It can be used for reclocking if the phase noise is sufficiently low. The exact frequency is unimportant as long as it is much higher than the frequency of the signal to be reclocked.
Let me share my solution: I bought a 1 MHz Motorola TCXO on eBay. It is about 2" x 2" size and the metal can is fixed by 4 screws at the sides. Internal there was a 4 MHz crystal and a 2-transistor Colpitts oscillator, followed by a CMOS divide-by-4 circuit. I opened it and replaced the 4 MHz crystal with a 16.9344 MHz crystal. Initially it did not oscillate, so I replaced one capacitor and it worked. This is now the 4.2336 MHz master oscillator in my Marantz CD-84 player, and I use it also for reclocking the Latch Enable signal of the DAC chips, and also for producing a Dynamic Element Matching (DEM) clock for the same. The DEM clock is divided by 8 and it is 529.2 kHz (12 times the sampling frequency of 44.1 kHz). This is a non-oversampling DAC, and I replaced the TDA1540P 14-bit chips to a TDA1541A.
Let me share my solution: I bought a 1 MHz Motorola TCXO on eBay. It is about 2" x 2" size and the metal can is fixed by 4 screws at the sides. Internal there was a 4 MHz crystal and a 2-transistor Colpitts oscillator, followed by a CMOS divide-by-4 circuit. I opened it and replaced the 4 MHz crystal with a 16.9344 MHz crystal. Initially it did not oscillate, so I replaced one capacitor and it worked. This is now the 4.2336 MHz master oscillator in my Marantz CD-84 player, and I use it also for reclocking the Latch Enable signal of the DAC chips, and also for producing a Dynamic Element Matching (DEM) clock for the same. The DEM clock is divided by 8 and it is 529.2 kHz (12 times the sampling frequency of 44.1 kHz). This is a non-oversampling DAC, and I replaced the TDA1540P 14-bit chips to a TDA1541A.
Unfortunately I have no tools for measuring jitter. I am thinking about building one. I could make comparisons only be ear.I_Forgot said:
Crystal oscillators built around transistor circuits have generally better phase noise than those with ICs. I believe one requirement for low jitter is the low oscillator voltage (preferably sine wave) on the crystal. A tuned LC circuit in the collector also helps. The type of the oscillator (Colpitts, Pierce, etc.) also has influence. Those DIP clocks are built around TTL or CMOS digital circuits, and they have full 5V square wave on the crystal.
It is not necessarily true that a discrete oscillator will have lower jitter than an integrated one. Especially not if you change the crystal out of the discrete oscillator's design range then start monkeying with caps to get it to start again. Did you use a crystal that is cut for lowest jitter? If the oscillator didn't start with the new crystal and you had to change caps to get it to start, does it now have lower or higher jitter than the original? How does the frequency change affect the spectrum of the jitter?
If you can't measure the jitter to see if there's some improvement it seems sort of pointless to be randomly swapping one clock for another. You might as well be swapping capacitors and reversing resistors to see if you can hear a difference. I suppose if you don't have anything else to do...
I_F
If you can't measure the jitter to see if there's some improvement it seems sort of pointless to be randomly swapping one clock for another. You might as well be swapping capacitors and reversing resistors to see if you can hear a difference. I suppose if you don't have anything else to do...
I_F
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.