[time-nuts] Advice on 10 MHz isolation/distribution / Phase Noise of 74AC gates (Bruce Griffiths)
GThorp at pascall.co.uk
Fri Feb 26 16:59:27 UTC 2010
A little more detail is required such as:
1) What was the divided down output of the 74AC163 compared with?
The E5052B contains 2 uncorrelated test systems, and uses its 2 internal
synthesisers (with separate 10MHz ref OCXOs) as references. The signal
under test is split at the input, and cross-correlation is used to
reduce the effect of reference noise.
2) An image of the breadboard would also be useful.
It was 2 years ago, and it doesn't exist any more! The IC was an SO-16
package, stuck upside down to a piece of copper-clad board.
3) A circuit diagram showing component values and manufacturer's part
Circuit attached. I don't remember the supply filter resistor value, but
it was probably ~100R. I adjusted the supply to give 5V on the IC when
it was operating.
I've read the E5052B manual but there's insufficient detail to have
confidence in the calibration technique used when a square wave input is
The E5052B doesn't tune the DUT, but digitally locks its synthesisers to
the signal, so there is no VCO constant to calibrate. It uses analogue
mixers as phase detectors. I assume it measures the slope at zero
crossing as it appears able to work with high harmonic levels, e.g.
passing undistorted signals through amplifiers driven into compression
and re-measuring doesn't change the measured close-in phase noise (in
the 1/f^3 region). Below 100Hz where the oscillator's phase noise
dominates, the 12.5MHz curve tracks it 18dB lower, which gives
confidence that the instrument has got the calibration right.
More information about the time-nuts