My discussion concerning run time here will revolve around the XP-L2 emitter (used in the H53c and H53w) compared to the XM-L2 of the former generation (used in the H52w).
Total gains in luminous power are
9% going from the XM-L2 to the XP-L, and
7.7% going from the XP-L to the XP-L2. Taken together there is a leap of 17.393%. Since there is no H52c, I will compare the "w" versions between the two generations. The H52w had a max output of 280 lm, and the newer H53w has an output of 330 lumens--a 17.86% increase in power. That seems consistent.
I made OTF efficacy vs. OTF Output regressions between the H52w and H53w to demonstrate the increase. Unfortunately Zebralight seems to have had some problems with a couple data points that are way out of range with the expected distribution. One seems to be due to a typo, a confusion between 17 hours run time and 12 hours run time, and the other is possibly due to overheating or a general step down shortening the run time.
Plotted (at FooPlot) are
Efficacy (lm/W) verses Output (lumens) for the
red data points of the H52w and the
pink data points of the H53w. The red curve is a regression with the corrected 12/17 hour run time data point. The pink curve is simply a multiplication of the red curve by 1.25. That is fairly consistent with the 18% difference in efficacy we expect from the two different LEDs.
The black and blue curves are merely relaxed regressions that were derived with the errant/deviant points maintained in each set of data. Regression allows a scientist to form curves through scattered data to find a generalized trend. Scatter can be the result of poor experimental procedure or variables that the scientist isn't interested in tracking.
You can turn off the relaxed regressions by deleting them. You can change colors of the data points and curves.
This plot makes the curves and data more clear. You can now see how far off ZL missed the 330 lm output data point of the H53w as well as the 25 lm output data point of the H52w.
If you want run time at any particular point on each curve (pink and red curve), use the Fooplot point tool (point mouse over graph and select point on line icon, then click a point on a curve) to find values of interest on each curve. Then multiply that number by 3.06 (for Eneloop Pro capacity) or 2.4 (for Eneloop capacity). Then divide by the output lumens to get run time. (see equations at bottom of this post).
Conclusions: ZL raised the max output between generations 18% with no little or no run time gain. Had the outputs always remained the same between generations, and similar batteries are compared, you would expect to see an 18% increase in run time. If any outputs are close, you can expect a run time increase close to 18%. For example: looking at the 12 lumen output on both data sets (the only equal output), the compared efficacies are 161/135 = 19.3% increase (translates to increased run time), which is close to the output improvement of 17.4% between generations.
Note: It is unclear if the same rules were observed by ZL in determining run time we observed in both headlamp tests.
The following data were used to compile curves:
ZL-H53w 4500K Eneloop Pro AA (1.2x2550 Capacity)
(OTF lm, hours, OTF lumen-hours, OTF lumen/watt)
330 0.9 297 97.1 THIS WAS OFF. Probably (but maybe not) due to a thermal or general step down shortening run time. Ignoring temperature or battery performance, the OTF efficacy would probably be around 125-140
275 1.6 440 144
198 2.3 455.4 149
122 4.2 512.4 167
65 8.5 552.5 181
30 21 630 206
12 41 492 161
1.2 13x24 374.4 122
0.31 1.1x30x24 245.52 80.2
0.07 2.2x30x24 110.88 36.2
0.013 3.3x30x24 30.888 10.1
ZL-H52w 4400K Eneloop AA (1.2x2000 Capacity)
(OTF lm, hours, OTF lumen-hours, OTF lumen/watt)
280 0.9 252 105
172 1.7 292.4 122
108 3 324 135
50 7.5 375 156
25 12 300 125 THIS WAS OFF Should be around 172 lm/W with 17 hours of run time (not 12)-- probably a ZL misread typo between 12 and 17.
12 27 324 135
2.7 4x24 259.2 108
0.34 3x7x24 171.36 71.4
0.06 2x30x24 86.4 36.0
0.01 3x30x24 21.6 9.0
Run Time Equations
HOURS = (OTF lm/W)*(3.06) / OUTPUT LM (Eneloop PROs)
HOURS = (OTF lm/W)*(2.40) / OUTPUT LM (Eneloop)