L
Admin
[Technical article / FAQ, rfforum Editorial Team]
After a disappointing S11, it is tempting to renormalise the port to 45 Ω or 55 Ω "because the datasheet says the chip is not 50." Sometimes that is correct physics. Often it is a way to hide a launch, via, or calibration mistake.
1. When renorm is legitimate
- The device is specified in a non-50 Ω system impedance and you are comparing to that reference
- You have already validated a thru / TRL fixture on the same stackup and the raw waves look causal and passive
- You document the target Z0 in the same plot title as the S-parameters
2. When renorm is a smell
- Port impedance reported by the solver is wild, or the thru phase/loss is nonsense — fix the port / mesh / launch first (/t/301)
- Renorm "fixes" S11 but makes |S| non-passive or blows out group delay
- Different engineers pick different Z0 until the curve looks shippable — that is presentation, not measurement
3. Habit
Order of operations: (1) sane port and mesh, (2) fixture / de-embed with a same-stackup standard, (3) only then renorm if the product reference impedance demands it. Archive both 50 Ω and renormed Touchstone when marketing needs the latter.
Discussion: on your last "datasheet Z0" debate, did renorm survive a same-stackup thru check, or did it fall apart once the launch was modelled honestly?
— rfforum 编辑部 / Editorial Team
After a disappointing S11, it is tempting to renormalise the port to 45 Ω or 55 Ω "because the datasheet says the chip is not 50." Sometimes that is correct physics. Often it is a way to hide a launch, via, or calibration mistake.
1. When renorm is legitimate
- The device is specified in a non-50 Ω system impedance and you are comparing to that reference
- You have already validated a thru / TRL fixture on the same stackup and the raw waves look causal and passive
- You document the target Z0 in the same plot title as the S-parameters
2. When renorm is a smell
- Port impedance reported by the solver is wild, or the thru phase/loss is nonsense — fix the port / mesh / launch first (/t/301)
- Renorm "fixes" S11 but makes |S| non-passive or blows out group delay
- Different engineers pick different Z0 until the curve looks shippable — that is presentation, not measurement
3. Habit
Order of operations: (1) sane port and mesh, (2) fixture / de-embed with a same-stackup standard, (3) only then renorm if the product reference impedance demands it. Archive both 50 Ω and renormed Touchstone when marketing needs the latter.
Discussion: on your last "datasheet Z0" debate, did renorm survive a same-stackup thru check, or did it fall apart once the launch was modelled honestly?
— rfforum 编辑部 / Editorial Team
