PLEASE HELP WITH CRYSTAL LAYOUT

[lawrence11]

So this is abracon 8mhz 4 pad smd crystal, pad 2 and 4 are "gnd".

On wich ground would this be cinnected for best performance,

global pink, or local blue ?

Thank you.

[lawrence11]

So this is abracon 8mhz 4 pad smd crystal, pad 2 and 4 are "gnd".

On wich ground would this be cinnected for best performance,

global pink, or local blue ?

Thank you.

[ataradov]

Does not matter, they are not connected to anything internally, and crystals are not susceptible to reasonable levels of EMI. There is an ongoing debate whether you should connect the can to the GND at all.

[bd139]

They are susceptible in some circumstances i.e. if you're using them as crystal filters on low level RF signals. Grounding the case actually kills off a lot of crazy. But this is rare and you usually buy them canned and matched already so no one gives a crap about that other than the hams.

In the circumstance above the rationale for grounding it is to stop RFI escaping the crystal. If it's a digital system, meh. I'd connect it to the lowest impedance ground you have. If it's an RF system with very low level (uV) signals then it needs to have a local decoupled power supply etc and a carefully designed grounding "tree" to stop unintentional feedback from the XO into the rest of the system.

Ergo it depends on your application. I am in the middle of developing a transceiver and on an earlier board rev, the synthesizer LO was being picked up by the front end AGC amp and turning this into IMD products at the first mixer (birdies). YMMV but stuff like this is important but only if you're dealing with very low signal levels, RF etc.

[Siwastaja]

There shouldn't be any "local" or "global" grounds like that. Add a ground plane and connect every ground connection to that, using traces as short as possible.

[David Chamberlain]

Pin 1 is connected to ground also?

[CJay]

They are susceptible in some circumstances i.e. if you're using them as crystal filters on low level RF signals. Grounding the case actually kills off a lot of crazy.

I'm of the belief it changes the capacitance but I'd be interested to see some results of grounded vs. ungrounded from some 'proper' test equipment to prove or disprove that belief

[Gribo]

Pin1 should not be connected to ground. Your MCU has XTAL in and XTAL out pins. 1 and 3 should be connected to these pins, doesn't matter which one. Can you post the schematic?

[Rerouter]

Having a ground ring tied to the oscillators ground pin is recommended by a few manufacturers guidebooks, in this case you would also tie both of the pads to ground, they do not actually connect inside, but are used to ground the case to provide sheilding from above.

Next up, I see thick nets underneath the crystal, Ideally there should be nothing other than an optional ground fill under the 2 crystal pins (1 & 3), my normal approach is to have nothing there, and another ground ring on the other side, filling it just increases capacitance, meaning you need to compensate with slightly larger matching capacitors.

pins 1 & 3 will need to connect to the microcontrollers oscillator pins,

[bd139]

Indeed. When I get my hands on some I shall test it. Unfortunately the stuff is slippery so far

[Simon]

I have merged the two identical threads, please do not post the same topic twice.