Why do I have to touch the crystal to start oscillation?

[cellularmitosis]

So I had a bag of crystals lying around labelled "3G1431A".  I assumed that meant "14.31MHz", but I thought a good learning exercise would be to actually measure the frequency to be sure.

I googled for crystal circuits and came up with figure 1e from Jim William's AN12: Circuit Techniques for Clock Sources.

I gave it a go on the breadboard and the circuit worked.  I measured 14.29MHz on my DSO.

However, what puzzles me is that I have to touch the crystal (one finger on the case, then tap the leg which connects to the 2N3904 base pin) in order to get the oscillation started.

What's up with that?

[Nerull]

Crystals generally have to be 'started' to work reliably. This is normally achieved via load capacitors, placed between each terminal and ground. The spec sheet for a given crystal will give the capacitance required.



A quartz crystal works like a tuning fork, vibrating at a given frequency. But only once struck. The capacitor values in that diagram aren't even in the ballpark for load capacitors, nor in the right places. I'm not sure if they're left out with the assumption that you will know to put them in, or this application simply doesn't use them, resulting in the startup problems you see.

[T3sl4co1l]

He's got the loading caps.

Probably just not enough gain.  Change the 1k resistor to a nice RFC (radio frequency choke), something more than 1k impedance at the resonant frequency (i.e., more than 11uH, with less than 11pF parasitic capacitance, i.e., a self resonant frequency 14MHz or higher).

Tim

[Tac Eht Xilef]

I googled for crystal circuits and came up with figure 1e from Jim William's AN12: Circuit Techniques for Clock Sources.

Which is essentially a Pierce oscillator design. It may help to look at other transistor-based Pierce oscillators designs & explanations (which can be a bit difficult, because if you Google "pierce oscillator" 95% of results are based on CMOS gates...)

However, what puzzles me is that I have to touch the crystal (one finger on the case, then tap the leg which connects to the 2N3904 base pin) in order to get the oscillation started.

What's up with that?

Without doing the numbers, the 100pF cap from base to emitter/gnd looks a bit high - particularly for one built on breadboard. I'd try 1/2 of that or less and see what happens...