I agree it would be nice to have an easier way to get more linear regulated current. Some discussion of this 5710 chip here.
It regulates on the high side of the LED so it’s not a simple drop in for the 7135.
I agree for higher current the efficiency is usually not bad compared to a typical buck driver. But for low current the efficiency of linear regulation is worse. An extreme example is the 2mm white flat whose voltage is ~2.75V at 0.5A which equates to about 2.75/4.2= 65% efficient.
EXCEPT for the fact the no cell remains at 4.2V under load… so with sag considered efficiency will be something better than that 65%, not much but still…
Also, a lot of chargers terminate at 4.16V or so, in that case efficiency would take another step forward…
True, it was just a simple approximation for low current.
The efficiency improves as the voltage difference decreases, so as both the voltage drop from circuit resistance increases and the LED voltage increases with current. Linear driver efficiency approaches 100% at direct drive, when it doesn’t drop any voltage.
It’s putting a strain on the cell when it can least afford it. Cell-voltage for sure will drop like a rock, and current-draw from a boost circuit will almost certainly want to maintain output, drawing even more current from the cell.
That’s why at least with linears, they’ll saturate and pull out of regulation as cell voltage drops below Vf + 0.1V, drawing less current, dimming the LED, letting you know it’s on its way out.
Yeah, but that seems to be way too soon when you’re dealing with a lot of 7135’s in parallel. For example, on my Convoy C8’s that have 8 x 7135 chips, they drop about 25% of their output by the time the cell is about half-drained (3.75v). This is using 30Q’s, so they should not have much voltage sag at 2.8A current (about 0.1v sag).
I am unimpressed with using linear drivers in high-output lights. Boost-drivers are the way to go. Or FET (for cheap lights), and just accept the fact that it’s going to drop output quickly.
Bad? What is bad from your perspective? What is the maximum amount of cell stress you are willing to let be?
I do not see how drawing more current at the lowest voltage levels can be any more bad. As long as you aren't overheating the cell or going much above its rated discharge current there's nothing really bad involved. The slight increase in drawn current at the end also trips the low voltage warning/protection a tad sooner, this can be seen as a good thing. :-)
Let me say that charging the cell to 4.2V and letting it rest for hours (or more) before use is in fact more detrimental for it than some increase in current draw just before end of discharge. And let me also say that the increase in current draw is not just at the end, but a general phenomenon happening from start to end due to the more or less constant power output and corresponding more or less required constant power input. More or less because driver efficiency progression matters, of course.
Rating: no big deal. :-D
Actually, that might be because all that current is being pushed through thin pc board traces and vias, not actual wires, through 2 sides of the board, yet.
Kinda like the difference between an exhaust manifold and exhaust header. Lots backpressure in the manifold, not much in the header.
Ie,
vs
It seems like the OP’s question was answered pretty thoroughly in the first page of comments.
FET, FET+1, FET+N+1, and 7135-only drivers are popular because:
- They’re simple, easy, compact, cheap, and well-understood.
- Designs for them are openly published.
- They’re widely available.
- They’re supported by a bunch of open-source firmware.
- They’re easy to swap in to almost any light.
- They work well enough for most common purposes.
Other types of drivers offer a few benefits though. They increase efficiency, they enable higher-voltage or in-series LEDs, they improve output regulation if done well, they can reduce heat on medium-high modes, and they decrease the frequency of posts complaining about FETs, 7135 chips, and PWM.
Someone could make fancier drivers more popular, but it hasn’t happened yet because:
- They’re harder to design, so fewer people are able to.
- They’re harder to fit inside small hosts.
- Hardware development costs money.
- The designs mostly aren’t open-source.
- Aside from a couple common sizes, each light tends to need its own custom driver designed.
- Although some have been created, the good ones haven’t really been widely available.
- They’re generally not compatible with open-source firmware, and the code for them has mostly remained closed.
This can change though, if a skilled electrical engineer is willing to publish a bunch of open designs for several different form factors, make them compatible with open-source code (or at least open up the code written for them), and work with a manufacturer to make them easy to obtain. Perhaps get in on some BLF community projects. Basically, donate a whole lot of time, effort, and money to the community. And BLF has a tendency to burn out the most skilled circuit designers, because their work is in such high demand and they don’t really get paid for their work.
I’m at least trying to make firmware ready for newer drivers, whenever those drivers come into existence. Like, creating a hardware abstraction layer and porting it to a variety of different MCUs and types of drivers. That way, new hardware can be immediately compatible with several interfaces, and new interfaces can be immediately compatible with several types of hardware, and users can mix and match as desired. But I’m not doing the hardware side of that.
As mentioned, some of us (Mike C and myself, at least) are actively working with this chip.
And a few of my recent designs (not promised to be perfect, still in testing… please don’t order unless you feel like gambling): 3A single sided, 17mm and 5A single sided for Emisar D1S
Interesting chip, but not a direct replacement for the 7135. More powerful and flexible, for sure. But it also has a slightly higher voltage drop. And if anyone knows of any chips similar to the 7135 / CN5710 / CN5711… feel free to share (or PM)!
Very nice, look like another fun project! Do you happen to have forum page you written about this drivers?
It looks pretty interesting. Any idea how it handles lower-voltage LEDs? Like, a red XP-E2 or XQ-E, for example, is down around 2V. I’m curious if it might be able to handle that or if it’d get too hot burning off so much extra voltage.
Voltage delta times current equals dissipated power. In this case it may be too much for a linear, though it depends on available cooling.
A buck converter is a lot better for low Vf emitters. Easy current output tuning and high efficiency. Used one a couple months ago for a friend's custom red XP-E2 SK98 build.
Can understand you may want to port your software, though. Porting to PIC or related microcontrollers found on available buck converters would help. Good luck in any case.
Cheers ^:)
The first few post and TK pretty well covered why there so popular but the history of the driver has alot to do with it to. These linear drivers came about in a completely different era than what we have now. Alot of batteries back then could hardly sustain 3 amps , especially the cheap fire brands. The top leds back then had a higher vf than now. Couple the higher vf leds with the tremendous voltage sag of most batteries back then and the linear driver was quite efficient in that era. It was the single best single cell driver out there. Easy to change the current by adding more 7135’s or unsoldering a few for less current. They were easy to mod and very very popular. As things progressed it’s hard to let go of what you know works good even if it’s outdated and starting to not be the best option. As it’s already been pointed out with today’s lower vf leds and the higher capable voltage of batteries under load it’s becoming not as efficient. But it’s still cheap, dependable, easy to find and highly moddable. Back when they came out they were the hottest thing since sliced bread. The very first one I remember looked like this.
It had no mcu for modes just on or off controlled by the switch. We use to master, slave two or more to achieve higher current’s, then the simpler 7135 piggy back method was born.
You have to keep in mind back when these came out, a led running at 1.5 amps was over driven. My memory’s a little foggy on this but I think it was the multi-die emitters that finally exceeded that. The SSC-P7 and the Cree MC-E, both had a high vf compared to today and the T6 xml was in there somewhere close.
Times change and their still very popular but I do feel its just a matter of time before the truth sets in and something better is developed.
I hope it comes from the good people right here on BLF. 
Thanks for the narrative, and well said! That’s why I welcome ideas such as what dsk3 brought up in the OP. Even if it’s not the exact idea that gets brought to life, it sparks good discussion and might get some wheels turning in many minds. 
Why would you want to know “it’s on it’s way out” when the battery is still 70% full…
Because that’s how it is most of the time.
There already are drivers which dim or blink when a certain voltage is reached as a warning, there is no need to have a constantly decreasing brightness over half or more of the usable runtime.
If you have an LED with a low enough Vf that it doesn’t decrease most of the time then you end up with low efficiency like 75% or worse, which is stupid. That could be another 30% longer runtime if you used a proper regulated driver.
Yes, it is. But I have a specific use in mind which is very small and needs four channels… and it wouldn’t have room for four separate buck converters. It also needs to be very resistant to impact, so I’d like to avoid any large components if possible.
I haven’t done too much testing yet, but Mike C has done a bit. He said their built-in temp regulation seems to do a good job of dialing down current automatically to keep things in check.
What can we do to support our developers?
Financially, there was a R&D fund that to my knowledge was never used to help driver developers. And really flopped - which may make people resistant to support another of its kind.
Is there some other way to help their budgets?
Is there non-financial support that we could provide?
Yeah, temperature throttling works pretty nicely, at least during my initial test, post #39: Driver giveaway: Constant current 17mm drivers, winners (finally) announced, post #2. - #39 by Mike_C
I had a fair amount of voltage to burn off, worked nicely and was very smooth. I’ve played with CAT4104s and they have temperature shutdown which is very annoying. They overheat rather quickly and then start a blinky party. Too bad really, they are rated up to 25V.
ToyKeeper, I’ll get around to responding to your post in my driver thread shortly. It was pretty heavy and I’m heading out the door. I’ll be back in a day or so!