PLL Loop Filter Calculator

Design the passive second- or third-order loop filter of a charge-pump PLL from the charge-pump current, VCO gain, divider, loop bandwidth and phase margin, then check the crossover and phase margin from the actual parts, rounded to E24 or E12 values.

Calculator Electronics Updated Oct 15, 2026
How to Use
  1. Pick a 2nd-order filter (R2, C1, C2) or a 3rd-order one, which adds R3 and C3 to filter reference spurs.
  2. Enter the charge-pump current Icp in mA and the VCO gain in MHz/V from the datasheets, at the tuning voltage you will lock at.
  3. Enter the output frequency and the phase-detector frequency; the divider N is their ratio.
  4. Enter the loop bandwidth and phase margin. Gamma stays at 1 unless you are following Banerjee’s optimised designs; for 3rd order also enter the pole ratio T31.
  5. Read C1, C2 and R2 (and C3, R3). The Bode plot shows where the gain crosses 0 dB and the phase margin there, re-computed from the parts; the dashed curve uses the nearest E24 or E12 values.
PLL & loop targets
mA
MHz/V
MHz
kHz
kHz
degrees
optional, 1
T3 / T1, 0–1
Presets
Loop filter & open-loop response
C1
—
C2
—
R2
—
Divider N
—
Crossover, E24 parts
—
Phase margin, E24 parts
—

Worked Example

Banerjee’s second-order example (ch. 38, Appendix A). 1960 MHz from a 50 kHz phase detector, so N = 39,200. Icp = 1 mA, Kvco = 60 MHz/V, loop bandwidth 10 kHz, phase margin 49.2°, gamma 1.024.

ωc = 2π × 10 kHz = 62,832 rad/s, tan 49.2° = 1.1584.
T1 = [√((2.024 × 1.1584)² + 4 × 1.024) − 2.024 × 1.1584] / (2 × 62,832) = 5.989 µs.
T2 = 1.024 / (62,832² × 5.989 µs) = 43.31 µs.
A0 = (1 mA × 60 MHz/V) / (62,832² × 39,200) × √((1 + 2.721²) / (1 + 0.3763²)) = 0.3877 nF × 2.713 = 1.052 nF.
C1 = A0 × T1 / T2 = 0.145 nF; C2 = A0 − C1 = 0.906 nF; R2 = T2 / C2 = 47.78 kΩ. These are the book’s own results (0.145 nF, 0.906 nF, 47.776 kΩ).
Rounded to 150 pF, 910 pF and 47 kΩ, the loop crosses at 9.84 kHz with 48.8° of phase margin. The coarse lock-time rule 4 / fc gives about 400 µs.
At fPFD / 5 this loop is wider than the fPFD / 10 that Banerjee’s chapter 29 gives as the limit of the continuous model, so the calculator flags it: the sampled phase detector will leave somewhat less than 49.2° in practice.

Banerjee’s third-order example (ch. 39, Appendix A). 1392 MHz from 60 kHz (N = 23,200), 4 mA, 30 MHz/V, 2 kHz, gamma 1.136, T31 = 0.6. The book prints the phase margin as 47.1°, but its time constants (T1 = 20.333 µs, T2 = 221.12 µs, T3 = 12.200 µs) correspond to 47.16°, so the preset uses 47.16°. Solving for T1 numerically gives C1 = 6.581 nF, C2 = 85.59 nF, C3 = 0.466 nF, R2 = 2.584 kΩ and R3 = 33.88 kΩ, against the book’s 6.5817 nF, 85.5896 nF, 0.4660 nF, 2.5835 kΩ and 33.8818 kΩ.

The common mistake: dividing by 2π once. The phase-detector gain is Icp / 2π per radian, but the VCO gain in rad/s is 2π × Kv, so the two cancel. Use Icp / 2π with Kv still in MHz/V and the first example becomes C1 = 23.2 pF, C2 = 144 pF and R2 = 300 kΩ. With the real loop gain those parts cross over at 37.0 kHz, not 10 kHz, with only 30.0° of phase margin.

Show Work

Enter the charge pump, VCO, frequencies and loop targets to see the working.

Formulas

Pole T1, 2nd order (Banerjee Eq. 38.7)
T1 = [√((1+γ)² tan²φ + 4γ) − (1+γ) tanφ] / (2ωc)
With γ = 1 this is (sec φ − tan φ) / ωc
Zero T2 (Eq. 38.6, 39.6)
T2 = γ / (ωc² (T1 + T3))
T3 = 0 for second order; T3 = T31 × T1 for third order
Phase margin, 3rd order (Eq. 39.8)
φ = atan(ωc T2) − atan(ωc T1) − atan(ωc T3)
Solved for T1 numerically; the closed-form approximation is T1 ≈ (sec φ − tan φ) / (ωc (1 + T31))
Total capacitance (Eq. 38.8, 39.13)
A0 = Icp Kvco / (ωc² N) × √((1 + ωc²T2²) / ((1 + ωc²T1²)(1 + ωc²T3²)))
Icp in A, Kvco in Hz/V: the 2π of Kpd = Icp/2π cancels the 2π of Kvco in rad/s
2nd-order parts (Eq. 38.9–38.11)
C1 = A0 T1/T2,  C2 = A0 − C1,  R2 = T2 / C2
Equivalently C2 = C1 (T2/T1 − 1)
3rd-order parts (Eq. 39.20–39.24)
C1 = A2/T2² (1 + √(1 + T2/A2 (T2 A0 − A1)))
A1 = A0(T1+T3), A2 = A0 T1 T3; C3 = (−T2²C1² + T2 A1 C1 − A2 A0) / (T2²C1 − A2), C2 = A0 − C1 − C3, R2 = T2/C2, R3 = A2 / (C1 C3 T2)
Open-loop gain (the check)
G(s) = Icp Kvco Z(s) / (N s)
Z(s) = Vtune / Icp of the actual network; crossover where |G| = 1, phase margin = 180° + arg G there
Lock time (Eq. 28.5, 28.6, 28.26)
tlock ≈ 4 / fc,  ωn = √(Icp Kvco / (N A0)),  ζ = T2 ωn / 2
A coarse rule of thumb; peak time ≈ 0.8 / fc, overshoot ≈ frequency jump / 3

How a Charge-Pump PLL Loop Filter Works

A phase-locked loop frequency synthesizer compares the reference, divided down to the phase-detector frequency fPFD, with the VCO output divided by N. The phase-frequency detector drives a charge pump, which pushes or pulls a current Icp for as long as the two edges disagree. The loop filter turns those current pulses into the VCO’s tuning voltage, and when the loop settles the output is exactly N × fPFD.

The charge pump into a capacitor is already an integrator, and the VCO is a second one (frequency in, phase out), so the loop has two poles at zero frequency and 180° of phase shift. R2 in series with C2 adds a zero that lifts the phase back up near the loop bandwidth: that lift is the phase margin. C1 smooths the current pulses and adds a pole above the bandwidth; a third-order filter adds R3 and C3 for another pole that attenuates the reference spur further out. The loop bandwidth fc is defined here, as in Banerjee, as the frequency where the open-loop gain falls to 1 (0 dB).

The design equations come from Dean Banerjee’s PLL Performance, Simulation, and Design (5th edition, Texas Instruments SNAA106C, 2017; earlier editions at National Semiconductor), which builds on William Keese’s application note AN-1001 (National Semiconductor, now published by Texas Instruments). Setting the phase margin at fc and the slope of the phase there gives closed-form time constants for second order; for third order Banerjee solves the phase equation numerically and chooses C1 to make C3 as large as possible, which keeps the VCO’s input capacitance and R3’s thermal noise from mattering.

Phase marginDamping factor ζNatural frequency ωn
30°0.6580.760 ωc
40°0.7320.683 ωc
45°0.7770.644 ωc
50°0.8290.603 ωc
60°0.9660.518 ωc
70°1.1910.420 ωc

From Banerjee’s Table 28.1 (second-order filter, γ = 1); the calculator reproduces each row from ωn = √(Icp Kvco / (N A0)) and ζ = T2 ωn / 2.

What the model leaves out. The equations treat the charge pump as a continuous current, which Banerjee says holds while fc is between fPFD/100 and fPFD/10; the tool warns outside that range. They also ignore the VCO’s input capacitance (it adds to C3, or to C1 in a second-order filter), charge-pump leakage and mismatch, and the variation of Kvco across the tuning range, which moves fc in proportion. The lock-time figure is Banerjee’s coarse rule of thumb, not a transient simulation.

About This Calculator

This calculator designs the passive loop filter of a charge-pump PLL synthesizer, second order (R2, C1, C2) or third order (adding R3 and C3), from the charge-pump current, VCO gain, output and phase-detector frequencies, loop bandwidth and phase margin, using Banerjee’s published equations including the gamma parameter. It reproduces both of the book’s worked examples.

Instead of trusting the design equations, it rebuilds the open-loop gain from the resistor and capacitor network, finds where it crosses 0 dB and the phase margin there, and does it again with the nearest E24 or E12 parts so you can see what rounding costs. It draws the filter and a Bode plot of magnitude and phase, warns when the bandwidth is too wide for the phase-detector frequency or the phase margin is low, and gives a lock-time estimate. All calculations run in your browser; nothing you enter is sent anywhere.

It is for RF and mixed-signal engineers laying out a synthesizer, students learning control loops through PLLs, and anyone checking the loop filter values on an evaluation board or an old design.

Related calculators: Phase Noise to Jitter Calculator (for the integrated jitter of the locked output), Harmonics & Spur Calculator and Colpitts Oscillator Calculator.

Frequently Asked Questions

How do I choose the PLL loop bandwidth?

It is a trade-off. A wider loop locks faster (lock time is roughly 4 / fc, so 10 kHz gives about 400 µs) and suppresses more VCO phase noise close to the carrier; a narrower loop filters more reference noise and spurs. Keep fc between fPFD/100 and fPFD/10: Banerjee notes the continuous model these equations use only holds in that range, and below fPFD/100 cycle slipping tends to occur.

What phase margin should a PLL loop filter have?

Most designs use 40° to 60°. Banerjee’s simulations show the fastest lock at about 48°. At 30° the damping factor falls to 0.66, so each frequency step overshoots and rings; above about 70° the loop is overdamped, locks more slowly and needs larger capacitors.

When should I use a third-order loop filter?

When the reference spur or other noise is ten or more times the loop bandwidth away, a third pole from R3 and C3 attenuates it further. In Banerjee’s example (2 kHz loop, 60 kHz phase detector) the R3–C3 pole alone adds 13.5 dB at 60 kHz. The cost is a little phase margin to share and thermal noise from R3, and C3 should be at least four times the VCO’s input capacitance.

Do I need to divide the charge-pump current by 2π?

Only if you also give the VCO gain in rad/s per volt. The phase-detector gain is Icp / 2π amps per radian and the VCO gain is 2π × Kv radians per second per volt, so the 2π cancels and Icp × Kv (MHz/V) is what matters. Divide by 2π once and keep Kv in Hz/V, and the capacitors come out 6.28 times too small: in Banerjee’s example the real loop then crosses at 37 kHz instead of 10 kHz with only 30° of margin.

What is gamma in Banerjee’s loop filter equations?

Gamma (γ) sets how the zero and the pole sit around the loop bandwidth: T2 = γ / (ωc² × (T1 + T3)). With γ = 1 the phase margin peaks exactly at the loop bandwidth, the classic design. Banerjee tunes it slightly above 1 for faster lock, for example 1.024 with 49.2° in his second-order example; it changes the parts by only about 1%.

How do I use the PLL Loop Filter Calculator?

Simply type your numbers and read the result, which refreshes the instant you change something. There is nothing to submit and nothing to wait for.

Does it cost anything or need an account?

No. The tool is completely free, there is no account to create, and it keeps working offline after the page first loads.

Is anything I type uploaded?

No. The tool works entirely on your device, so the values you enter never leave your browser.

Common Use Cases

Integer-N RF synthesizer

A 2.4 GHz output from a 1 MHz phase detector (N = 2400), 2.5 mA and a 40 MHz/V VCO with a 100 kHz, 50° loop needs C1 38.4 pF, C2 252 pF and R2 17.4 kΩ; with 39 pF, 240 pF and 18 kΩ it crosses at 102 kHz with 49°.

Narrowband channel synthesizer

Banerjee’s 1960 MHz example steps in 50 kHz channels with a 10 kHz, 49.2° loop: C1 145 pF, C2 906 pF, R2 47.8 kΩ, and a lock time of roughly 400 µs.

Reference spur suppression

A third-order filter with T31 = 0.6 on a 2 kHz loop and 60 kHz phase detector adds R3 33.9 kΩ and C3 466 pF, which on their own cut the 60 kHz spur by a further 13.5 dB.

Checking an existing board

Re-solve the loop with the parts actually fitted: rounding Banerjee’s 145 pF, 906 pF and 47.8 kΩ to 150 pF, 910 pF and 47 kΩ moves the crossover to 9.84 kHz and the phase margin to 48.8°.

Debugging a ringing PLL

A loop designed for 30° phase margin has a damping factor of 0.66 and overshoots on every channel change. Redesigning for 45–50° at the same bandwidth trades a little filtering for a clean settle.

Last updated: