>I've found the DigiKey resonators (ECS I think) to work great at 4MHz
>and
>10MHz and not at all at 20MHz.  Haven't tried other frequencies; any
>idea
>what the problem at 20MHz is?
>


        After finding my prototypes worked with the ECS resonator at 16
MHz, we went ahead and built 100 units using them.  They all either
didn't work at all or failed under elevated temperature.  I talked with
ECS, Integrity and muRata.  Both ECS and Integrity said that the
resonators worked in their sample circuit (using a cmos inverter) and if
they didn't work in my circuit (the 16c74a) the problem was my circuit.
In talking with muRata, I found that they work with the semiconductor
companies, getting samples of the chips and modifying the resonator
design as necessary for reliable operation with that chip.  When I spoke
with them a little over a year ago, they had not characterized a device
for use with Microchip as that time.  However, I found that the existing
devices seem to work fine (we've probably shipped about 1,000 systems
with their resonators on the 16c74a).
        As to "what the problem is", I imagine there is not enough gain
in the 16c74a to reliably get over the losses in the other resonators.  I
talked with Microchip about getting plots of input Z, output Z, gain and
phase vs frequency for the oscillator.  They did not have that data and
said it would be expensive to produce.  So, we end up with resonators
that don't work and I wasn't able to give the resonator companies specs
on what WOULD make it work, since Microchip had not completed the tests.
Microchip did correctly point out that they provide about as much
oscillator data as other microcontroller manufacturers (very little).
        So...  we ended up using muRata.

Harold