Get rid of the 30 pF capacitors. The normal loading on "watch" type crystals is 6 pF or 12.5 pF depending on what type you have. Your circuit will supply enough stray capacitance to properly load these crystals.

Joe

----------
From:   Stephen Birchall (Stephen Birchall)[SMTP:N.S.Birchall@LBORO.AC.UK]
Sent:   Tuesday, July 08, 1997 4:44 AM
To:     PICLIST@MITVMA.MIT.EDU
Subject:        Re: Trying to get 32KHz crystal to spin up!

I've often found that connecting the crystal can to ground will get an
oscillator to function

Steve

>
>I'm working on my PIC "Power Manager" which is set up much like the circuit
>shown in AN582, with a fast main crystal between OSC1 and OSC2, and then
>the watch crystal between T1OSI and T1OSO.  This is a pretty neat thing
>that the PIC can do - allowing a sleeping PIC to sleep, but continue to
>keep time via the Timer1 and the 32KHz crystal.  I have much of my firmware
>done.  The problem is - I can't get the 32KHz crystal to oscillate reliably.
>My boards don't come in for another week, so, I'm prototyping on a vector
>board.
>
>I'm not an analog guru, but I've tried keeping the wires short, etc.  If I
>lick my pinky and touch around the circuit (seriously...) I can get a few
>hundred ticks into Timer1 but nothing sustained.
>
>I have two 30pF caps across the crystal to ground.  I've tried some large value
>resistors (e.g. 500K to 10M) across the crystal, and can often get
>several dozen ticks.  No nets are greater than, say, 1/2 inch.
>
>I'm surprised this is all *that* sensitive.  Am I doing something wrong?
>