A bit of help with a crystal oscillator

You snipped too much for this conversation to make sense.

snipped-for-privacy@highlandsniptechnology.com wrote

Unbelievable :)

The power level must be miniscule - just the gate current and what gets coupled to the gate via parasitic capacitances (Miller effect etc).

It's fairly common to leave jfet gates open, and let them bias themselves, like in electret microphones. This one even AGCs itself. The drain pullup came from the TTL gate input circuit. So the DC situation isn't totally absurd.

But why it oscillates is still a mystery to me. It did make a nice clock at the TTL gate output.

That's a complaint you don't hear often.

If you look at the issue from a 'math' POV: For an UART that samples at half the bitrate for 1 bit the error can be plus and minus 1/2 bit, so +-50 %. Normally you sent like 1 start bit 8, data bits, and 1 stop bit. That makes 10 bits and the error accumulates (sample point shift) so then is +- 5%. If you assume a similar error for the other side you need +- 2.5 %.

Check it with your chips's internal oscillator spec over the temperature range you will be using. Or just send some character in a loop and measure the length - and change of of length of a frame while using the heat gun or fridge on the chip ;-). All that just in case you design for space or something ..

Don't you dare! ;-)

Jeroen Belleman

Actually, with a bit of care, this could be made into a nice 155 MHz Clapp oscillator. See 04.ASC JFET 155MHz Clapp Osc in Oscillators.zip

I thought it would be cool to make an XO from a schmitt trigger inverter, a crystal to ground at the input, and a feedback resistor. This must oscillate, and always does, just nowhere close to the expected frequency. The crystal is just an expensive capacitor.

Why "nowhere close to the expected frequency"? This is in essence a digital oscillator and has been made to work. Heck, it's been done through software connecting an input port to an output port... on the same pin.

You are doing it wrong. You have a RC oscillator. You need to get 180 degree phase reversal for positive feedback. A pi network will do this. Just add the series load capacitors and stick the crystal between them.

See file 08.ASC, "Pierce crystal oscillator with B-source", in Oscillators.zip, at

It is designed for 1/(2 * pi * sqrt(10e-3 * 0.1e-12)) = 5,032,921 Hz, not counting the series load caps and crystal capacitance. You can find the calculations in Vig.

It runs at 5 MHz according to LTspice.

Jan Panteltje wrote

Yes; UARTs don't need an accurate clock.

Well, until the other end is doing auto baud detection ;) Modems used to autobaud on the "AT" characters and with many of them you need a clock accurate to a fraction of 1%.

I'm not doing it. It's just something I tried once.

What's wrong with trying things?

I was referring to your schmitt inverter. If you knew anything about Barkhausen Criteria, you wouldn't even waste your time and come to the wrong conclusion. I fixed it for you.

Classical concepts like Barkhausen are useless in strongly nonlinear circuits, like Schmitt oscillators.

I tried the Schmitt XO before Spice was available. It might actually work, with a fast Schmitt and the right crystal.

I suppose you've got a bit of an impedance mismatch there. The crystal can resonate all it wants, but the Schmitt trigger remains impervious to that.

I've always wondered why ICs invariably use the Pierce. Wouldn't they rather use a topology that has one end of the crystal at GND and use the pin thus freed for something interesting? OK, you'd need a pair of caps inside the package, but is that a real problem?

Jeroen Belleman

False. You determine the power input to the crystal through the drive resistor, The input is a square wave, the signal at the crystal is a sine wave. The other end of the crystal is also a sine wave of approximately the same amplitude but opposite phase. The difference in amplitude is due to the input capacitance of the inverter.

If the voltage swing to the input of the inverter is sufficient to drive the output, you meet Barkhausen and the circuit will oscillate.

Without phase inversion through the pi network, it cannot work.

All you will get is a RC oscillator.

Cost. An inverter in a digital chip is essentially free. A Colpitts or Clapp would require a bipolar or JFET, which may be incompatible with the process.

An extra pin may be a problem on a small ic. Just go to the next larger size.

There's no reason why you can't make a Colpitts with MOSFETs. There are plenty of MOSFETs in an IC. Package pins are a precious limited resource.

Jeroen Belleman

Do you know of any examples of Mosfet Colpitts? The only MOSFET oscillators I know are plain inverters or balanced LC push-pull pairs that use on-chip inductors.

I have never seen a single-ended MOSFET crystal oscillator.

Additionally, the voltage swing at the output may be too low to drive digital logic.