Crystal Tips (no Alistair)

[PerranOak]

I have salvaged two crystals from an old piece of kit (trailcam). All they have on them is a freq indication: 12.000 and 32.768.

I have setup a Colpitts (see attached) and got the 12.000 to oscillate at 12MHz.

However, I can't get the other one to oscillate. I assumed it was "MHz" but, having searched a bit, think it might be "kHz" as the trailcam had a clock.

Can I use this circuit to get it to oscillate? If so, where can I find the best values for the caps and resistors?

Cheers.

[Benta]

It's most likely a 32.768 kHz "tuning fork" crystal. What does it look like? A small cylinder?

These are very sensitive to overdriving, drive power should be no higher than 1 uW. Load capacitance is usually 10...20 pF.

[PerranOak]

Yes, a very small cylinder.
Oops! I've simply put it in the same circuit as the 12MHz one; hope I haven't killed it.

[Benta]

Nah, it'll most likely be OK.

32.768 kHz crystals normally operate as Pierce oscillators.

The simplest is a 74HCU04 (the "U" is important!). Use a second inverter to buffer the output.

R_f = 2...10 Mohms
R_d = 330 kohms
C_d, C_g = 10...12 pF

Never had any problems with this set up.

[PerranOak]

Cheers. I guess I'll have to buy a 74HCU04 as my 74LS04 didn't work!

[Benta]

Cheers. I guess I'll have to buy a 74HCU04 as my 74LS04 didn't work!

74LS will never work. Far too much input current, it has to be CMOS.

[PerranOak]

... and unbuffered. I tried a 74HC14 which seemed to osc at 40MHz! 

[Benta]

NEVER! use Schmitt-trigger inverters for crystal oscillators.

[PerranOak]

Agreed.

[ledtester]

For the 32kHz crystal you might try this circuit:

http://www.discovercircuits.com/DJ-Circuits/ultralowpwrxtlosc1.htm

[T3sl4co1l]

The 32kHz crystal has a very high impedance, 100s of kohms.  (In general, crystal impedance is comparable to the reactance of the loading capacitors.)  Basically it's shorted out by the bias resistors.  Try increasing the resistor values by 10 or 100x.  The output level will be pretty weak as well, but you can add another transistor to buffer it if needed.

Tim

[PerranOak]

Brilliant. The "discovercircuits" circuit worked a treat (see attached). 

Thank you very much ledtester.

I think I will need to buffer the output to trigger my Schmitt trigger, cheers.

[PerranOak]

Now that the xtal is oscillating nicely, how do I divide the freq down to a nice accurate 1Hz.

Surely it's not a case of fifteen flip-flops in a row?

[oPossum]

CD4060 will get you to 2 Hz and it has an oscillator circuit that you may be able to use.

[T3sl4co1l]

If you wish to continue in discrete form, there are a few simplifications that can be made -- a full D-type flip-flop is quite a lot of work, but a pulse-triggered toggle requires only two or three transistors.  Also, an astable multivibrator can be used as an injection locked oscillator, giving oddball ratios -- not just 1/N but also N/M if you get it to lock just so.  (Needless to say, some ratios are pretty tricky to tune in, and you're best off going with simple ratios; small powers of 2 would be most useful here I guess, such as 4 or 8.)

Otherwise, just getting the all-in-one block (see above) is a very convenient way indeed.

Tim