I can completely understand that.
Regarding to your question about missed clever tricks: Her is a not so tricky one:
By nature the HL1606 strips have shift register where you shift through the values and latch if all values are set. I am driving 2112 LEDs with that. It is organized in 32 LED strings (chains of shift registers) with 66 LEDs (currently).
By parallelling the shift register I deal with 32 shift 'entry points'. Those values can be easily put out over 4 8 bit pins, with some 1 bit pins for clock signals and latch signals.
So only the length of my strips defines my 'image clock' to put out a complete write cycle to all the leds. Thats is the reason why I can live with such low kHz of pixel timing.
I only try that much bits as I am able to do (it has to work regardless if it is perfect). I have not calculated the real performance (or even measured) but from my look at it it is quite good - especially with dark values (which are always a bit problematic with too slow PWM).
But then again I do not really think that my solution is better than yours (it si more or less a hack) but perhaps one idea helps you. Or better it trigger a better idea in your mind ;)
Perhaps this can help you.
Multi-channel PWM for LED controller
-
- XCore Addict
- Posts: 216
- Joined: Wed Feb 10, 2010 10:26 am
-
- Junior Member
- Posts: 6
- Joined: Thu Mar 31, 2011 4:44 pm
Thanks.Interactive_Matter wrote: I only try that much bits as I am able to do (it has to work regardless if it is perfect). I have not calculated the real performance (or even measured) but from my look at it it is quite good - especially with dark values (which are always a bit problematic with too slow PWM).
But then again I do not really think that my solution is better than yours (it si more or less a hack) but perhaps one idea helps you. Or better it trigger a better idea in your mind ;)
Perhaps this can help you.
I have now been looking into multi-channel Bit Density Modulation and Bit Angle Modulation using software,
as if my figures for PWM generation are correct, I can't see that it is practical. The requirements on interrupt rate for BAM are much reduced, so there's hope.