So the driver will work with 3 or 4 nimh or alkalines to power a p7, but the efficiency of 3 vs 4 nimhs is the problem being discussed?
Is there any heatsinking needed for this board, especially if it is running 4 alkalines (6V)?
I doubt that alkalines can deliver 2.8A at say 3.6V. You need NiMHs.
Efficiency is a secondary, or even tertiary, factor. In my mind, the primary issues are running in regulation, and running in regulation for a decent amount of time.
What that means is delivering enough Vbatt to reach regulation and to stay in regulation as the cells discharge and their voltages drop.
Read my post #44 in that AMC7135 thread. You need to satisfy two criteria -- Vf > 2.7V (with the reverse polarity diodes removed), and Vbatt,min > Vf + 0.12V. The first criteria should be no problem. At 2.8A drive, a P7 should have a Vf of at least 3.2V.
For the second criterion, Vbatt,min > perhaps 3.3V if you have a low Vf P7 (will be more if you have a higher Vf emitter). In reality, you probably need a few tenths more because of parasitic resistance in your flashlight, due to things like contact resistance (e.g., springs, cell-to-cell). So let's say that your Mag has a total parasitic resistance of 0.1 ohms. At ~3A, that is ~0.3V drop. So, you might need a Vbatt of about 3.6V to reach regulation for this hypothetical case.
Based on Silverfox's NiMH shootout data, 3xD Accupower NiMH (at least the Accupowers that he tested) can hold over 1.2V per cell at 5A draw. So those cells should give you near-max run time in regulation at the ~3A draw of an 8x7135 driver.
But if your P7 has a high Vf such that you need a Vbatt,min of say 3.9V, just to pick a value that is unfavorable, then you will need your cells to hold at least 1.3V under a 3A load. Looking at Silverfox's D Accupower curves, you can't get there from here, and so 3xD NiMH is out. You will need 4xNiMH to reach regulation and run in regulation for a satisfactory amount of time.
Also note that if you plan to use 3xAA Eneloops in D shell holders, the cells probably won't be able to hold above 1.2V per cell for very long and thus your run time in full regulation will be short. So, you will probably need 4xAA Eneloops for this batt configuration.
Alternatively, you could drive the P7 at a lower drive current, say 2.1A with 6x7135. At the lower drive current, LED Vf is also correspondingly lower and might allow you to use 3xNiMH, when you would have needed 4xNiMH if you chose an 8x7135 driver (2.8A).
To assess your situation, you need to be able to determine your LED's Vf at the relevant drive current.
Once you've established how many cells you need to run and stay in regulation, you're all set (assuming that Vbatt doesn't exceed the Vin rating of the 7135). Now efficiency comes into play. For the 7135, efficiency is essentially Vf/Vbatt. If the amount of waste heat isn't too bad, you might get away with no thermal management of the driver. IMO, if the waste heat starts to exceed 1W, I'd start to consider heat sinking the driver.
The large tab on each 7135 chip is the desired contact point for heat sinking. But, sinking to the 7135 case is better than nothing and probably easier to do. I'd probably use various convenient pieces from copper plumbing (easy to cut, bend, file, and shape) to thermally glue to the 7135 chips and then sink (thermally glue) the copper to the Mag tube.
Thanx!
