duty_cycle getter doesn't return what setter was given

[DrRob]

Hi, why does the setter have value + 1 in line 91?:

value = (value + 1) >> 4

There isn't a corresponding adjustment in the getter, so for example if I set duty_cycle to 0x7fff and then read it back I get 0x8000.

Thanks!
Rob.

[tekktrik]

@tannewt Looks like this was in the original commit for the library, curious if you know or remember why the getter and setter would be different?

[tannewt]

I don't remember. Sorry!

It would make more sense to read the same value back.

[tekktrik]

Thanks @tannewt!

Looks like it might be a bug @DrRob, are you interested in submitting a PR to fix it?

[DrRob]

I'll try, but I'm puzzled about why the value + 1 part doesn't result in a requested duty cycle of zero actually being on for 1/4096th of the period. (It doesn't - I checked with an oscilloscope.)

[DrRob]

Actually I got that wrong - if a duty cycle of zero is requested, then (0 + 1) >> 4 == 1 >> 4 == 0, so both the LEDn_ON and LEDn_OFF registers are set to zero, which probably explains why the output is off all the time as expected...

...HOWEVER, the datasheet says, "The LEDn_ON and LEDn_OFF count registers should never be programmed with the same values." LEDn_OFF should be set to 0x1000 for fully off.

I'll try to fix it according to the datasheet instructions.

[DrRob]

I'm a bit rusty with git but I think I've just submitted a PR for this.