I didn't try with two fans, but the PCB had very good temp too. I think that the most important thing with this cooling is a huge air flow at the heatsink. The same air that cools the heatsink is going to cool the VRM plate, so I think you don't need a second fan for that plate.
As you can see, when I use the fan regulator to set the delta to the minimum voltaje of the regulator (I supose 6v), I have 13ºC more than with the 12v applied to the fan. At full speed is a ~130CFM fan. So it's possibly that a high CFM fan do better job that two regular CFM fans.
Well, here is other idea that I tryed last week. First I asked to a friend, that has get a MK13 heatsink for his 470, for his shroud. I bought a 9cm delta fan AFB0912HH and the single slot bracket used in the liquid cooled EVGA 470 FTW. My idea was to test if the famous high flow bracket was useful or not, and most of all, to see what happens if I attach the 9cm fan to the shroud, for helping the graphic card fan.
The single slot bracket needs some modifications:
I've cutted a hole over the heatsink of 4,5 X 9cm (air flow direction is not prependicular) and attached the fan.
And then I started the tests.
The enviroment was the same, except for the 8800 ultra that I use for physx (with the 120x38 fan it doesn't fit). the room with constant 24º with AA. Unigine Heaven in window 8xaa ande tesselation extreme loop, 800/950 OC, ten minutes per test.
here are the results:
With the unmodified shroud and the original bracket:
Max 82º Fan 84%
With the unmodified shroud and the modified single slot bracket:
Max 81º Fan 82%
With the modified shroud, the 9cm fan and the single slot bracket:
Max 75º Fan 77%
As you can see the EVGA high flow bracket seems to be unuseful, but that 9cm fan helps a lot. I supose that with the 9cm fan and the graphic card fan set to 84%, there would be a couple of grades improvement.
bye.
