Absolutely. The 7135 is a great companion.
Continuing where the OP left off, if you do put too many 7135s, all of your modes will suffer dimming, even moonlight. This is because PWM is just a percentage ON vs OFF. So you are running full volts even in moonlight. As those volts fall your output does too. So if you have too many 7135s, your moonlight will probably not work worth a darn as the battery voltage reduces. You need to give the 7135 regulators something to regulate 
That is probably worth repeating… Your regulators need some room to regulate. Or there will be dimming.
um, this is news? I thought it was common knowledge on here that 7135s are current limiters not current regulators. The limit the current that can be drawn - when the LED is trying to draw more current at a voltage that is higher than the battery can supply that current, it will not. It's pretty simple. Solutions are pretty simple too - add more cells in parallel or use a higher drain cell.
The PWM thing with multiple 7135s is a good point, but one that several members on here have addressed with new 7135 based drivers, where single or small numbers of 7135s are used in sequence. So 1 7135 (with PWM) at low currents, then 2, then 4, then 8 etc. Or even the 1x 7135/FET drivers.
By all accounts (not mine for sure) buck or boost drivers are exceedingly non-trivial to design, even where the light set up allows for their use. I think I've read of 2 on here in the years I've been reading this forum that have reached fruition.
This is a very diverse group. Many new comers and learners. Basically, the next generation.
ah, the OP read as if this was a newly discovered phenomenon, not a new user PSA. Might be worth asking for it to be made a sticky then, otherwise it'll fall off the front page within a few days and its purpose will be defeated.
Well I didn’t think my choice of words put it like that. It wasn’t written for a electrical engineer that’s been around the forum for 4 years.
um, yes, it did. No worries though, it's a helpful PSA and it doesn't hurt for people to be reminded about stuff. Sticky would be good though, otherwise this will somewhat be in vain in a couple of months time.
ps. I'm a geneticist :)
I’d be interested to know who among us are actual electrical engineers.
Yea I’d be interested too.
And I’m good with the thread being sticky’d. I wasn’t going for that though.
So the cool thing about the 219C, which we do have data for is the forward voltage curve. It only takes about 3.4V to run it at 4.5A. That’s about 70% of the capacity of a Sanyo UR18650FM under a 5A load.
Yes, a current-controlled buck converter is going to perform better in low modes. The driver required is larger and more complicated, with a bulky inductor on it. Lots of manufacturers and more than a few DIY lights use drivers like that, but it’s hard to find one that fits nicely in a Convoy S-series host. Being able to fit inside the host you’re using is an important characteristic for drivers.
A boost, or buck-boost driver would be ideal. It can keep the output regulated until the battery hits whatever safety cutoff voltage the firmware author has chosen. Single-cell 1.2/1.5V lights use them, as do Li-ion lights from Zebralight and Armytek. I’m not aware of any boost drivers available for higher-powered Li-ion flashlights, though some with the necessary current and voltage ratings do exist in the laser community. Those appear to be based on Ti 258x chips, and like buck drivers, do need some diodes and an inductor.
That's somewhat true, but we should stick into very crude basics here without making unnecessary assumptions, right? The topic starting post is so trivial that should it surprise someone, they might arrive at false conclusions in case we cut corners here. A 7135 only driver can be made to operate in a way that lower modes only activate some chips, even one.
But that's just a random number, isn't it? It's admittedly typically less efficient to run a larger number of parallel regulator chips, but by how much, varies by situation. The increased efficiency of linear regulation in when approaching full power setting(or when discharge deepens) theoretically mostly results from the battery voltage drop. Part of the theoretical problem might still be there even when the driver is of the switched-mode type. Depends.
A lot of the time the low mode efficiency is quite comparable no matter which simple 7135 driver is chosen. Any linear regulated circuit will waste the excess voltage as heat, but that's still a fraction of the total energy consumed. Even using a smaller or less capable cell will sometimes(a lot of the time) result in improved (linear-regulated) driver efficiency. And that's not a practical method or goal most of the time.
Yes, it’s hard to find any in such a small form factor. I built one S2+ with a buck driver, but this very example is quite different from what we’re looking for here and I admit I did not test this with a “normal” S series setup. This driver is quite affordable, needs just a simple 17mm-17mm driver space extension ring and the LED wires routed differently to fit properly, and works with 2*18350 and MT-G2. This might(should) work with a more typical config of, say, XP-L or 219 and a single 18650. Someone’s probably built one(or a bunch) already as we’re on the BLF.
It's a matter of definition if this "fits nicely". After all, the driver retaining ring could not be used, driver had to be soldered in place and and a depth extension ring added. However, the light would mechanically take even quite long protected 18650 cells. This is after modifying the reflector, so original head might be too long. But for a triple setup there would be no problems.
Here’s a tiny inductor for that buck that would fit on the smallest of boards. It has a 4.31mm x 4.7mm footprint.
76mOhm inductor..... What is that you want to do with that IC? Everything sounds just underwhelming for high currents (4A 5A) while the input is under 4.2V
Put your plan on the table don't let me guess what you have in your mind.
Hmmmm …. 
Matches the boots …
Nevermind. But let’s end here. What started as a kind way to anonymously educate some BLFers with crazy wish lists that they need to adjust their expectations… has turned into another chest pounding thread.
That ST chip is really not well suited for single cell lights driving a single LED. A LiIon cell voltage is rather close to the LED Vf voltage. With circuit a component losses/overhead you don’t have enough headroom for a buck converter to be effective. It would work if you had at least one more cell in series than you had LEDs in series.
I don't really understand what you mean, how have you gained nothing? Surely you have gained the increased output until such time as the voltage of the battery falls below that required to pull the max regulated current from the battery, at which time you'll slowly dim until you switch off the torch.
So instead of running 3 7135 at 1A 500lm, I can run 6 7135 at 2A 850lm, with no dimming due to voltage until 3.2 volts -which is perfectly fine.
I also can't see how this adversely affects moonlight mode, moonlight mode would be brighter at high voltages due to the increased current at a set pwm right? Then once I hit 3.2 volts I'd be slowly decreasing until I flatten my battery, just as I would if there were no current regulators - or do I completely misunderstand how the driver pwm works, you've got me a little confused.
EDIT: Ahh don't worry, I see where you were going, not going to stop me stacking 7135s though I want the output even if it means I get dimming under 3.2v, hell 3.3v I'm crazy 
, just confused me with the "the perfect number of 7135" line, perfect only if you want no dimming until until cutoff and are willing to take reduced lumens to get it.
Cheers
Nik
I think what the original post is trying to say is that a FET plus some small number of 7135s like the 17DDm driver is more efficient than a large number of 7135s, and that the 7135s won’t stay in regulation very long as the voltage drops, which will result in declining output in all modes. This is essentially true, though the 7135s provide a certain degree of predictability (the current will not exceed X) and are definitely safer with something like the 219C, which can easily pull an unhealthy amount of current in a direct-drive configuration.
It read to me as if there is something wrong with setting the chip number to get a certain max current when it would fall out of regulation right away. It’s a matter of choice. It should be obvious that maximizing output will come at the cost of run time and efficiency but nobody ever said floosh was free.