Not always the same VF numbers — look at http://www.luxeonstar.com/assets/downloads/DS68.pdf and page down to the “Electrical Characteristics” list (page 7)
There are three different part numbers each for “red” and “red-orange” and two different part numbers for “amber” — each part number has different forward voltage numbers listed, for the same color emitter. So getting three colors that should be identical electrically means selecting only the part numbers I listed.
You are right it is better to pick those who are closest to each other but I don’t think it is a big problem, i believe the excess energy will burn off in the emitter as heat. Also the ones i have seen for sale have been the ones with higher Vf.
And at start they will get 4.2v (not accounting for Volt sag) from the cell and the excess will burn of as heat there also. I have used a red XP-G2 on a sinkpad running straight from the cell (a Efest IMR10400) and the only side effect besides a lot of heat was the massive output :bigsmile:
Thanks Richard for suggesting it is possible to use different 7135 groups, i wanted to use the rebel lime also but i couldn’t figure out how to limit its output so it wouldn’t overpower the mix, but that could work. I had been thinking i needed to use a resistor at the bridge on the star to limit the lime one. I still thing that is a good possibility for some control, but then again it might be based on wrong assumptions.
I am going to have to eat my words a bit about what I said before about the BLF A6 firmware. While I still don't think it is for everyone, I do think it is really cool. I put it on a flashlight and have been playing with it for a few days now and I think that it's pretty great. A lot of my prior problems were due to me having the wrong setup when I flashed it before, but TK helped me get that all sorted out. Thank you ToyKeeper for all of your hard work! 
So — would it be possible to fix up a ToyKeeper driver to run the three electrically matched colored emitters Cajampa and I are interested in?
I’m not up to doing this but I could pay for the work, it’s an area I think widely enough interesting to be worth doing.
I can mail you two of those triples (red, red-orange, and amber Rebels on a 20mm Sinkpad2) if you’d be interested in building two flashlights, at my expense of course. We could pass one around to Cajampa and whoever else might be interested in finding out how it works for them, if anyone else is trying the “low-or-no-blue” evening lights to help with sleep.
The question Cajampa raised is whether having three somewhat different Rebel color emitters is going to give better color vision than the single PC Amber Rebel emitter. (The phosphor in the PC Amber gives a typical wide bell curve between 510nm and 710nm, peaking around the amber 590
to be compared to the coverage from the red, red-orange, and amber standard emitters
Alternatively, you could build a light with the phosphor-pumped PC-amber and PC-lime emitters — those have different electrical requirements and would need tapping off pins as you describe — both ought to be usable for light that doesn’t disrupt sleep, I think.
—- maybe a quad board with two of each?
_
People vary* (variations in DNA and repetitions of genes make variations in proteins — for the visual system that can make slight variations in the receptor molecules between one person and another) — but broadly speaking it’s the 440-500nm wavelength band that’s tied to the sleep system — http://ursula.chem.yale.edu/~batista/publications/ShivaJACS.pdf
_
- Aside: gene variations multiply when a gene is on both copies of a chromosome, if both copies are active. For those people who have two X chromosomes (so two different variations on those genes operating to make receptors) a rare few have different enough genes to build four rather than three peaks in the spectrum of their color receptors — look up Tetrachromacy - Wikipedia
(People also vary a lot with age as the color of the natural lens becomes more yellowish: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.278.764&rep=rep1&type=pdf)
Extensive summary of what’s known: Protecting the Melatonin Rhythm through Circadian Healthy Light Exposure - PMC
Yeah, i am not an envious person for anything………but Tetrachromacy sounds extremely special.
It is just a reality extremely few of us humans can ever experience & from the description in interviews of those women that have it it sounds absolutely AMAZING.
But then i am a guy who really enjoys nature and the delicate shades of patterns that make up our natures visual experience in flowers, foliage & the changing shades of the forest, so the possibility that i am missing out of most of them is a bit sad.
The horrible backside for the women who are blessed with Tetrachromacy, is that the gene mutation that gives them 4 peaks and a vision of up 100,000,000 different shades of colors, also rob many of the sons & men in there genetic lineage of the color vision 
hank, i did not know the melanopsin effected limit only went to 500nm, i have always thought it was 550nm. But you have probably checked more research than me, but i have to ask is the 550nm limit wrong then?
Because i was a bit disappointed in the rebel lime, i thought it was just a strong peak of 20nm wide, but it was a PC led with a 100nm wide peak band.
Do you know if there exists any colour led’s in 550nm or higher but still green, but that is not a phosphor led but a single thin peak like the red/amber and such led’s? I have though for months that the lime was the solution but i have reconsider that i think. But then again it doesn’t hurt to experiment with the lime if that is all there is in the top end of green emitter, it really should improve color vision a lot. And lower the barrier for entry for the less health conscious of us who want to try it out.
EDIT I wonder if not the PC amber + red-orange + red or swapping the red-orange for the deep red, because the red-orange is so close to the red one. One good reason for doubling up on red & red-orange could be that because the eyes sensitivity drops so much for the higher nm it might make the light more even. For example my setup with amber & red ledstrips, gives a bit of an amber push and i think i will experiment with adding more reds to compensate. Need to order more strips first though.
Another possibility if the limit really is 500nm, is to use the lime + PC amber + one red or preferable as many of the red spectra led’s you can fit. That should give the widest possible color above 500nm and should be more pleasing in day to day work the last few hours before bed.
In my last order I ordered a few of the new 3XP Sinkpad MCPCBs. I could get them a little bit cheaper than the Noctigons, and they are already configured in parallel, so it would save me the work or bridging or adding jumpers. I tried a few out last night, and I am not impressed.
Here are the cons compared to the 3XP Noctigon, from least important to most important:
- They don't look as cool.
The Red Noctigons look great underneath the optics. - They aren't gold plated. This makes the solder not flow as well, and the backside tends to oxidize during reflow.
- They are too big. They measure 20.6mm maximum diameter, which is too big to fit in an S2 or S2+. I sanded mine down so it would fit, not a huge deal, but another step with the potential for a screwup. The Noctigon is 19.7mm maximum diameter.
- The center hole is too small. You can't fit 22 AWG silicone wire in there, and 20 AWG Teflon is a very tight fit. Drilling out the hole then deburring on each one is a show stopper.
- The traces are weak. I put a triple 219C together last night with an FET driver, turned it on, then a few seconds later it went dark and I smelled that nice burn smell. I figured that I had screwed up and shorted something, but no, the trace burned up! If you do use these, use the smaller pads instead of the big ones, since they seem to be able to handle more current. This probably won't be an issue for regulated lights or maybe even triple XP-G2 S4s with an FET driver, but it is obvious that the traces are much smaller. The Noctigon uses 3oz. huge copper traces.
In summary: I think that while it was a valiant initial effort, the Noctigon 20mm 3XP MCPCB clearly outshines the Sinkpad for flashlight use.
Burned Trace
Thanks for testing Richard, very good to know i will scratch the sinkpad 2 based triple option of my list and just use the good old noctigons instead 
And hank, i read the research you posted & yes over 500nm seems to be no problem for melanopsin, nice that means we can use the whole green spectrum also and potentially make much nicer looking after dark lights.
But i have to say i really enjoy dimming down the red/amber ledstrips lower & lower when it gets closer to bed, it really give that going to bed in the glow of embers feel & it feels very natural & a pleasant way to end the day.
Cajampa, see http://ursula.chem.yale.edu/~batista/publications/ShivaJACS.pdf — a few years old, I haven’t read all the citing papers to see what’s newer. There’s no specific number applicable to everyone (or every species, for that matter, they’re comparing mice, squid, people, cows …).
“λmax” I think refers to the peak, and there’d be a decreasing sensitivity above and below wherever the peak is, so yeah, it could well go into that 500-and-above range.
I’m trying a few of the PC-lime emitters at night now to see if they make us feel wakeful or not. I’d guess it’s going to vary with age and each person.
I was looking to get a FET driver for a JAX-MTG2 host I never got around to assembling and I wonder if it’s possible to get the 22mm FET driver with ToyKeeper’s BLF EE A6 firmware and then make things more challenging by being able to run it in H-M-L with memory disabled?
Yes we could do that. Right now I just have the FET 22mm, but I have some prototype 22mm + 7135 on the way also if you wanted to wait a few weeks. That said, we can have the same functionality with just the FET, just without as low of a moonlight mode.
Ok, i missed that they where peaks, so it could be very possible be some effects higher up then. Ok back to considering 550nm a good safe limit again then
The Protecting the Melatonin Rhythm through Circadian Healthy Light Exposure - PMC had an interesting test on a blind man “fig .8” where they tested 555nm and could found no at least claimed effect on melatonine production. But wonder how they produced an exact 555nm light? Maybe a diffused laser or a filter or something.
To bad there is so little available human tests, it should be very easy to test melatonine level & production to the end of the sleep cycle under various light frequencys, maybe not so easy to construct a lot of various peaked light sources though. But somewhere in the 500-550nm range, it probably is and as you say it probably varies from individual to individual anyway.
RMM any idea when you will be getting some more Nanjg 101-AK-A1 drivers in?
Wanting to build a green and/or red UF-T20 to hunt with. Bought those drivers and LEDs from you before and they are working good.
If you are not going to be getting them any time soon, would the 1.52A 17mm driver you have listed on eBay work as good?
That UF-T20, do I understand right that the C8 pill will fit in it? I'd like to be able to swap red, green, and white LEDs into the same host as needed.
I have a question concerning the use of low vf LEDs like the 219C and 71235 drivers, I don’t have clear understanding of how these drivers deals with the volt differences.
I want to build a triple nichia 219C with a 12x7135 driver, if I run it in med mode with a high drain cell and the LED remains with a low voltage, does the driver “consume” the excess power? And what happens in the case of a fet driver when the vF and battery V is about 0.8V difference?
Thanks!
It will still be about 7-10 days. Yes, that driver will work. It is just the qlite with 4 chips removed (you can order it from the site for less money). I am not sure about the C8 pills, since there are so many different ones.
The 7135 is a linear regulator, so it operates under the same principal as any other linear regulator, which is that it is basically a variable resistor.
So, at 100% PWM the difference in voltage needed for the given output current is burned off as heat. The greater the difference, the more heat is generated. With a 1S cell situation and a 3V LED, the difference isn't generally that large, so the efficiency isn't that bad, especially since the cells sag voltage a bit. At a medium 33% mode, you are just pulsing the 100% power, so the dissipation isn't that bad. If you were using an analog dimming, where you vary the constant current amount, then the losses would be quite high on the lower modes.
With the FET driver only the FET and driver resistances are doing anything. Usually, you get about 0.005-0.010 ohms of resistance through a good FET driver, depending on the current draw, FET gate voltage, temperature, etc. The batteries usually sag enough under load to keep the current within the safe zone for most LEDs on a DTP. The 219C will pull a ton of amps with an FET driver, but the LED can handle a ton of amps so it works.
Thanks for the heads up on price. I saw the removal chip add/remove option, but didn't pay enough attention to the prices.
I misplaced my main light (Solarforce L2N) somewhere and haven't been able to find it for about a week, going to order some parts to build another. It'll probably reappear as soon as I get the new one put together, but never hurts to have a spare (or three), right? Said last year that I was going to try the T20 and never did, figure since I'm ordering now is as good a time as any.
I'll be placing an order soon!
Memory is simple, but changing the mode order isn’t trivial. I think the simplest way to change it would be to reverse the actions of short and medium press, and make long-press reset to the last/highest mode instead of the first. That way, it still uses in L-to-H order internally for LVP and turbo step-down. But I may be forgetting other areas where it has hardcoded assumptions which could cause issues.
However, the code I’m working on now allows either mode order, among other things. It doesn’t care about mode order since it uses a completely separate mechanism for LVP/turbo/thermal adjustments. The catch is it doesn’t fit on a tiny13, so it needs a tiny25 or bigger.
Thanks Richard! Exactly what I wanted to know.
I couldn’t tell a tiny13 from a tiny25 if my life depended on it. 
I imagine a tiny25 adds as many problems as it resolves if past experiences hold true. Assuming the chip is bigger because of the bigger number, would I have to get a board designed around a tiny25 with different traces etc or can a tiny25 be plopped in place of a tiny 13?
One of the advantages of the 25 vs the 45/85 is that the 25 is available in the exact same package size as the 13a we normally use. Plug & Play.