I have built and have been testing the following circuit based on the TPS922052. I used a draft tool shared with me to compute some of the values, and then looked at the EVAL schematic for further information. I'm putting the chip in flexible dimming mode and then combining analog and PWM dimming (mostly PWM on the low end, mostly analog on the high end) to increase the dimming ratio.
I've configured PWM to 4 Khz / 10-bit for both DIM and ADIM (I know that e.g. ADIM only has 8-bit resolution, but that shouldn't factor in).
The weird thing I'm seeing is that at the very darkest end, I'm observing hysteresis behavior where I can dim down to 0 quite smoothly, but then when fading in, until my DIM PWM hits between 2-3 microseconds of pulse width, no light is produced. I've tested this across a range of PWM frequencies, the timing is quite consistent where until I hit 2-3 microsecond pulses, it's dark. During this time ADIM duty is kept at 15%.
Note that once I hit this value, I can then reduce the width of the pulse e.g. to 500 ns and it produces (dimmer) light. This makes sense given the chip's stated 150-ns pulse width minimum. So in summary, on the way down, smooth; on the way up from zero, no output until it hits 2-3 us PWM and then smooth.
I tried removing C8, C9 and C10 just in case that was having an effect on narrow pulse widths, but this didn't make a difference.
Second question is around completely turning off the light. With PWM and ADIM both sending 0 duty cycle signals, there is still a tiny bit of light on the output. Note that I bumped up R7/R8 from the eval based on a recommendation I read here, but that didn't seem to do the trick. I'm using an LED which is 18v/500mA if that makes a difference.
Any suggestions would be much appreciated, the chip is otherwise looking quite good!
