Adjacent Channel Leakage Ratio (ACLR) is one of the first compliance plots people stare at on an FR1 PA or small cell. The DUT matters, but the instrument setup can move ACLR by a decibel or more if you are careless.

Signal side
- Use the correct channel bandwidth and SCS for the waveform you claim to test. A 20 MHz NR waveform measured with LTE ACLR masks is not the same test.
- Set crest factor / power back-off the way the product will run (DPD on or off). Quoting ACLR at P1dB without DPD is a different story than ACLR at rated conducted power with DPD.
- Confirm the signal generator's own ACLR floor with a loopback. If the source is only 5 dB better than your DUT target, you are measuring the source.

Analyzer side
- Noise correction / noise subtraction helps when you are near the analyzer noise floor, but it can also hide real DUT noise if misapplied. Know when it is on.
- Attenuation and preamp: too little attenuation and the analyzer's own IMD eats ACLR; too much and noise floor hides the leakage. Sweep attenuator until ACLR stabilizes.
- RBW / integration bandwidth must match the standard's definition for the adjacent channel integration. Do not freestyle a narrower RBW to "clean up" the plot.
- External filters ahead of the analyzer can help reject out-of-band drive noise, but characterize their loss in the adjacent channels.

DUT and fixture
- Cable and fixture loss belong in the power reference. ACLR is a ratio, so flat loss cancels, but ripple across the main and adjacent channels does not.
- Temperature and supply sag under the high PAR waveform change ACLR a lot. Log Vcc at the board, not only at the PSU screen.

A useful lab habit: save three traces — loopback floor, DUT at rated power, DUT at −3 dB. If the middle trace sits on the loopback floor, fix the setup before tuning the PA.

What analyzer attenuator / preamp combo do you standardize on for ~23–30 dBm FR1 conducted ACLR?