L
Admin
[Technical article, rfforum Editorial Team]
Group-delay plots are easy to over-interpret. A "rippled" delay curve can be the filter, the fixture, the gating, or simply an aperture that is too narrow for the SNR you have.
1. What group delay is measuring
Group delay is the negative derivative of phase vs frequency. Any noise or phase glitch becomes a spike after differentiation. Narrow frequency apertures make the plot look dramatic; wide apertures smear real features.
2. Aperture and smoothing
Match aperture to the feature size you care about (passband tilt vs narrow suck-outs). If changing aperture by 2× rearranges the ripple peaks, you are looking at measurement noise or processing, not a stable DUT property. Record aperture next to the screenshot.
3. Fixture and gating
Time-domain gating that clips the filter's own ringing will invent passband delay ripple. Connector savers, adapters, and a poorly de-embedded fixture add their own phase curvature. Always compare gated vs ungated, and fixture-only thru, before blaming the filter vendor.
4. Habit that saves arguments
- Same cal, same cables, same aperture when comparing lots
- A thru or airline sanity check after cal before the DUT
- Report aperture, IFBW, and whether gating was on
Discussion: for your last SAW/BAW or cavity filter sign-off, what group-delay aperture did you standardise, and have you ever rejected a lot that later proved to be a gating artefact?
— rfforum 编辑部 / Editorial Team
Group-delay plots are easy to over-interpret. A "rippled" delay curve can be the filter, the fixture, the gating, or simply an aperture that is too narrow for the SNR you have.
1. What group delay is measuring
Group delay is the negative derivative of phase vs frequency. Any noise or phase glitch becomes a spike after differentiation. Narrow frequency apertures make the plot look dramatic; wide apertures smear real features.
2. Aperture and smoothing
Match aperture to the feature size you care about (passband tilt vs narrow suck-outs). If changing aperture by 2× rearranges the ripple peaks, you are looking at measurement noise or processing, not a stable DUT property. Record aperture next to the screenshot.
3. Fixture and gating
Time-domain gating that clips the filter's own ringing will invent passband delay ripple. Connector savers, adapters, and a poorly de-embedded fixture add their own phase curvature. Always compare gated vs ungated, and fixture-only thru, before blaming the filter vendor.
4. Habit that saves arguments
- Same cal, same cables, same aperture when comparing lots
- A thru or airline sanity check after cal before the DUT
- Report aperture, IFBW, and whether gating was on
Discussion: for your last SAW/BAW or cavity filter sign-off, what group-delay aperture did you standardise, and have you ever rejected a lot that later proved to be a gating artefact?
— rfforum 编辑部 / Editorial Team
