Thanks for the correction. I’m certainly not one to be writing about this stuff with correct terminology. I just say what I think sounds right. 
The emitter insulation gaskets actually help the light not to drift into the gap between the emitter and reflector. I don’t think I’ll be replacing the board with a copper one or adding copper in between to raise the emitter height. I just don’t have the skills to do a project like that yet. Either way, I’m loving the added lumens by this “easy” mod. Plus the host is great for dissipating heat with all those heat sink fins. Really liking this light now.
Also, I did a quick read from the tailcap using my unmodded DMM. The only thing I was able to do was compare it to the readings from my Convoy C8. Here’s what I got:
Convoy - high, mid, low - .95A, 0.39A, 0.04A
Roche - high - 2.22A, 1.05A, 0.25A
I KNOW the C8 is using a driver that outputs 2.8A to the LED… so what does that mean I’m getting with the Roche on high? Maybe my numbers are wrong??
QX9920 Buck controller
Resistors - R200, R300, R330, R330
2x18650s
Should I be worried about damaging the LED?
It doesn’t mean anything. The point of the excercise was not to collect accurate measurements, but to see what the behavior was when you get “no output” cycling the light from ON/HIGH to OFF to ON again. Also the point was to see if that behavior manifests when you do not have the tailcap in place. (If it doesn’t then we’re blaming the switch!)
Since last night, I couldn’t reproduce the problem. I’ve turned it on and off from high mode a few dozen times and it seems to be ok. No idea what could have changed overnight. I know the numbers aren’t accurate because my leads are cheap and thin, so as long as you say there’s nothing to worry about, I can rest easy.
The numbers are probably accurate for the buck driver (not voltage drop sensitive, lots of headroom above dropout), but are certainly 100% inaccurate for the linear driver in your C8. The driver in your C8 just doesn’t have enough headroom to operate properly with your unmodded DMM in the way.
I’m glad the problem went away. Post up again if it comes back and we’ll take a closer look.
Quick question. Since this is a modded flashlight and the led is being driven fairly hard, is it safe to use my Samsung ICR unprotected 18650s? Of all my 18650s, these are my highest quality, but they’re also unprotected.
I would only do it if they are matched cells:
-same cell
(-same batch)
-same charge/disscharge cycles
-same age
-have the same voltage when discharged simiutaniosly
That means they usually have to be used in the same configuration (2 cell, 3 cell, 4 cell)-light all the time.
I bought them at the same time from the same store and have only ever used them with the Roche. So I think that qualifies nicely. :bigsmile: how do you know the batch?
I’m just guessing, that if you order them together from one seller you should most likely get cells that are from the same batch/age.
But basically those rules also apply to protected cells - IMO. I do have some sets of matched cells for every series/parallel configuration, marked with a lableprinter.
I tend to use safer unprotected cells lately (real IMR/LiMn chemestry), at least they wont blow up if something goes terribly wrong.
On/Off problem came back and it was the same: has trouble turning back on when turned off on high mode. I turned it on using my DMM and there was no problem. I did about a dozen or so times and it was fine. Is it possible that the switch has a hard time dealing with the increased output after the mod?
I think so, yes. Have you disassembled the tailcap? It’s somewhat common to see tailcap design/construction which is lacking. I like to see the switch mounted on a PCB which is then clamped in place with a retaining ring. In that case the electrical path is SPRING > PCB > SWITCH > PCB > BODY and the PCB > BODY joint is held with clamping force by the retaining ring. In that case the spring is soldered to the PCB.
In other cases I’ve seen something involving no solder with press fits involved. More like BRASS POST > SPRING > SWITCH TAB > SWITCH > SWITCH TAB > BODY. In that case you will likely find no solder joint between any of those things, and no clamping force holding the switch tab against the body, just a press fit.
IOW please take apart the tailcap, show us pictures, and think about what you are seeing.
I’ll post pictures when I can but when I disassembled it, the layout was brass post > spring > pcb > switch, all held in with a plastic washer type thing and a metal retaining ring. The switch is soldered onto the pcb and looks kind of like this kind of switch http://www.fasttech.com/products/1610/10001865/1135000-15a-250v-reverse-clicky-switches-2-pack
It might not be the same brand/quality, but it looked like that.
Let’s see what the pics reveal. Really I’d say that the construction sounds OK to me, but maybe the switch itself is the issue if it’s not a real Omten. (Even if it is a real Omten, it could be defective.)
Remember, the current draw through the switch is actually pretty low, so this should be no trouble…
After playing with the switch some more, and also doing testing with the DMM, I deduced that it must have been either my soldering job or somehow having 2 more sense resistors stacked on the existing one. I came across the same problem when I turned the light on or off with the DMM (seemed like the problem was getting worse) in any mode. So I decided to take off the new resistors and solder on a new single R330 and cleaned everything up as much as I could. The result so far is great. The light is still significantly brighter than it was stock, and it hasn’t had any problem turning on at all after a whole night of using it.
LS01, is this the same? I bought this shell from Convoy.
yeah, that looks to be the same host. Do you have the link to that? Can’t find it.
Do you use a buying agent?
I gather that I’m suppose to be able to purchase without one now… I tried to select 海外 but the shipping showed the same as Chinese destinations.
I’m a Chinese
Whole light:
Host:
I have the one from BangGood - but modded with a MT-G2.
The two Head-parts are glued together, so I would go for the host.