Can we replace the AMC7135?

Here is my thinking: the 7135 showed up on cheap Chinese-made drivers in early-ish LED flashlights. We’ve stuck with it for quite awhile, even making new drivers to utilize it and stacking dozens of chips to increase the current. The fact that it comes on very low-cost mass-produced drivers tells me that it must be a very cost effective choice, but probably not the most high performance choice.

I know FETs are all the rage right now, but I prefer regulation. We found high performance FETs and built drivers around them; isn’t there a high performance regulator out there we can build a driver around? My goal is to get a regulated 5-6 amps out of a 17mm driver (Attiny controlled, of course), and single-sided if at all possible.

Unfortunately, I don’t have the skills to read a datasheet and know what it means. Can anybody help me out?

This is out of my league but very good question sir!

I wish I could help, but I don’t have computer hacking skills.

How about the 7136? It is the same linear regulation circuit but it breaks out the FET and the current sense resistor so you can size it for any current you want.

I actually have some of those. Wight did a driver utilizing it, but it was kind of forgotton because the PWM was too low.

After re-reading the entire thread ( Wight’s 7136 driver, ) it’s evident most of the issues were caused by one particular “feature” of that QX7136. Maybe chips from a different manufacturer wouldn’t have that “feature”? Or maybe there is a different part that performs the same function but would be better suited to higher frequencies? I don’t see a problem with having a separate controller and external fet to achieve regulation.

Pretty sure this is what led4power has done with the LD-1/2 drivers.

Kind of. They have gone full CC non-PWM, which is nice. But they’re also not Attiny based, not an open design, and because I can’t build it myself, more expensive than I’m interested in paying.

I’m looking for an upgrade from the 7135 that’s still Attiny based and easy to build boards for.

If you want 6A to an XML2 (for example), can’t you just add a 3.9V zener between +/-?

That would make it constant voltage instead of constant current. As the LED temperature changes (and effects the Vf) you would get a change in the current.

Sorry to jump in, but yes, I think what we’d really need is a constant CURRENT regulator, which is what the 7135 is, right?

Single cell and 5-6A…a fet Driver is the best way possible because it has the lowest resistance and loss…
With the high forward voltages of LEDs these days it won’t even overdrive so hard as the battery is limiting…so a limiting(not regulating) driver is not neccessary.

All regulating drivers use a resistor and measure the voltage drop over it to measure and then regulate current… that means losses

There are plenty of regulating drivers around here, TK made some, RMM has a buck driver and the LD2 seems also nice. Of course all can be rebuild with an tiny as mcu…

I have the 7136 circuit in my supfire m6 if I am right and it was a bit weird with the Pwm values and of course it will never be a power horse like a direct driven Fet…

The main point is the price a nanjg for 3 $ with custom firmware is the best thing we have. I am sure someone could build fet drivers for the same money in big counts. But the assembly and making costs are huge so you have to build a lot so that the price drops. I had hope that the Fet drivers from the a6 will make it to market with the initial run of 1000 pcs they should be able to sell them for a few dollars…

Yes I understand that Vf would limit the top end with most LED’s, but there are some with lower Vf’s like the 219c. I’m not trying to replace the Nanjg for low cost, just find another option.

I had a go at this a while ago, and got some different components to try out. However, it got over my head and when looking at the specs these 7135s ain’t that bad and they’re easy to use. By setting them up with 1 on pin 1, 2 on pin 2, 4 on pin 3 and so on let’s me turn on any amount of them in constant current with steps of 0.35 or 0.38 amps, with an additional one for PWM between these steps. From what I could make out, doing this with other components will not save me space, cost or provide additional flexibility.

So I’m sticking with them for now. They are pretty flexible, can be made smaller by cutting of the ground tab (with adjusted footprint) and they seem to handle heat pretty good. But I’m all ears if you find a candidate for replacement.

An adjustable voltage regulator can be used as a current regulator, as in this example. A high current LDO adjustable voltage regulator such as the TI TPS75901. can potentially be used.

The challenge is that the FB pin needs to get to 1.22 V. This would require too big a voltage drop across the sense resistor at the high desired current. An op-amp can be used to amplify the sense voltage. Or perhaps the attiny25 can serve as an amplifier using an ADC input pin and an analog voltage out pin.

An alternative is just to use a FET with a sense resistor and program an attiny25 to regulate the current through the sense resistor. So it would be an LD-2 clone but with open-source hardware and software.

That sounds promising. Do you have a schematic for that?

The sense resistor is in series with the LED and the FET.
The voltage drop over this resistor has to be very small to be useful for a 1-cell flashlight.

The resulting signal would need to be amplified to something that is useable by the MCU.
This is not trivial, as the voltage we are measuring would typically be very close to either of the supply rails of the opamp.
It needs some careful design and opamp chip selection.

I suspect the LD-2 is closed-source because there is a PID or similar control programmed in the MCU that could easily be broken if the software is customized. It probably also requires some calibration procedure for each unit.

That sounds like it could be minimized by adequate heatsinking. Any guesses about how much output would drop, on average, between 25 and ~85°C? The official datasheets doesn’t feature graphs for 6A (10% drop at 85°C with XML2 700mA).

It looks like someone has done this using ATTiny84. There is a complete write-up including schematics and source code. The challenge is how to scale the current up to 6A and use an MCU with fewer pins.