DIY constant current 2-LED remote, powered by 18650

[Peabody]

I recently built a single-LED TV remote that works ok, but would like to design a version with two LEDs in series to produce twice the IR power from the same current.  And I want it to be constant current.  The basic setup includes a 3.3V 8MHz Adruino Pro Mini with the power indicator LED and voltage regulator removed, with both the Arduino and the LED string powered directly from an 18650 Li-ion cell.  The output would be typical NEC format 38KHz with a 30% duty cycle.  The LEDs are TSAL6100, and the LED current is to be about 500mA.

The circuit I've come up with is shown below.  I will need to order some parts for it, but before doing that I wanted to see if anyone sees anything that obviously won't work, or which could be done better.

Of course the woolly mammoth in the room is the voltage headroom as the battery discharges.  The remote I built uses a TSAL6200, which I believe is the same LED but with a different lens.  At 500mA, its Vf is 1.7V.  So two of them would be 3.4V, and add another 0.1V for whatever, and in theory the circuit should still work down to 3.5V battery voltage.

But for constant current, I can't use the typical circuit of a source resistor that drops 0.7V in order to feed the NPN transistor base that provides negative feedback.  That 0.7V headroom "waste" would make it impossible to use two LEDs.  Instead, I put one of the LEDs in the source circuit, and depend on the idea that the Vf of the LED increases with current.  And I also provide for, but may not need, a 0.2R resistor which will drop that extra 0.1V at 500mA.

I would tune the other four resistors so the NPN has no effect at 3.5V battery voltage, but at any higher voltage the constant current feedback would kick in to take the mosfet out of saturation.  It doesn't have to stay at exactly 500mA, but hopefully pretty close.  I have no idea what the other resistor values should be.  And actually, I'm not sure that  the feedback will work at all, and not oscillate, but seems like it ought to - it just needs to damp the increased current that would otherwise flow, on up to a 4.2V battery voltage.

It looks like the mosfet will easily handle 500mA with 1.7V GS voltage, which is half a volt above its maximum threshold voltage, and almost a full volt above typical threshold.

I would appreciate any comments and suggestions.

[Benta]

The circuit you've drawn won't work, unless you've found a MOSFET that will turn fully on at 1.5 V (unlikely, but perhaps. Sounds expensive).
Why don't you use the A/D for the current sense resistor? Much simpler.
I know nothing about the Arduino A/Ds, but that's where I'd go.

[MikeK]

The circuit you've drawn won't work, unless you've found a MOSFET that will turn fully on at 1.5 V (unlikely, but perhaps. Sounds expensive).

Unless I've misread the datasheet, it'll pass 10A with 1.5Vgs.

[Peabody]

The circuit you've drawn won't work, unless you've found a MOSFET that will turn fully on at 1.5 V (unlikely, but perhaps. Sounds expensive).

It's 50 cents at Digikey, which claims to have half a million in stock.

http://aosmd.com/res/data_sheets/AOSS32136C.pdf

Why don't you use the A/D for the current sense resistor? Much simpler.
I know nothing about the Arduino A/Ds, but that's where I'd go.

I guess this is where it becomes obvious that I'm not an engineer.  I don't know what you mean.

[Peabody]

Why don't you use the A/D for the current sense resistor? Much simpler.
I know nothing about the Arduino A/Ds, but that's where I'd go.

Can anyone explain what Benta meant in this quote?  I can't figure it out.

In general, if there's a better way to do the constant current, I'd like to know about it.

[tooki]

Why only 500mA? I’m pretty sure that IR emitter is capable of doing significantly more, around 1A.

[Peabody]

I certainly wouldn't rule out going to 1A.  But the challenge is to get two LEDs working in series using a single lithium-ion cell.  Two LEDs in series at 500mA is the same IR power as one LED at 1A, so I think I would still get a lot of range, but with less battery voltage sag and other possible complications.  And of course twice the battery life, nominally.  But if I can get it working at 500mA, I will certainly try taking it to higher current.

I have only one TSAL6200, but last night I found in the garage an old DVD player remote. I'll try the LED from that as the second LED.  I'll have to see what its Vf is, but I think I should be able to see if the constant current part works at all.  I have some doubts about that.  The question is whether an IR LED acts enough like a resistor (higher Vf at higher current) to be usable in controlling the current.  The TSAL datsheet suggests it might, but I'll just have to see.