Sometimes crystal cases are grounded and sometimes not. Why is this?

 

I'm not sure but I think it may be a mechanical mounting thing.

 

The old school crystal cases had maybe 1 or 2 pf of case capacitance so the effectiveness of grounding the case for frequency stability increased as the frequency of the crystal increased. For the old tube type radios I worked on it was more for mechanical stability and EM shielding than absolute frequency stability.

 

I have experience with small cylindrical 32 kHz crystals, sometimes they will not run if shell is not grounded.

 

I was on a project that used a microcontroller, on the eve of on-chip RC clocks became a clocking option, consequently we use a crystal. The analog circuitry associated with the microcontroller is very sensitive to noise pickup, particularly a continuous signal. It turned out that one of the types of crystals we had in prototypes was doing a nice job of coupling the oscillator frequency into the analog circuit. Grounding the crystal can helped a lot.

To trebla's experience with watch crystals. Yep, it seems to me that some require more load capacitance than others. Grounding the can would add to the load capacitance.

 

So grounding the can benefits:
mechanical support for the crystal,
Reducing interference from the crystal to the rest of the circuit, and vice versa.

Are there any disadvantages?

 

I have experience with small cylindrical 32 kHz crystals, sometimes they will not run if shell is not grounded.

I find this hard to believe.

 

I put a loop of wire around the case of watch crystals and solder it to PCB holes, but that's just to hold it in place until I can solder the leads.

 

I have noticed that all the populated prototype/development boards from Microchip that I have, do not have the 32Khz or system crystal case bonded to GND.
I have never grounded them on projects either.
Max.

 

I find this hard to believe.

So was i when first time stucked to non-working 32 kHz crystal several years ago. I struggled with this crystal about week until i accidentally discovered that crystal shell grounding fixes the poblem. Not all 32kHz crystals behave similarly, even with same MCU. I have an Meterman/Amprobe simple contact thermometer which worked for years until stopped. I changed CR2032 cell but the thermometer was starting and then resetting shortly. After connecting crystal can to common wire it started normally and is in good shape today too. I have'nt found good explanation for it but i leave now for this type crystals small grounding area on PCBs.