L
Admin
[Technical article / FAQ, rfforum Editorial Team]
Integer-N PLL spur fights often start and end on the loop-filter spreadsheet. On a real board, charge-pump mismatch, supply coupling, and the reference path share the same "spur" column and can beat a textbook filter.
1. What actually sets the spur floor
- Classical CP/PFD ripple through the loop filter is real, but so is CP up/down mismatch, VCO pushing from a noisy rail, and reference harmonic coupling into the VCO tank or buffer
- A "perfect" 4th-order filter on paper does nothing if the VCO supply shares a return with a switching regulator, or if the reference CMOS edge rings into the RF ground
- Offset vs integer boundary: close-in fractional stories are different; this note is for integer-N / integer-mode spurs at N×f_ref and friends
2. Freeze list before you blame the filter alone
- Measured spur level at the RF port with CP current, loop BW, and phase margin on the same sheet
- VCO / CP / digital rail tree and whether the reference buffer returns on RF ground or digital ground
- Whether you can toggle CP current or loop BW and watch the spur move as theory predicts (if it does not, look off-filter)
- Board photo or stack note for reference routing length and shield cans over the VCO
3. Habit
Archive spur plots with loop BW, CP current, and a one-line supply/return sketch. Redesigning the filter three times while the spur is supply-coupled wastes a spin.
Discussion: on your last integer-N spur hunt, did the fix land in the loop filter, the CP/supply, or the reference path — and do you require a rail-toggle test before approving a filter respín?
— rfforum 编辑部 / Editorial Team
Integer-N PLL spur fights often start and end on the loop-filter spreadsheet. On a real board, charge-pump mismatch, supply coupling, and the reference path share the same "spur" column and can beat a textbook filter.
1. What actually sets the spur floor
- Classical CP/PFD ripple through the loop filter is real, but so is CP up/down mismatch, VCO pushing from a noisy rail, and reference harmonic coupling into the VCO tank or buffer
- A "perfect" 4th-order filter on paper does nothing if the VCO supply shares a return with a switching regulator, or if the reference CMOS edge rings into the RF ground
- Offset vs integer boundary: close-in fractional stories are different; this note is for integer-N / integer-mode spurs at N×f_ref and friends
2. Freeze list before you blame the filter alone
- Measured spur level at the RF port with CP current, loop BW, and phase margin on the same sheet
- VCO / CP / digital rail tree and whether the reference buffer returns on RF ground or digital ground
- Whether you can toggle CP current or loop BW and watch the spur move as theory predicts (if it does not, look off-filter)
- Board photo or stack note for reference routing length and shield cans over the VCO
3. Habit
Archive spur plots with loop BW, CP current, and a one-line supply/return sketch. Redesigning the filter three times while the spur is supply-coupled wastes a spin.
Discussion: on your last integer-N spur hunt, did the fix land in the loop filter, the CP/supply, or the reference path — and do you require a rail-toggle test before approving a filter respín?
— rfforum 编辑部 / Editorial Team
