L
Admin
[Technical article / FAQ, rfforum Editorial Team]
Mixer datasheets quote a recommended LO level. On the bench, that number is a compromise among conversion loss / NF, IP3, LO leakage, and sometimes odd spurs — and your LO chain flatness across band often violates the assumption behind the single number.
1. What LO drive actually buys
- Too little LO: diodes / FETs under-switched → higher conversion loss, worse NF, and IP3 that collapses faster than the datasheet table
- Too much LO: heating, LO leakage / feedthrough, compression in the LO buffer, and sometimes worse even-order behaviour; "more LO" is not monotonic for every mixer family
- Band edges: if your LO amp rolls off 3 dB, the mixer at the high end is not the mixer you measured at mid-band
2. Freeze list for a mixer claim
- LO power at the mixer pin (not at the synth output) across the LO band you ship
- NF and IP3 (or P1dB) at the same LO drive points — one row in a table, not two slides from different days
- LO-RF / LO-IF isolation and whether leakage fails a desense or spur mask before NF does
- Whether the LO path needs leveling / ALC so the mixer sees a constant drive when the synth or cable changes
3. Habit
Do not publish a mixer NF or IP3 without the LO dBm at the device and the frequency. A "per datasheet LO" note is not enough when the LO chain is ±2 dB across band.
Discussion: on your last mixer bring-up, what hit the gate first as LO drive moved — NF, IP3, or leakage / spur — and do you level the LO or just freeze a mid-band setpoint?
— rfforum 编辑部 / Editorial Team
Mixer datasheets quote a recommended LO level. On the bench, that number is a compromise among conversion loss / NF, IP3, LO leakage, and sometimes odd spurs — and your LO chain flatness across band often violates the assumption behind the single number.
1. What LO drive actually buys
- Too little LO: diodes / FETs under-switched → higher conversion loss, worse NF, and IP3 that collapses faster than the datasheet table
- Too much LO: heating, LO leakage / feedthrough, compression in the LO buffer, and sometimes worse even-order behaviour; "more LO" is not monotonic for every mixer family
- Band edges: if your LO amp rolls off 3 dB, the mixer at the high end is not the mixer you measured at mid-band
2. Freeze list for a mixer claim
- LO power at the mixer pin (not at the synth output) across the LO band you ship
- NF and IP3 (or P1dB) at the same LO drive points — one row in a table, not two slides from different days
- LO-RF / LO-IF isolation and whether leakage fails a desense or spur mask before NF does
- Whether the LO path needs leveling / ALC so the mixer sees a constant drive when the synth or cable changes
3. Habit
Do not publish a mixer NF or IP3 without the LO dBm at the device and the frequency. A "per datasheet LO" note is not enough when the LO chain is ±2 dB across band.
Discussion: on your last mixer bring-up, what hit the gate first as LO drive moved — NF, IP3, or leakage / spur — and do you level the LO or just freeze a mid-band setpoint?
— rfforum 编辑部 / Editorial Team
