I'm looking at the 19.44MHz VCXO 3.3V in our system.
I replaced the tuning voltage with a resistive divider to eliminate a source of error. I used our scope (TDS2024) in FFT mode to sniff around. (Hanning, 64 averages) I expected to see the fundamental and its odd multiples at the output. Instead I see the second harmonic being higher in amplitude than the fundamental, and each spike has sidebands +-1.2MHz. And the craziest part is there is a spike at 1.2MHz. The power supply at the pins of the VCXO seems relatively clean. I expected to see the 19.44MHz but there's actually nothing. It's pretty quiet... So I took a VCXO chip from inventory, and connected it to a lab power supply by itself. The output has the same spectrum. So it's not some modulation caused by our board.
1) What is the output spectrum supposed to look like with a 19.44MHz square wave that can vary from 40-60% duty cycle? The 50-50 case is all odd harmonics, right? The second harmonic should be lower than the fundamental, right???2) What is the significance of the 1.2MHz spike? That's the last thing I expected to see.
3) Is there something about VCXOs I'm overlooking? Yes it drifts by a few Hz every second, I'd hardly expect that to show up on a spectrum display where each division is 5MHz.4) Probing technique. I usually use home brew Zo probes. A foot or two of RG-174, a feedthrough 50 ohm terminator at the scope end, and a 1K at the signal end so I don't load the signal too much. All soldered in. Is this OK? (I've been using this for years, but hey, I might be wrong)
I feel like ditching this ASVV part. The output signal is jittery enough that you can see the signal wobble back and forth by a pixel on the scope. I'm no expert here, but that seems like a lot of jitter. The output of my OCXO is rock solid on the screen.
Maybe another approach? Maybe a Pericom PI6CX100-27 and a 19.44 crystal? I know the Pericom part works above 27MHz, I see no reason why it shouldn't work at 19.44.
TIA