Critique my LED PWMer

[magic]

Quite the opposite: R9 provides path for ESD to reach Q8 gate when Q7 is off and break down the gate insulation. Such resistor should be connected straight to the source.

[Siwastaja]

Seeing you seem to worry about the inaccuracy in LED current due to nonlinearity of PWM, do you realize LED Vf varies all over the place with temperature etc., and your 5V supply varies, too, so you never get regulated, accurate current with simple series resistor driver anyway. If you need accurate LED current, you need to measure it and feedback. If you need accurate output, you need to sense the light and feedback on that.

[robzy]

Quite the opposite: R9 provides path for ESD to reach Q8 gate when Q7 is off and break down the gate insulation. Such resistor should be connected straight to the source.

Oh! That's a good point! I was aware that ESD on the gate was an issue, it didn't really click with me that ESD's impact on Vgs could be an issue. In this case my chosen MOSFETs have inbuilt zener(?) diodes between gate and source, would that've saved them?

Either way I'll modify my design to put the fuse before the current limiting circuitry.

If you need accurate LED current

To clarify my previous point - I don't need accurate LED current, but I want to know the relative change in power from nominal 1uS pulses to 2uS pulses. And its easiest to do this if the actual pulse length is close to the nominal pulse length (i.e. not eaten up by MOSFET switching time).

[StillTrying]

but I want to know the relative change in power from nominal 1uS pulses to 2uS pulses.

Why don't you view the 1 & 2us light pulses with a photodiode, a couple of us have done it here.

https://www.eevblog.com/forum/chat/20w-halogen-bulb-viewed-by-a-photodiode/msg2411274/#msg2411274

With 1A through the LED you wouldn't need any photodiode amp.

[robzy]

Why don't you view the 1 & 2us light pulses with a photodiode, a couple of us have done it here.

https://www.eevblog.com/forum/chat/20w-halogen-bulb-viewed-by-a-photodiode/msg2411274/#msg2411274

With 1A through the LED you wouldn't need any photodiode amp.

Oh that's cool! Thanks for the link!

I don't have any measurement equiptment with the required bandwidth, though, and even if I did I'd still like to invest engineering effort up-front to make sure the circuit switches fast enough that 1uS pulses are approx half the length of 2uS pulses.