I am testing the growth between two 40Bs. One has a single 250wattDE, IC ballast, pheonix14,000K. The T5 is a 6x39watt Tek. I will change out all the bulbs at once in a coupe months (after the newer T5 setup should be done cycling) and take PAR readings. I am using the ATI/D-D/Geisemann bulbs (all the same), with 2 actinic03, 2 blue plus, 2 11,000K.
I just want to point out that the IC ballasts arent worth it. There is a limited amount of phosphors in any bulb, and once they reach their limit, pumping more electricity through them results in little additional output, and more heat. I remember tests done on VHO & PC bulbs that showed that a 10-20% increase in voltage increased output by 5-15%, and more voltage than this didnt result in more light.
The tests I have seen with T5 support this. While the IC ballast pumps 50% more juice through 3 and 4ft bulbs, it only increases their PAR by 30%. A better use of that electricity, as well as the high cost of a IC ballast, would be to simply buy 50% more reflectors, 50% more ballast, 50% more bulbs, and run them within operating spec for 50% more electricity = 50% more output.
The idea that IC throws out there that 'soft start' somehow extends bulb life is absurd to us reefers. We replace out bulbs long before they cease to start and create black deposits at the ends. While this might be useful for commercial uses, for reefers, we replace our bulbs when the phosphors wear out...and this is due to heat. The IC ballast creates more heat.
I believe your bulbs are losing that much intensity that fast if they are running hot. You might consider a better cooling setup to counter the effects, or swap out ballasts for more bulbs.