Wednesday, March 7, 2012

A New Master Clock - Cont'd2

To make sure that my previous casual observation was correct, I re-installed the 100MHz Master Clock Chip in the Propeller (see previous posts). This time I included a 0.1uF By-Pass Cap at the terminals of the Oscillator for good measure. The results were as I had remembered - BAD.

The 100MHz Master Clock produce terrible Hash on the RF output, as recorded and seen at the receiver.

Vendor Supplied 5MHz Crystal
vs
100MHz Master Clock XO

The left half of the screen grab, shows the signal from the vendor supplied 5MHz Crystal (clean), and the right half is the signal from the 100MHz Master Clock (hash).

I paused the Display to capture the two signals, so they appear side-by-side.

Ignore the apparent frequency ramp on the right, after about 4 minutes, the Master Clock stabilized rock solid on the expected frequency (without a correction applied). I should have waited a few minutes before recording the results. If you look inside the hash, you can see the data is the same for both oscillators.

Note: a -320Hz correction was applied to the vendor supplied 5MHz (x 16) clock, just to bring the signal on sub-band, that is about -32ppm error (well within crystal spec's, but I want better, with much less thermal drift).

See; Jeff -KO7M's efforts with his Propeller Frequency Stability, at his blog.


Conclusion

The 100MHz Master Clock is much more accurate then the open air vendor supplied 5MHz Crystal, but it produces MUCH more phase noisy (jitter).

Note: the problem I am trying to solve is not apparent in the above capture, because the test Beacon code does not provide a rest period (cooling as the cpu is inactive) between transmitted Beacon signals.


So, the quest goes on, I am still looking for a stable, low noise, solution.


UPDATE
I put a 22pF DC blocking cap between the Oscillator and Prop, thinking that maybe DC bias was pulling the Oscillator - No change, still Hash.

--

No comments:

Post a Comment