Oscillator pulling in dense PA layouts

VCO pulling 2MHz when the PA next door transmits. Guard rings and distance help but board space is tight. Besides physical separation, what actually fixes injection pulling?

2 MHz of VCO pulling when the PA keys up is classic injection pulling / supply–ground bounce, and board space limits usually mean you attack the coupling path, not just "move it farther."

Checks that often fix more than another guard ring:

1. Separate VCO / PLL supply with its own LC + LDO island; never share a switcher phase with the PA bias. Measure pulling with a current probe on that rail — if the rail moves when PA TX, fix supply first.
2. Magnetic / E-field: rotate the VCO inductor / tank away from PA matching currents; a shorted turn or fence only helps if it breaks the actual H-field loop you measure with a near-field probe.
3. Ground: a single narrow neck between PA return and VCO GND is a classic injector — widen or stitch with many vias, or give the VCO its own local GND pour tied at one point.
4. Buffer / isolation amp after the VCO before the PA driver chain; pulling often enters through the first buffer's reverse isolation, not through the air.
5. If it is PLL push rather than free-running pull, check loop BW vs the PA envelope — widening or narrowing the loop can move you out of the sensitive region.

Share whether the VCO is on-die with the PA, a discrete PLL, and the offset in frequency between VCO and PA carrier — that changes which of the above dominates.

— rfforum 编辑部 / Editorial Team