Grounding Crystal cases?

[westfw]

So this came up on another forum, but those are mostly software people, so I thought I'd ask here.
Some times you see the case/can of a crystal in a circuit nicely (or not so nicely) soldered to the ground plane of a PCB.  But not really that often.  Sometimes you even see crystal packages with a lead or tab on the can designed to make this easy.  But those are not very common, and for a lot of the smaller "modern" crystal packages, making such a connection ranges between "inconvenient" to "nearly impossible" (OTOH, there are also modern packages that include ground pads...)

So ... when is it important to ground the can of a crystal?  Offhand, I'd guess something like "if the crystal is being used in a low signal-level filter and is nearby other components running at similar frequencies, ground the can may be useful.  In most digital circuits where the crystal is being used in a relatively high-power simple oscillator, it's not important..."

[AmmoJammo]

Sorry, don't know the answer, but I'm curious about the reasoning behind it too

[Tac Eht Xilef]

In theory it can minimise coupling between adjacent crystals (in old radios you'll often see them staggered at right angles for the same reason), and minimise the parallel parasitic capacitance of the case (it becomes a component of the load capacitance).

Again, in theory the result is increased stability & higher Q. In practice it makes a slight difference, but is probably really only necessary where you want absolute maximum Q e.g. crystal filters.

On the downside it's easy to damage the can seal, though that's probably most likely in the case of old - or cheap Chinese - crystals.

A lot of old-school RF guys do it simply out of habit. Some of the DIY audio/synth guys, etc, believe in it too, but I wouldn't take too much notice of that 

[dannyf]

when is it important to ground the can of a crystal?

When the straigh capacitance is significant. Given that the crystal's pullability is very limited, I wouldn't worry about it unless you are into hundreds of Mhz - a rarity for crystal oscillators.

[GEuser]

Also , crystals drift slightly with heat whether just ambient heat (the air say 5c to 30c) or the heat if any of what it is attached too , on some certain devices a drift of a few hundred hertz can be critical .

So , grounding or attaching the crystal housing might/can either help disassociate the crystal from the ambient by keeping it at the actual device temperature , or the opposite of helping isolate the crystal from the device temperature and holding it at the current ambient , it'll also depend on how warm it is inside the box if its in one ..

[peter.mitchell]

Also , crystals drift slightly with heat whether just ambient heat (the air say 5c to 30c) or the heat if any of what it is attached too , on some certain devices a drift of a few hundred hertz can be critical .

So , grounding or attaching the crystal housing might/can either help disassociate the crystal from the ambient by keeping it at the actual device temperature , or the opposite of helping isolate the crystal from the device temperature and holding it at the current ambient , it'll also depend on how warm it is inside the box if its in one ..

This can be part of it, but also, if it is directly soldered to a ground plane, eg a blob on the side of the can going to a pad on the pcb, it can reduce mechanical stress and vibrations which also upset some crystals.

[JackOfVA]

As has been said, it's a matter of reducing unwanted coupling in crystal filters.  A good crystal filter can exhibit an ultimate rejection > 80 dB so it takes only a tiny bit of stray capacitive coupling between crystals to leak a signal at -80 dB down from the input level.  Grounding the crystal case to the PCB reduces the unwanted capacitive coupling.  It also takes attention to detail in the physical layout of the crystal filter components, routing leads, etc., but grounded crystal cases seems far more prevalent than not.

In the days when HC-6U crystals were used with plug-in pins (0.093" diameter so they would fit into an octal vacuum tube socket) the crystal would often be held in place with a metal spring clip, grounded to the chassis. These were oscillator circuits, not filters, however.