L
Admin
[Technical article, rfforum Editorial Team]
After a SOLT or TRL calibration, every degree of phase drift in the test cables is an error in your measured S-parameters. "Phase-stable" cables are sold for exactly this reason, but the datasheet number is easy to misread.
1. What the spec usually is
Vendors often quote phase change versus flex for a defined bend (for example a 90° wrap around a mandrel of stated diameter), sometimes also versus temperature. It is not a promise that the cable never moves; it is a bound under that specific abuse. A cable that is excellent on the bench can still drift if you hang it, twist the connector, or let it rest against a hot PA heatsink.
2. Why flex matters after calibration
Calibration stores the error terms with the cables in one geometry. If you then move the cable to connect the DUT, the electrical length and the connector interface change. At 10 GHz, 1° of phase is only about 83 µm of equivalent air path — easy to get from a soft bend. Reflection measurements (S11) and any phase-sensitive parameter (group delay, balanced phase, differential match) feel it first.
3. Habits that keep cal alive
- Dress cables once after calibration and do not move them again until you are done.
- Use phase-stable (often semi-rigid or specialised flexible) jumpers for the last metre to the DUT; keep long cheap cables only on the analyser side of a stable reference plane if you must.
- Torque connectors to spec; hand-tight plus a quarter turn is not a torque wrench.
- Recheck a known thru or a gold-standard attenuator after any cable move. If S21 phase or S11 moved more than your uncertainty budget, recalibrate.
- For temperature work, either control the cable temperature or use cables with a published ppm/°C phase coeff and budget it.
4. Connector and adaptor stack
Every adaptor is another interface. A "temporary" 3.5-to-SMA stack for one measurement is a common source of non-repeatable phase. Prefer a single calibrated interface to the DUT.
Discussion: what phase drift (degrees at your highest frequency) do you tolerate before you force a recal, and do you log cable serial numbers in the test report?
— rfforum 编辑部 / Editorial Team
After a SOLT or TRL calibration, every degree of phase drift in the test cables is an error in your measured S-parameters. "Phase-stable" cables are sold for exactly this reason, but the datasheet number is easy to misread.
1. What the spec usually is
Vendors often quote phase change versus flex for a defined bend (for example a 90° wrap around a mandrel of stated diameter), sometimes also versus temperature. It is not a promise that the cable never moves; it is a bound under that specific abuse. A cable that is excellent on the bench can still drift if you hang it, twist the connector, or let it rest against a hot PA heatsink.
2. Why flex matters after calibration
Calibration stores the error terms with the cables in one geometry. If you then move the cable to connect the DUT, the electrical length and the connector interface change. At 10 GHz, 1° of phase is only about 83 µm of equivalent air path — easy to get from a soft bend. Reflection measurements (S11) and any phase-sensitive parameter (group delay, balanced phase, differential match) feel it first.
3. Habits that keep cal alive
- Dress cables once after calibration and do not move them again until you are done.
- Use phase-stable (often semi-rigid or specialised flexible) jumpers for the last metre to the DUT; keep long cheap cables only on the analyser side of a stable reference plane if you must.
- Torque connectors to spec; hand-tight plus a quarter turn is not a torque wrench.
- Recheck a known thru or a gold-standard attenuator after any cable move. If S21 phase or S11 moved more than your uncertainty budget, recalibrate.
- For temperature work, either control the cable temperature or use cables with a published ppm/°C phase coeff and budget it.
4. Connector and adaptor stack
Every adaptor is another interface. A "temporary" 3.5-to-SMA stack for one measurement is a common source of non-repeatable phase. Prefer a single calibrated interface to the DUT.
Discussion: what phase drift (degrees at your highest frequency) do you tolerate before you force a recal, and do you log cable serial numbers in the test report?
— rfforum 编辑部 / Editorial Team
