When I did first zenclock I thought I was going to have to design a 'jitter adding' circuit to add some
random level of low level jitter to make the clock sound musical. As it turns out just tweaking the power supply
was enough.
Do i understand that your aim was to develop a low jitter clock (to ignore for a moment the complex spectral properties of jitter in general), but that the jitter of your clock as you initially developed it was too low, so you had to modify the power supply to add some jitter back...?
Not quite.
I developed a second generation zenclock which had substantially lower jitter by measurement.
Subjectively it wasn't quite right. It was better in every 'technical' sense but less musical.
After liaising with the oscillator manufacturers, I found they had used a low noise, low dropout IC regulator inside the can.
Lots of people use these for clock supplies. I replaced this with a discrete, low noise reg of my own design and the sound was right.
Theoretically, my discrete reg actually has lower noise than the IC one they used. But I don't have measuring equipment
good enough to measure the jitter after reg change. Also the layout was different - it could have gone slightly either
way (more or less jitter).
Regardless, it retained all the previous positive attributes and also sounded musically right.
So in fact, I am not driven totally by measurements, the final arbiter is listening.
I just used measurements to get the foundations right and did final tweaking based on how it sounded.
Z