The P90 and P91 are both used with three lithium 123's = about 9.6v (fresh cells), so using two 17500's would not overdrive them as the 17500 x 2 voltage would be about 8.4v fully charged. However, the 17500's/17670's have a flatter discharge curve than the lithium 123's (not as much voltage sag under load), which gives kind of a semi-regulated effect, causing the P90 or P91 to stay whiter longer than on lithium 123's. I don't notice much dimming until maybe about 5-7 minutes before the cells shut off.-socom1970
there seems to be some confusion about li-ion giving regulated effect, and some confusion about the difference between li-ion and lithium primary configurations as it pertains to the "overdrive effect."
For the sake of explanation, I'll use a P90 and P91, 2x17500 vs 3xCR123.
CR123s, into the load of a P90, don't spend any longer than a fraction of a second maintaining 3.2V per cell (though they often do measure 3.2V per cell OPEN-CIRCUIT). Within the first second, they drop to about 2.75V per cell, and then level off at about 2.5V per cell after within a short moment. A CR123, will then continue delivering very near 2.5V for about the first 30 minutes, only dropping to about 2.4V at around the 30 minute mark. At which point it starts to gradually decline to around the 2.15V mark before suddenly falling off the map.
So.... what does this do for output? Well, lets assume that the first few "moments" don't count for anything besides an expensive thrill, so the "starting point" that we'll call 100% output is occurring with 7.5V worth of cell voltage available. After 30 minutes, output drops to 87%, not bad, at the ~60 minute mark it drops to about 60% output, then somewhere around that point it "falls off the map" (goes totally dim and useless within a few moments). So we go from about 100% down to around 60% output before the batteries are about to "die." That's not too bad.
Lets compare that to a pair of 17500s on the P90...
A 17500 does come off a charger at 4.2V, but like the primary cell, this is an open-circuit-ONLY voltage. Under the load of a P90, a 17500 cell immediately sags to ~4.0V and begins a relatively steady decline in voltage through the length of the run. In the first ~30 minutes or so, they drop to about 3.7V per cell, then over the next ~30 minutes (just before "falling off the map") they drop to about 3.45V each. So we actually start off quite a bit higher than the CR123s, at about 124% of the primary configurations output to start with, declining steadily to about 95% of the primary configurations initial output in the first 30 minutes or so, then declining to about 75% of the primaries initial brightness at about the 1 hour mark (give or take a few minutes depending on cell brand).
In comparing these 2 situations, we see that the lithium-ion actually runs brighter for the majority of the run. But this can vary depending on what brands of cells are compared. BOTH configurations drop to about 60% of their initial level just before the cells "die." So lithium-ion really isn't all that much flatter than a primary configuration... But, keep in mind, that the initial "spike" on CR123s (of around 2.75V per cell OR HIGHER in some cases) is actually probably as hard, or harder on the bulb as running with a more controlled overdrive on li-ion cells. (the li-ion would rarely if ever hold above 8.1V in this configuration at the cell, where CR123s could theoretically deliver 8.3+V at the cells for a short moment before they settle down)...
Now.... When dealing with the P91, we are getting into some serious power. 2.5A has been reported on 3xCR123 configs, and 2.7A is routinely reported on 2xli-ion configs (I can verify this through personal experience)... this fact alone tells us that there is some substantial overdrive going on with li-ion cells in play.... but we can simply refer to the charts for further enlightenment (Silverfoxes battery charts)...
Interesting thing when you get into loads like this, CR123s really don't have a chance, literally half of their available power is wasted as heat, they start off into 2.5A at about 2.2V and drop to somewhere around 1.6V in about 20 minutes. So, assuming a P91 starts at "100%" at 6.6V on primaries, it drops to something like 40% output around the 20 minute mark, then falls on it's face.... but keep in mind here, that at these loads, there are HUGE variations in performance from brand to brand, and the discharge characteristics will even vary heavily from cell to cell quite dramatically as well. It's really a great way to flush CR123s down the toilet, lol. Sorry kinda getting off on a tangent here... but moving to a 4xCR123 configuration can actually deliver very similar lumens, but for a full hour as opposed to only 20 minutes. (because "12V" style lamps have filament designs that happen to tend to be more efficient, and you can drop down to a lower current, at which point the cells hold their voltage better)
Anyways... take the same P91 on 2x17500 and you actually start off around 3.75V (or higher) into the load per cell, dropping to around 3.2V per cell before falling on it's face. So on 17500s a P91 actually starts off at about 150+% of the initial brightness on primaries, and actually ends up at about 90% of the primaries initial brightness at about the 20 minute mark. Which is totally awesome, this does shorten bulb life. And there is still a substantial decline in output through the run, but in this case, it's much better than primaries.
What would you suggest in terms of cell extenders? Anything from Leef?
