PLL Loop Filter Designer
Design a type-2 second-order PLL passive loop filter. Calculates time constants, capacitor and resistor values for target loop bandwidth and phase margin.
Charge-pump PLL, 2nd-order passive filter · exact phase-margin fitAPI slug pll-loop-filter
Specification
Circuit
| Ref | Position | Ideal |
|---|---|---|
| R1 | Zero | 393.2Ω |
| C1 | Zero | 623.4nF |
| C2 | Pole | 62.34nF |
Response
- Crossover as Built
- 5kHz
- Actual Phase Margin
- +45.0°
- BW / f_ref Ratiofc/f_ref
- 0.05PASS
Formula
| Symbol | Quantity | Your value |
|---|---|---|
| ωc | Loop crossover frequency (2π × f_BW) (rad/s) | — |
| T₁ | Zero time constant, the root of PM = arctan(ωc·T₁) − arctan(ωc·γ·T₁) (s) | — |
| γ | C₂/C₁ = T₃/T₁: at most 0.1, smaller for high phase margins | — |
| Icp | Charge pump current (A) | 500 µA |
| Kvco | VCO gain (rad/s/V) | 20 MHz/V |
| N | Feedback divider ratio | 100 |
Assumptions
5 items
- 01A charge-pump PLL whose open-loop gain is G(s) = (I_cp/2π)·Z(s)·K_vco/(s·N), K_vco in rad/s/V, with R₁ in series with C₁ to ground and C₂ to ground beside them; the loop bandwidth entered is where |G| crosses 0 dB.
- 02The model approximates the circuit it draws: it takes Z(s) = (1 + s·R₁C₁)/(s·C₁·(1 + s·R₁C₂)), using C₁ where the circuit has C₁ + C₂ and R₁·C₂ for the pole’s time constant where the circuit has R₁·C₁C₂/(C₁ + C₂), so the circuit’s pole sits higher than the model’s, by 10 % at C₂ = C₁/10.
- 03C₂ = γ·C₁ with γ = 0.1, or 0.8 of the largest ratio that can still reach the margin where that is smaller (above about 51°); ω_c·T₁ then solves atan(ω_c·T₁) − atan(γ·ω_c·T₁) = φ exactly, with T₁ = R₁C₁, and C₁ makes |G(jω_c)| = 1.
- 04The charge pump, divider and VCO are ideal and linear: the phase detector’s sampling delay, the VCO’s input capacitance, R₁’s noise and the variation of K_vco over the tuning range are not modelled, and a loop bandwidth above a tenth of the reference frequency is flagged for reference spurs.
- 05With a part series chosen, C₁, R₁ and C₂ are each rounded to the nearest value of that series, and the page finds where the loop the rounded parts build crosses 0 dB, in the same model, and its phase margin there, atan(ω·R₁C₁) − atan(ω·R₁C₂); the BOM and KiCad export are the rounded parts’.
Worked example
2.4 GHz synthesiser, 100 kHz loop, E24 parts
A 2.4 GHz synthesiser divides by N = 48 to a 50 MHz reference; its VCO tunes at 50 MHz/V and its charge pump gives 1 mA. Design the filter for a 100 kHz loop bandwidth and 50° of phase margin and round it to E24: where does the loop the rounded parts build reach unity gain, and with what margin?
| Step | Expression | Result |
|---|---|---|
| Output frequency | 48 × 50 MHz | 2.4 GHz |
| Phase margin as a tangent | tan(50°) | 1.1918 |
| Largest pole ratio the margin allows | (b − √(b² − 4)) / 2, b = 2 + 4 × 1.1918² | 0.1325 |
| Pole ratio γ = C₂/C₁ | min(0.1, 0.8 × 0.1325) | 0.1 |
| x = ω_c·T₁ from the phase margin | 0.1 × 1.1918 x² − (1 − 0.1) x + 1.1918 = 0 | 5.839 |
| Zero time constant T₁ | 5.839 / (2π × 100 kHz) | 9.294 µs |
| C₁ for the crossover | 1 mA × 50 MHz/V / ((2π × 100 kHz)² × 48) × √(1 + 5.839²) / √(1 + (0.1 × 5.839)²) | 13.5 nF |
| C₁, E24 | E24(13.5 nF) | 13 nF |
| R₁ = T₁/C₁ | 9.294 µs / 13.5 nF | 688.5 Ω |
| R₁, E24 | E24(688.5 Ω) | 680 Ω |
| C₂ = γ·C₁ | 0.1 × 13.5 nF | 1.35 nF |
| C₂, E24 | E24(1.35 nF) | 1.3 nF |
| Crossover as built | |G(j2πf)| = 1 | 100.1 kHz |
| Phase margin as built | atan(2π × 100.1 kHz × 680 Ω × 13 nF) − atan(2π × 100.1 kHz × 680 Ω × 1.3 nF) | +50.7° |
How it works
The PLL Loop Filter Calculator computes component values for Type-2 phase-locked loop filters — essential for frequency synthesizer design, clock recovery circuits, and communication system development. IC designers, RF engineers, and embedded developers use this to achieve target loop bandwidth and phase margin. Per Best 'Phase-Locked Loops: Design, Simulation, and Applications' (6th ed., McGraw-Hill) and Banerjee 'PLL Performance, Simulation, and Design' (5th ed.), Type-2 PLL uses charge pump with passive RC filter providing second-order response. PLL noise specifications relevant to communication systems follow ITU-R SM.1538 and IEEE Standard 1139-2008 (IEEE Standard Definitions of Physical Quantities for Fundamental Frequency and Time Metrology — Random Instabilities). Loop bandwidth omega_c determines lock time (t_lock ~ 2*pi/omega_c) and phase noise filtering — wider bandwidth tracks input faster but passes more reference noise. Phase margin phi_m controls overshoot: 45 deg yields 23% overshoot, 65 deg yields 5%. Per Banerjee "PLL Performance, Simulation, and Design" (5th ed.), optimal phi_m = 48-55 deg balances speed and stability. Component equations, with Kvco in Hz/V: C1 = Icp*Kvco/(omega_c^2*N) * sqrt(1 + (omega_c*T1)^2)/sqrt(1 + (omega_c*T3)^2), R1 = T1/C1 and C2 = gamma*C1, where T3 = gamma*T1, gamma is 0.1 (smaller above about 51 deg of phase margin) and omega_c*T1 solves arctan(omega_c*T1) - arctan(gamma*omega_c*T1) = phi_m.
Practical tips
- Per Best, target 48-55 deg phase margin for optimal settling time vs. stability tradeoff
- Use C2 = C1/10 or less for adequate pole spacing (the calculator reduces it above about 51 deg of phase margin) — a larger C2 increases reference spur rejection but costs phase margin, per Banerjee
- Verify loop stability with Bode plot simulation — 6 dB gain margin minimum per control theory standards
- For low phase noise, minimize R1: thermal noise contribution = 4kTR1*Kvco^2/(omega_c*N)^2 per Egan
Common mistakes
- Neglecting phase margin impact on settling — 30 deg margin causes 50% overshoot and 5x longer lock time vs. 55 deg
- Forgetting that the design model approximates the circuit — the circuit's time constants are tau1 = R1*C1 and tau2 = R1*C1*C2/(C1+C2) with C1 + C2 integrating, where the model takes R1*C2 and C1; at C2 = C1/10 the real pole sits 10% higher, so check the final loop in a simulator
- Overlooking VCO gain variation — Kvco can vary 2:1 over tuning range, degrading phase margin by 20 deg per Texas Instruments SCAA030
Frequently asked questions
2 questions
What determines PLL loop filter performance?
Per Banerjee: (1) Bandwidth omega_c sets lock time (t = 10/omega_c typical) and noise filtering. (2) Phase margin controls overshoot — 45 deg = 23%, 55 deg = 14%, 65 deg = 5%. (3) Component tolerance — 10% capacitor causes +/-5 deg margin shift. (4) R1 thermal noise adds to output phase noise floor.
Why use a Type-2 passive loop filter?
Per Best: Type-2 provides second-order response with zero at omega_z = 1/(R1*C1) ensuring stable tracking. Passive implementation (no op-amp) eliminates active device noise contribution. Charge pump interface sources/sinks current, simplifying filter design. Standard for all modern fractional-N synthesizers per Analog Devices, TI, and Maxim reference designs.
Shop Components
As an Amazon Associate we earn from qualifying purchases.