LED61,
No problem. It IS confusing. The issue is that a lamp can be driven over a range of voltages, from where it produces only heat, to where the filament is just about to vaporize because it is so close to the melting point of tungsten.
The MN61, driven by 4 primary 123's is less powerful than the MN21 driven by two parallel stacks of 3 primary 123's (6 total).
True. However, we are not using 4 primary 123's. A primary 123 under near maximum load (about 2.5 amps) holds around 2.2 to 2.3 volts each. So 4 of them will hold about 9 volts or so, depending on their temperature and state of charge. But we are using (aren't we cool!) 2 parallel stacks of THREE LI-ION rechargeable cells. I don't have SilverFox's graphs in front of me, but IIRC a Powerizer of Lighthound R123 will hold about 3.5 volts with a current draw of around 1.2 to 1.5 amps (i.e. each stack only needs to provide half the total current).
SO, three cells at 3.5 volts/cell is 10.5 volts. Quite a bit more than 9 volts, isn't it? This will mean higher current draw. Just exactly how much really should be determined by measuring it--both voltage and current in the setup being considered.
However, there are handy dandy rules of thumb for re-rating a lamp of known parameters, to a different voltage. These are known as the re-rating formulas. Here they are:
Ir = (Va/Vd)^0.55 * Id
Cr = (Va/Vd)^3.5 * Cd
Lr = (Vd/Va)^12 * Ld.
Here the "r" 's represent the re-rated values and the "d" 's represent design values. The "I" is current draw, the "C" is mean-spherical candlepower, which is proportional to total bulb lumens (multiply by 12.5 or something like that), and the "L" is lamp life--number of hours of operation you are likely to get. It's a statistical thing. At the L rating, half of the lamps in a group sample will have burnt out. But half will be still going.
Notice that the Life re-rating is HIGHLY non-linear: a small relative change in applied voltage will mean a large change in lamp life.
*cough* OK. More info maybe than you wanted. Let's apply it.
If the MN61 at design is 9 volts, 2.5 amps, 350 lumens and 35 hours of life, what happens at 10.5 volts. OK. First, let's adjust the 350 lumens figure. SF always under-rates by at least 20 percent. So let's just crank that up to 420 lumens, which is still conservative.
Ir = (10.5/9)^.55 * 2.5 amps = 1.16666^.55 * 2.5 amps = 1.08 * 2.5 amps = 2.72 amps.
Lr = (9/10.5)^12 * 35 hours = .157 * 35 hours = 5.5 hours.
Cr = (10.5/9)^3.5 * 420 lumens = 1.715 * 420 lumens = 720 lumens.
So yes, it does look as if the MN61 at this voltage will out output the MN21. But let's hope that 35 hours of life at 9 volts is very conservative! Otherwise we're looking at a scarily short life span.
Going back to the current, 2.72 amps disagrees with the CPF info I found in one of the incandescent forum sticky threads of interest. They claimed 2.85 amps on three Li-ion cells. A measurement will be required in any case, but I should point out that the re-rating formulas are only approximate, and the farther away you go from design the less accurate they are. On top of that, we actually don't really know the design values of this lamp. I just guessed them into the formulas.
Also, keep in mind that power is NOT going to tell you lumens, unless you also know the CCT or the life-rating. In other words, the harder you drive a lamp, the shorter the life and the greater the efficiency. At 3300 CCT, you get about 30 lumens/watt. Each 100 CCT up or down adds or subtracts about 3 lumens/watt. The practical maximum you can achieve is somewhere around 37 lumens per watt. More than that and the filament is so close to melting that it is too fragile and too short lived to be useful.
So, short answer: MN61 brighter because it is being driven harder than stock, equates to more power coupled with greater efficiency.