Across 4 measurements by KB, post #108 here, it certainly does show consistent rise in output from 15A to 20A. I should get some more data points in this range as well. I only tested at 11A and 18A here at post #94.
I will just give up for the moment, a proper re-test would require a new $16 dollar led that takes weeks to arrive, soldering a fresh ledboard onto a 30mm copper disc and mount that on my test-rig.
In the meantime I guess that over many people combined enough knowledge is gathered on this led that a test has little to add anymore.
KB mentioned his lux measurements were with bypassed tailcap, so the amp values associated with those lux measurements could be 1-4A lower, depending on the tailswitch resistance (ranging from 0.01-0.035mOhms). So the lux measurements could possibly be over 12-16A, not 15-20A. This could potentially explain a big part of the discrepancy with djozz’s measurements.
I get higher lumens measurements with the tailcap on consistently over the bypass wire, 3 times out of 3 with 3 different sets of cells. Not by much but between 100 and 200 lumens. I use a 14 AWG bypass wire, 8" long between the batt- and the end edge of the tube.
Good switches are pretty efficient and you have a stronger ground contact in the assembled, threaded tailcap, plus a 20 AWG bypass for the spring.
Interesting, thanks for the measurements. But they could be interpreted in different ways. It could be that you are actually getting higher current with the tailcap. Or it could be that you are getting less current with the tailcap, and the output is higher because you are in the downward slope portion of the output curve.
I just did a quick measurement of contact resistance of a 12 gauge wire pressed on the edge of a 1.5mm thick copper disk, and with just a bit of pressure it was below 1mOhm. So unless your 14 AWG wire end is unclean the contact resistance is very low. The 8” 14 AWG wire has about 1.7mOhm by itself.
I have measured bypassed-spring tailcap and switch assemblies and they have ranged from 10mOhms to 35 mOhms.
I have solder blobs at each end of the 14 AWG wire. Dunno, could be some degradation, I've had it happen before on jumper wires - resolder, and get higher readings. This one has been used a lot.
I hear the theory but in practice, I've seen over dozens and dozens, maybe 100's of measurements where the tailcap assembly does better, pretty much every time. Granted many don't have a switch (e-switch lights with no tail switch). So maybe my wire ends are beat up. I got plenty of wire around, so maybe I'll make up a clean new 12 AWG jumper and compare.
I'm surprised the surface contact area makes no difference in the formula? With the tailcap you get the full circumference of the tube making contact even with an anodized threaded tail, while the wire has a tiny fraction of that for contact.
Update: I just cut a 10 AWG stranded coated wire (all I use is stranded), 6", soldered the ends, cleaned up the ends with isop. alcohol - same result: jumper wire has lower lumens readings every time compared to the L6 switch assembly - repeatable, went back and forth a few times. Not much of a difference but I'd say about 60 lumens.
The way I measured the tail cap resistance was by passing a known current from the (bypassed) spring to the point where it would make contact with the battery tube. Then I measured the voltage at these two points to find the resistance. So any contact resistance between the tail cap and battery tube would be in addition to that resistance, but I think the contact resistance would be very small.
With your updated measurements I would say there is almost certainly more current flowing when you bypass with the AWG10 wire. There is just no way there is more resistance (contact resistance included) than in the tailcap. If that is true then that means you are at or near the maximum in the output curve.
Could you try the same measurements (with tailswitch and with AWG10 bypass) with a single XPL light? In this case we would be on the upward side of the output curve, and you should see more output with the bypass. (If my understanding is correct)
Ok, tested the 10 AWG wire across 3 different FET based lights, and again with the L6 at lower amps, about 12A. On the L6 the wire makes little difference. Have you seen an L6 switch? It's huge! Believe it's on of those 6A rated switches.
On the other lights:
- U21 w/dedomed XPL, no tail switch: wire does slightly better (~7 lumens out of 1,600)
- Warsun X60 w/XPL2, tail switch: wire does much better (~190 lumens out of ~2,000)
- Lumintop SD26 w/XPL2, no tail switch: wire does slightly better (~30 lumens out of ~2,100)
- L6 using EVVA cells, about 12A @tail: it's about a draw over 3 trial tests, all within 1% one way or the other (~7,300 lumens)
So for a bad switch, like the X60, can be a big difference. The L6 has one of the best switches we have available to us though.
All the more reason to get rid of the tail switch 
. If I get around to upgrading the X60, I'll look at eliminating the tail switch - don't need it with NarsilM.
You can simply bypass the switch like we do the springs. Or just bypass the spring directly to the outgoing side of the switch.
I modded a couple X60's for BLFers, but not sure what I did - probably swapped out the switch - that's what I usually do when I see a significant drop from it. My own stuff though always gets lowest priority - I never seem to get back to my own to improve it.
I don’t think a tail switch would matter though, current is the same everywhere in a circuit, if he measures 20A at the switch then he would measure 20A at the LED too.
It is the voltage that changes between components in series.
Thanks for doing the measurements, Tom.
The issue is that the circuit is altered when the current is measured. The tailcap and switch are removed and replaced with the current meter/shunt.
You’re right, that’s why I use a clamp meter, not sure what those guys used to measure.
clamp meter - UT210E
I think all the serious testers here use clamp meters (UT210E mostly) and wire. Anyone still running the circuit through a multimeter should know that they will never get accurate results (above a few amps at least).
The issue I’ve been talking about here is that the bypass used with the clamp meter can have less resistance than the tailcap/switch, so the current measurement can be inflated relative to what will actually flow with the tail cap in place. This is the opposite effect of what we see when people use ammeters with high resistance leads.
Yes and this is accepted.
Just like people using homemade “lumen tubes” as opposed to calibrated light spheres. You can only get so much accuracy because we are hobbiests, not scientists with a lab full of expensive testing gear.
So when it comes to factoring in the resistance of the rear spring and switch, I say as long as the spring has a bypass wire on it, a wire and clamp meter should not inflate the output by very much if at all. Maybe a tiny bit. This is the best we can do as hobbiests.
If you want to take it to the next level you’ll need to buy an assortment of common switches and springs and test equipment to put various loads through them and measure the output at various levels of spring compression and chart the resistance levels as more current is applied. You may also have to account for switch wear due the contacts creating arcs every time they touch and over time adding more resistance.
I think that is a bit overkill.
The context here is just that I was trying to identify any factors that might contribute to discrepancies between the output/current trends observed by djozz and those observed by the couple users who have measured the XHP70.2 in flashlights.
In this case it looks like the L6 switch is low resistance and so it is probably not the source of a big difference.
I don’t think that is a concern here, the point is that those guys did their measurements and found an increase in output from 15 to 20 amps while DJozz did not.
So whether the current is the same with a switch or not is not really relevant to the fact that there was an increase from 15 to 20A.
UNLESS they measured current and measured intensity at different times, which honestly wouldn’t make much sense.
Then you’re right, if they measured the intensity with the flashlight in normal condition and then used a bypass for measuring current then there could definitely be a difference in performance due to current.
Yes, this is what happened, as I noted when I first talked about this issue on the previous page.