R
Member
A datasheet is a set of conditional statements. Before comparing two parts, write down the conditions each number was measured under.
Checklist:
1) Frequency, bias, temperature and supply voltage. Gain and noise figure quoted at one bias point do not transfer to another.
2) Test fixture and reference plane. Some figures are de-embedded to the package leads, others to the die; that difference can be several tenths of a dB.
3) Typical vs min/max. Typical values are not guaranteed; design with the guaranteed limits for production.
4) Linearity definitions: P1dB, IIP3/OIP3, and the tone spacing used. Two-tone results depend on tone separation and termination.
5) Stability and impedance conditions: an LNA characterised in a 50 ohm system may behave differently with a real antenna match.
6) What is NOT specified. Absence of a phase-noise, VSWR or temperature-drift spec is information too.
A useful habit for interviews and design reviews: pick one parameter and explain the measurement setup that would produce that number. If you cannot describe the fixture, calibration and conditions, you do not yet know what the number means.
When building a comparison table, add columns for conditions, units (dB vs dBm vs linear), whether the value is guaranteed, and the source document revision. Re-check the revision before freezing a design.
Takeaway: RF engineering is largely about knowing the conditions behind a number. The ability to reconstruct the measurement is more valuable than memorising typical values.
Checklist:
1) Frequency, bias, temperature and supply voltage. Gain and noise figure quoted at one bias point do not transfer to another.
2) Test fixture and reference plane. Some figures are de-embedded to the package leads, others to the die; that difference can be several tenths of a dB.
3) Typical vs min/max. Typical values are not guaranteed; design with the guaranteed limits for production.
4) Linearity definitions: P1dB, IIP3/OIP3, and the tone spacing used. Two-tone results depend on tone separation and termination.
5) Stability and impedance conditions: an LNA characterised in a 50 ohm system may behave differently with a real antenna match.
6) What is NOT specified. Absence of a phase-noise, VSWR or temperature-drift spec is information too.
A useful habit for interviews and design reviews: pick one parameter and explain the measurement setup that would produce that number. If you cannot describe the fixture, calibration and conditions, you do not yet know what the number means.
When building a comparison table, add columns for conditions, units (dB vs dBm vs linear), whether the value is guaranteed, and the source document revision. Re-check the revision before freezing a design.
Takeaway: RF engineering is largely about knowing the conditions behind a number. The ability to reconstruct the measurement is more valuable than memorising typical values.
