MORE. This shouldn't be as the power supply is Titanium rated and set up is closer to the 50% load rating.
Thanks for the results. In my experience it takes a lot longer than 5 minutes for hash rate to stabilize, 1 hour is more like it. The power readings may settle down earlier, however there will be some variations as things warm up, fans spin higher etc.
Cheers