Thanks a thick contact ring, should have known that you were going to say that 
There seems to be a wish to process on the Q8 with TF, the engineer even emailed on Saturday so let’s hope things can go faster from now on.
On a side note, yes using less then 4 cells still feels special for me in a 4P light, oh how I hated toys and things needing 4 batteries and only having 2 or 3 as a kid. This could be a reason for the intense feeling of joy operating a light with empty spaces in the tube 
I wired up the 22mm board from post #153 for testing - 2 long LED wires and a switch. created a new project called NarsilTriple, and created 7 mode sets, from 1 mode to 7 modes for supporting the triple PWM channels. I made the changes, in theory, to support the 3 output PWM channels like bistro_tripledown does, and updated all the uses of setting outputs for blinks, etc. to the new 3 channel setup.
For ramping, I cheated for now -- zeroed out the ramping table for the 7135 bank. Mainly I just need want to see if regular mode sets work with the new 3 PWM support. Just programmed the 85, gonna go try it now. Wish me luck!
Good luck! Can’t wait to see this come to reality.
This would complete my goals for these drivers as a do anything FET based driver. Although I do think I would add another 22mm+ version with an LDO for the larger boards, plus finish up the 46mm version.
So basically it would handle any flashlight up to a 30mm driver, Clicky, e-switch or dual switch, regulated and FET modes, thermal regulation (on the clicky for now at least), zener for use in multi cell lights and LDO on the larger drivers.
I am quite happy with how these are turning out.
O-K, it basically works! 
Wasn't bad - first time. Well actually, the first time it was flaky bad, looked over the code cleaned up some things, but mostly comments, naming, etc. Then went to dnld again, and dang -- it hit me! I downloaded the old Narsil, not NarsilTriple. I forgot to create a new BAT file to do the triple download... 
So, I think what I proved;
Didn't try PWM's on the bank yet because only one mode set uses it - the 7 mode group - easy to try though. Current mode sets are below. I roughly assume full single 7135 is about 10%, and the full bank of 8 7135's is 50%. This makes sense if you have about 1,500 lumens max on a full FET, and the bank of 7135's is about 750 lumens. I've have to gen'ing a ramping table for 3 channels. This might be some trial&error til I get one that looks good - need to test in a real light.
I totally disabled (commented out) the on-board LED support for now.
// 1 mode (max) max
PROGMEM const byte mode7135Set1[] = { 0}; // for single 7135
PROGMEM const byte mode7135sSet1[] ={ 0}; // for 7135 bank
PROGMEM const byte modeFetSet1[] = { 255}; // FET only
// 2 modes (7135-FET) ~10% max
PROGMEM const byte mode7135Set2[] = { 255, 0};
PROGMEM const byte mode7135sSet2[] ={ 0, 0};
PROGMEM const byte modeFetSet2[] = { 0, 255};
// 3 modes (7135-7135s-max) ~10% ~50% max
PROGMEM const byte mode7135Set3[] = { 255, 0, 0};
PROGMEM const byte mode7135sSet3[] ={ 0, 255, 0};
PROGMEM const byte modeFetSet3[] = { 0, 0, 255};
// 4 modes (1.2-10-50-max) ~1.2% ~10% ~50% max
PROGMEM const byte mode7135Set4[] = { 30, 255, 0, 0};
PROGMEM const byte mode7135sSet4[] ={ 0, 0, 255, 0};
PROGMEM const byte modeFetSet4[] = { 0, 0, 0, 255};
// 5 modes (1.2-5-10-50-max) ~1.2% ~5% ~10% ~50% max
PROGMEM const byte mode7135Set5[] = { 30, 120, 255, 0, 0};
PROGMEM const byte mode7135sSet5[] ={ 0, 0, 0, 255, 0};
PROGMEM const byte modeFetSet5[] = { 0, 0, 0, 0, 255};
// 6 modes 0.8-2-5-10-50-max ~0.8% ~2% ~5% ~10% ~50% max
PROGMEM const byte mode7135Set6[] = { 20, 110, 255, 255, 0, 0};
PROGMEM const byte mode7135sSet6[] ={ 0, 0, 0, 0, 255, 0};
PROGMEM const byte modeFetSet6[] = { 0, 0, 0, 0, 0, 255};
// 7 modes (0.5-2.5-5-10-25-50-max) ~0.5% ~2.5% ~5% ~10% ~25% ~50% max
PROGMEM const byte mode7135Set7[] = { 12, 63, 150, 255, 255, 0, 0};
PROGMEM const byte mode7135sSet7[] ={ 0, 0, 0, 0, 120, 255, 0};
PROGMEM const byte modeFetSet7[] = { 0, 0, 0, 0, 0, 0, 255};
Looks like the next thing to do is combine LVP and switch on Pin 7 to get the indicator LED back on Pin 2
Hhmm.. I know Mike ?? did that, or something similar, but didn't look into it in detail. Thought he mentioned it required his board design? Do you know if anyone did this on one of our std boards? I can see where possibly there's no conflict - only do a A-D conversion to read the voltage when the switch isn't pressed, or hasn't been pressed for a while - some sort of logic like that, I'd think. Then maybe just sharing the switch with the R1/R2 pair could electrically be done?
Another option is getting rid of R1/R2 and using the internal ref, ala Dr.Jones. Or, using the Atmel dev kit, then you can use pin #1 as an I/O pin -- some options...
MikeC did it, I want to say he combined LVP, e-switch, and OTC for dual-switch all on one pin. It “required” his board to work, but I’m sure you could air-wire it to test firmware if you were motivated. All it would really take is soldering your switch+ directory to the tail of R1 or R2
May be possible. Don't think he was 100% with it all though. Recall he posted his code/boards, but don't think it was fully stable yet. Would hate to get false low voltage warnings... Though, that's exactly what happened to me last night, using a flat top WindyFire cell in my modded H15 last night - it seemed to work at first, then went totally flaky, and batt check was working but showing low voltage levels. Turned out it worked fine when I simply added a solder blob to the top of the cell...
Worth look'n it to though, for sure...
Edit: even an intermittent false warning should be tolerable since we need several low readings in a row to trigger a drop.
Very nice, glad to see this progressing!
Far as the pins go, if the LVP could be moved to the internal reference and R1/R2 removed that would be my preference for sure. That would free up a significant amount of space on the boards on top of the free pin. Plus it sounds simpler to me then trying to combine a bunch of functions into a single pin.
Not to mention it would remove issues with low tolerance components used when china starts selling drivers based on these layouts (the triple channel general layout, not necessarily The Avenger drivers), it is only a matter of time based on past dealings.
That said I have no clue what the coding involved in either change would be.
Yea, I do agree - better bang out of using the internal 1.1V ref, plus reducing parasitic drain even further. It's been briefly discussed. I know DrJones uses it, so it's possible for sure, and don't think he said much about it being a PIA or anything... I just don't know where to start to try to implement it. I'll have to do some research on it.
I also tested the 7 mode grouping and confirmed PWM's are working on the bank of 7135's. So the firmware is 100% functioning, just not using the bank of 7135's in ramping yet.
Defining modes though in percentage of output is highly dependent on the light mod/configuration. Some lights have a max FET amps of 3.5A, others will have 20A, and everything in-between. So 2.8A may be 80% or may be 20%. Mode sets should really be tailored accordingly, if you want consistency from light to light. Ramping has a little more tolerance, but should still be tweaked for exaggerated cases. I'm using Narsil in a MtnE SRK 7135 driver, and it uses 380 mA 7135's, so the standard Narsil ramping table has a couple OFF levels in the transition from single 7135 to the bank of 7135's because 380's can't be run at the lower PWM settings, like 350's can be. If you like a low moon, should always use a 350 mA 7135 - I've been seeing even a PWM value of 1 works with the new design drivers, but suspect it will go OFF when the cell gets low.
Now I need to find an E-switch light with which to try one of these in…
O-L showed a way to convert the Mag-Lite switch to an e-switch. It could probably be done with any side-switch light. Basically just bypass the switch for power, connect it instead to the e-switch pads on the new driver you install, and disable the latching part of the clicky mechanism. In case you don’t have access to an e-switch light you’re ready to mod, maybe you have a side-clicky you’re less fond of. :innocent:
Yeah, I figured it would not be hard to use any side switch but I don’t really own any. I have nothing against them, just never seem to find one that peaks my interest enough to cause me to buy one over a clicky version of the same thing.
I have some maglights but lack of heat sinks for them means they still use the stock incan bulbs and sit at the bottom of the box of “to be modded” lights.
The only side switch lights I have at the moment are SRK style lights, L6, S70, NE01, DO1 and thats it that I can remember unless something is hiding in my box of spare lights.
Another reason for my lack of side switch lights is they are generally pricier then clicky and harder to mod. I actually won a U21 for $35 on ebay a few days ago but from the little I saw of the driver it doesn’t appear to be at all mod friendly, plus it needs a boost driver, so most likely I will leave it stock unless I can get a resistor mod to work.
There has also been little in the way of good options for hardware and firmware for e-switches till recently as well.
I am sure I will correct my lack of e-switch lights before long, just a matter of finding one that is cheap, interesting and worth buying. Or I might make up a driver for the DO1 and try it on that.
A JM35 is something I was considering while they are $22 at banggood but it once again needs a boost driver and with that shallow reflector I don’t think it would take well to a 3v emitter.
Say wut?? I luv my U21, fully modded running Narsil, using the RED LED under the button cover as a locator LED, and luv the dragon eye pattern.
Dang! If I would have seen that eBay listing, you would have had some bidding competition
I have a dedomed XPL V6 0D in it, does about 6 amps, 1,700 lumens, 240 kcd. Much better than the stock XHP35 HI. The Liitakola 26650 is made for it.
It's my favorite light at the moment. Was think'n - $65 is steep to pay for a host, otherwise I'd invest in a few, mod them, sell them...
The lack of a tail switch saves in length of the light big time. Has great heat sinking built in, no need to mod that. Perfect large pocket thrower.
Yeah, the U21 was the first side switch to catch my eye in awhile which is why I nabbed that one, I figured for that price I could hardly go wrong. That said I like the idea of the XHP35 in it and if I can resistor mod the driver I think that would be the best option for that light.
The JM35 is a host I would get if it didn’t need a boost driver, it looks kinda silly but the price is right. With an XP-L I just don’t think it would have a worthwhile beam pattern for the size.
There are just not that many worthwhile side clicky hosts out there that I have seen.
I am also looking for a good, easy to mod side switch 18650 host.
Just found out about Convoy L4, it uses a 23.5mm driver and 20mm MCPCB.
http://flashlightreviews.xyz/convoy-l4/
Unfortunately the coupon from m4dm4x expired
Ohhh, I got a ton of e-switch lights, turned hosts. Can't buy any hosts per say, but mostly you don't want to anyway - on most you need the stock driver as a contact board and switch mount. My favorite budget host is hard to find now - the ZY-T11 clones, I paid like $9-$11 for: nice compact size, throwy wide head/reflector - perfect 18650 EDC for some throw.
The SkilHunt H15 turned out real nice - like that the switch is on top, w/LED using as a locator LED, etc. The H15 and the U21 are the two lights I either have on me, or are close by.
Bunch of ones that worked out great w/mods:
I know I'm forgetting a few more...
I've modded the Convoy L4 - nice C8 size/quality, but wayyy too long.
The easiest to mod hosts are the SupFire M2-Z and L6 L5 (ooops!), because they got driver retainer rings, nice SRK style mounted switch in the side - easy to wire up. No need to piggyback a driver in. Might be more like that, if you have the right size driver. My favorite easiest is the L6 right now.
Convoy L4 with a LEGO fit SS bezel:
L4 driver mod:
New Supfire L5 also has a driver retainer ring.
Got the 17mm drivers today - they took longer, different panel. Wow, sure looks packed in there. Will take a closer look - should work well for the piggyback e-switch setups. Need to work on the ramping table...
Discussed the issue of dropping R1/R2 to my EE buddy at work, and using the ref voltage. He's not familiar much with the Atmels, uses the PIC's mostly, but said that is entirely possible to do, couple different ways in fact, depending on what the Atmel has.
He's still adamant that we pursue optical programming/dnld of the drivers. He believes the LED can be used to read light as well as transmit it, and we could use it to both set config settings, and do a full download from a cell phone. Other lights on the market do this, but not sure they are using the primary LED for this.