L
Admin
[Technical article / FAQ, rfforum Editorial Team]
Phase-noise plots of oscillators, synthesizers, and mixed-signal clocks are easy to over-trust. If the stimulus or the analyzer's own LO is noisier than the DUT close-in, the trace you archive is a floor measurement dressed up as a product number.
1. What actually limits the plot
- Source PN, analyzer LO PN, and cross-correlation settings set a floor; below that floor you are measuring the instrument path
- A "clean" DUT on a noisy generator looks like the generator; swapping to a quieter reference can drop the trace several dB without touching the DUT
- Cable microphonics, ground loops, and supply ripple add close-in skirts that look like oscillator noise
2. Freeze list before you publish a PN number
- Stimulus / reference PN specification (or measured) next to the DUT plot, with offset frequencies marked
- Analyzer mode (direct spectrum, PLL, cross-correlation) and correlation count / averaging
- Whether a quieter reference or a known-good oscillator was substituted as a floor check
- Supply, grounding, and mechanical isolation notes for close-in offsets
3. Habit
Never archive a DUT PN claim without a floor check. If the plot does not move when you improve the reference, you were already on the floor.
Discussion: on your last "DUT too noisy" chase, did the fix land in a quieter reference, analyzer settings, or the DUT itself — and do you require a floor-substitution check before freezing close-in PN?
— rfforum 编辑部 / Editorial Team
Phase-noise plots of oscillators, synthesizers, and mixed-signal clocks are easy to over-trust. If the stimulus or the analyzer's own LO is noisier than the DUT close-in, the trace you archive is a floor measurement dressed up as a product number.
1. What actually limits the plot
- Source PN, analyzer LO PN, and cross-correlation settings set a floor; below that floor you are measuring the instrument path
- A "clean" DUT on a noisy generator looks like the generator; swapping to a quieter reference can drop the trace several dB without touching the DUT
- Cable microphonics, ground loops, and supply ripple add close-in skirts that look like oscillator noise
2. Freeze list before you publish a PN number
- Stimulus / reference PN specification (or measured) next to the DUT plot, with offset frequencies marked
- Analyzer mode (direct spectrum, PLL, cross-correlation) and correlation count / averaging
- Whether a quieter reference or a known-good oscillator was substituted as a floor check
- Supply, grounding, and mechanical isolation notes for close-in offsets
3. Habit
Never archive a DUT PN claim without a floor check. If the plot does not move when you improve the reference, you were already on the floor.
Discussion: on your last "DUT too noisy" chase, did the fix land in a quieter reference, analyzer settings, or the DUT itself — and do you require a floor-substitution check before freezing close-in PN?
— rfforum 编辑部 / Editorial Team
