analogcapacitornoiseoscillatorstability
I saw in a schematic a while back that a polypropylene capacitor was used to generate a stable sine wave in an oscillator circuit. I suppose this gave a better 'frequency reference' for the oscillator.
Are some capacitor types better for lower noise applications?
10 22 33 47 68 series is pretty much all you will need. For audio/analog applications you will probably need values from 10p–1μ PP and 1μ–470μ electrolytic and some bigger ~10000μF caps.
That Colpitts oscillator will oscillate at the crystal's most active resonant frequency. Any crystal has resonant modes at many frequencies.
As Olin has mentioned, any crystal is resonantly active at harmonics of its fundamental resonant frequency. Since crystals above (roughly) 30 MHz become very thin and fragile, a manufacturer might modify a crystal to be more active at its third harmonic than at its fundamental frequency. In your Colpitts oscillator, such an overtone crystal will most likely oscillate at the higher frequency.
For a fundamental-mode crystal, it is possible to oscillate on an overtone, by adding LC components that either kill fundamental feedback, or promote feedback at the higher frequency (or both). A modification of your Colpitts oscillator to run at a third harmonic might look like this:
simulate this circuit – Schematic created using CircuitLab
The components L1, C1 suppress oscillation at 10 MHz. where the crystal's fundamental frequency would otherwise cause oscillation. Q1 has more gain at 10 MHz than at 30 MHz, so this overtone oscillator can be a bit fussy. The component selection of C2 and C3 for small values helps to promote oscillation at the higher frequencies. While third overtone oscillation is not guaranteed, the 5th or 7th crystal overtone modes usually have more losses.
Taking output from emitter will yield a distorted wave shape. If you wish to have a more pure sinusoidal wave shape that is free of harmonics, then consider taking output from the crystal. This will be a very small signal requiring amplification. A common-base amplifier can provide the low input impedance required, and give gain too:
Best Answer
HighK ceramics like X7R, Z5U etc have huge variation of capacitance versus voltage. Using them in filters or any kind of coupling application guarantees humongous distortion. They are piezoelectric: they are both good loudspeakers and microphones. Decoupling a high impedance node with them results in a nice vibration detector. Tolerance on values are not huge, rather they are hyuuuge: expect +20/-50% depending on DC bias. Also, they drift a lot with temperature.
They are truly excellent for power supply decoupling, though, because they are small, have lots of capacitance per volume, low inductance, and are cheap. For decoupling, who cares if it's 1µF +/- 50%?
Now, for filtering applications, or when you run a signal through a cap as in your oscillator application, you want...
Dielectric absorption and leakage will not matter for your oscillator, but they will for other applications.
Film caps and NP0 ceramic caps are excellent on this, although:
Your first choice should be NP0 ceramics if they are available in the value you need. They are small and cheap, and almost perfect.
NP0 ceramics and High-K ceramics like X7R/Z5U are completely different materials. High-K capacitance varies with DC bias, NP0 does not.