I have seen may posts talk about using an excess of 7135s to achieve constant current all the way to full power. This just is not possible. I will explain in a minute. Here’s one such recent post
A convoy is a single battery flashlight. And notice it says ‘a Nichia…’, meaning single LED.
I don’t have Nichia 219C data, so let’s use this XP and XM data for the example.
In this image we see either LEDs run at 3 volts when fed 1 amp. And a dead battery is 3 volt. This works. So the correct number of 7135s is 4. So when your firmware gives 100% PWM to the 7135s they regulate the amps to 1 no matter what the charge state is. And at 1 amp the LED runs at 3 volt no matter the battery charge. Producing a perfect brightness start to finish… You would probably want a FET on that board too for a high setting. But high would suffer a dimming effect.
That’s the right way.
What happens if you do it the wrong way? Say 12 7135s? That’s 4.2 amps. When your battery is full there is enough voltage to run the LED at 4.2 amps and 4.2 volts. Yay! But as the battery drains you lose volts. Eventually you only have enough VOLTS to run it at the 1 amp level. So you have gained nothing. And what’s worse is this affects ALL the levels, not just turbo.
It sounds like you understand the situation for single led and single cell and it’s important that people realize how the drop in cell voltage affects output. Linear drivers have always had this implicit in their design, you can add more chips until you max out cell voltage at the cost of run time in regulation or max out run time in regulation at the cost of output. Most people go for output and use lower modes to extend time in regulation even knowing that pwm isn’t the most efficient way to generate lower modes. This was one of the driving influences behind the LD-1/2 drivers and a good reason to use a good buck driver when possible. It also influenced the FET +1 drivers that maximize output on high and also runtime on the lower modes. Stacking chips is well understood and still a popular way to dial in a maximum current and with 2-channel pwm control also gives the same run time benefits on the lower modes. Lots of choices in how to run a light now and all good ones with oodles of programming choices.
Good info there! I used a FET+3 in my triple XP-L build and the med mode draws about 1A, with a 30Q it almost achieves full regulation the whole runtime.
It is also important for people to know that if you rarely run a light on full output, it’s better to use a FET+X driver instead of a 7135 only driver. Running 2x 7135 chips on 100% is more efficient than running 8x chips on 25%.
It used to be a rule that you shouldn’t run LEDs in parallel since the one with the lowest Vf would draw more current and poof. I guess binning has improved a bit. A good FET would feed more current but be unable to evenly feed multiple LEDs the way a board like that can do. Nor could it tailor output the to different types of LEDs. 7135’s are still pretty cool devices.
For anyone interested, there is better tech out there than a 7135. This chip for example is smaller than 1 7135 but can replace 16 of them! And it is fully adjustable from 0 amps to 4 amps. You set the amps with a couple resistors. And you can still dim it with PWM. It has a larger brother here if you think it’s too small to work with!
0.1 Ohm RDSon, inductor, sense resistor (wants to see 100mV). Just that you make it sounds like it would take the same surface area as a 7135. It is a buck converter, I think BLF is a place where we can say that, instead of "better tech"
Yes it is a buck and needs a few external components. There is a datasheet over there that has a usage example laid out on a board. Along with a parts list. Copying that should be easy. It is yet to be seen how small of a board it could be put on.
Let's say 4Vin taking into account all those things I have mentioned before there will be quite some voltage drop, 0.1V for the sense resistance, at 0.1Ohm RDSon , etc. not that attractive. If we are talking about higher input voltages then this is comparing apples to oranges, which we are doing already, linear to buck.
Yep that probably nails the intention the thread starter had.
—
In former times stacking some chips was a cheap and very good way to enhance the above mentioned limit more to the physical borders mentioned in the OP. 4.5A for a XML were a kind of standard for me, in former days.
A nanjg is unbeatable value for price 3$+ custom firmware which is free, gives you a very good and absolutely reliable driver for onecell flashlights.
I am all for maximum power so I mostly use fetdriver now which are optimized to be almost on the physical border of the possible, but can also have modes.
You can’t have anything so everyone has to choose what priorities he has mine are maximum power :evil:
The next step the fet+1 driver for 6.5$ is also an incredible step Up for us modders. It also is a one sided driver which fits most flashlights without further modding.
All the buck drivers/chips are not suitable for my kind of preference. Because they are only for “more cells than LEDs” flashlights and often lack power. If I have a single xpl single cell light which is driven with 5-7A(of course only as long as the battery voltage is high enough) I always think my bigger buckdriver light which is only driven at 4A is weak…
Yea the thread wondered a bit. The OP was warning against using more 7135s than your flashlight can handle. After post #5’s picture we started talking about the kind of flashlight that can handle dozens of 7135s, and how to avoid crazy amounts of them.
I just can’t tell what you’re suggesting? Are you saying for anything over 1amp everyone should just be using a FET or design a new driver? That’s kind of what it sounds like.
It just means you can’t force the LED to take more than 1 amp when the battery is at 3v. So any settings above 1 amp will suffer dimming. In the OP example.
