On Fri, 6 Nov 1998, Chris Eddy wrote:

> Starfire; (Former anger set aside)
>
> I haven't done a PLL, but may have to on another project soon.  I thought
> of coupling a PLL and PIC, but the customer wants an ultra low cost
> device.  Would it be possible (possible for sure, practical is the
> question) to couple an R2R or other DAC to a variable cap diode in an RF
> circuit, then come back through a prescale into the PIC so that the PIC
> can monitor frequency.  In this manner, the PIC can take charge of the
> many PLL tasks, except for the RF portions.  My application would only
> have to account for temp drift, and as such the PLL doesn't have to have
> massive performance.I want to give the user easy selection of five major
> transmit frequencies without any complex factory tuning.

I did something with a MC145171, with and without extra UHF prescaler. The
only trick is to clean the data control lines from the PIC to the
PLL/synthesizer of RF if any is present (i.e. use RCs and proper
decoupling on the data lines). If you fail to do this strange errors can
happen sometimes. The PIC and the synth shared the same clock osc. which
was implemented at the Motorola part.

A PIC-only PLL is best done using a PWM ouput from a single pin followed
by a dumb RC low-pass. A transistor (or two) is (are) sometimes used to
chop a 33V voltage for larger tuning ranges (often used in un-syntesized
TVs). Of course the slow output filter must be accounted for in the PLL
loop to prevent instability. PLLs implemented like this with a PIC are
slow reacting as a rule (is there a rule ?).

Peter