Single-inverter crystal osc. circuit guidelines?

Hi all,

A surprise requirement has bit me and I have to add a 16 MHz clock, 3.3V operation to my design. I'm planning on using a single-gate inverter + crystal to produce it. If it matters, crystal is +/- 60 ppm (freq+stability).

Does my single-gate buffer need to have a schmitt-trigger input?

I'm thinking of either : NL17SZU04-D or SN74AHC1G04 for the single-gate IC. Links for your ease:

formatting link

Any suggestions on how to determine the "Rs" (series resistance)?

My data sheet gives me a drive level of 0.1mW and an ESR for the crystal of 60 Ohms. Do I use P=VI, V=IR, P=I^2R and backsolve for R assuming my V = 3.3V??

Sanity check:

For the parallel resistance I'm thinking a 1 MOhm resistor across inputs is acceptable.

For the capacitive loading, what I've read says the two caps that hang off the crystal the value should be CL == C1*C2/(C1+C2) == 1/2*CL if C1 ~ C2.

I appreciate your feedback and help. Oscillator circuits are obviously not my specialty.

John.

Hello John,

No. Should be an inverter and preferably an unbuffered one.

Looks good.

Not so good for this case, has Schmitt inputs AFAIK.

Best is to go by what the crystal mfg recommends. The feedback resistor is mainly there just to get things started after VCC comes on. Then, ideally there should be an R at the output decoupling Cout from the inverter's output. The ONSemi data sheet to the 74HCU04 shows how.

Important is to stick to the recommended Cin and Cout. Mind the capacitance of the chip.

Best to read up on it in app notes. Cout (C2?) may need to be decoupled from the output via a resistor. But you'd still take the clock signal off the output pin.

BTW if in a real time crunch you can buy these whole oscillators in a can. 16MHz is very common. Then you don't have to worry about all the above, just make sure it has a nice decoupling cap next to it.