Logic gate prop delay

[JasonL220]

i was thinking of making a oscillator similar to this, **broken link removed** with a 50MHz crystal, and was looking at not gates, and found **broken link removed** but they have a propagation delay of 8ns seconds and the clock will be oscillating witha period of 2ns. does it matter that the prop delay is longer than the period of oscillation?

thanks

 

[Nigel Goodwin]

I don't see any particular connection between propagation delay and the frequency of oscillation - but from a practical point, can you easily source a 50MHz fundamental crystal?.

 

[audioguru]

The crystal's datasheet has the oscillator circuit they use to test their crystals.

 

[JasonL220]

well what i want is and oscillator that has one output line ( instead of 2 ) those diagrams looked like they would work but if the prop delay of the not gate is greater than the oscillator period would that mean that it would not oscillate properly.

 

[Nigel Goodwin]

Feel free to completely ignore my previous post!

 

[evandude]

Why don't you just use a 'canned' oscillator that has built-in drive circuitry? these seem to be more common for frequencies that high, and it's the simplest solution, especially considering it sounds like you are trying to reinvent the wheel and build exactly the same thing with an inverter anyway... digikey has some for under $2 apiece, like part no. 535-9183-5-ND , for example.

 

[Sceadwian]

I think the hardest part of getting an oscillator started at higher frequencies is finding a noise source for the initial feedback to start the oscillation. A long propogation delay simply means the intial noise pulse that starts the oscillator has to be longer than the delay, or the instrinsic noise at the input has to be at least that high. A simple tuned high gain amplifier stage just before the input to the logic gate would pretty well solve any difficuties with starting the oscillator.

The logic gate driven Crystal amplifier as described is going to have severe stability issues, especially with stray capacitance (a human finger is more than effective at disturbing oscillations at those frequencies) and the severe phase difference (What 700+ degree's out of phase?) that are generated.

Evandude is right, just use a canned oscillator.

 

[hjames]

These oscillator circuits are *ANALOG* and "propagation delays" don't matter. Be sure to get the unbuffered versions inverter chips that the crystal manufacturer recommends, otherwise the oscillator will be erratic.

(Normal inverter gates usually have multiple inverters in series. Crystal oscillators want 2 mosfets biased in their analog region, hence the feedback resistor. This doesn't work out too well with the buffered versions of the chips).