Self oscillating buck LED driver has too many components

[ronsimpson]

I don't see anything against using behavioral models to create a "IC" in spice.

It seems pretty clear as to how the ZXLD1366 works. The 600khz osc is the only part that is not clear.

 

[alec_t]

its just a representative simulation so that the students can see how its working..

In that case can't you just replace the comparators by two more behavioural voltage sources? Or is that a cheat too far?

 

[Flyback]

yes that's a bit too far,
The 600KHz max frequency is because above that, and you tend to violate the minimum on or off times of the fet. Also, at that higher frequency, the di/dt tends to get higher and the comparator delay becomes more significant and causes led current inaccuracy.

 

[alec_t]

Here's how I'd do the latching of the turn-off comparator (I don't see any point in latching the other):

 

[Flyback]

Thanks Alec_t...that's extremely impressive, I didn't realise it could be done as well as you've done it with so few components.

 

[alec_t]

..... even fewer if you use a single hysteretic comparator . For amusement, here's a sim of the twin-comparator and single-comparator versions side by side. Virtually identical LED current profiles can be achieved with component values as shown.
The single can be made to run faster than the twin (by reducing the hysteresis), because of less inherent delay, enabling use of a lower inductance value for a given ripple amount.