Analog

PLL design basics

By CourseTron Team ·  7 min read ·  Updated Oct 2026 💬 𝕏 in f
Table of Contents
    🖨  Download / Print this guide (PDF)

    In short: A phase-locked loop is a feedback system that forces a local oscillator's output to track the phase and frequency of a reference. Good pll design means balancing five blocks — phase/frequency detector, charge pump, loop filter, VCO, and feedback divider — so the loop locks quickly, stays stable, and keeps jitter and spurs low.

    What Is pll design?

    At its heart, pll design is the art of closing a loop so that one oscillator's phase becomes a faithful, scaled copy of another's. The reference might be a clean crystal at a few megahertz; the output might be a multi-gigahertz clock feeding a SerDes or a radio synthesizer. The PLL bridges that gap by continuously comparing phases and nudging the output until the error settles to a constant.

    Because it is a feedback control system, a PLL is governed by the same mathematics as any servo loop: loop gain, poles and zeros, phase margin, and settling time. The engineering challenge is that those control parameters are set by physical circuit blocks — a charge-pump current, a capacitor, a VCO gain — each with its own non-idealities. Getting them to cooperate is what separates a locked, low-jitter clock from one that hunts, drifts, or spews reference spurs.

    Key fact: For a classic charge-pump PLL, the output frequency is simply fout = N × fref, where N is the feedback divide ratio. The reference sets the resolution grid; the divider sets the multiplication. This single relationship drives almost every architectural trade-off in pll design.

    The Five Blocks of a Charge-Pump PLL

    The dominant modern topology is the charge-pump PLL. Signal flows around the loop like this:

    Charge-pump PLL: five blocks in one loopReferencef_refPhase/FreqDetectorChargePumpLoopFilterVCODivider÷ Nf_outfeedbackCourseTron · coursetron.com
    The classic charge-pump PLL: the divided output is compared against the reference, and the error steers the VCO.

    Phase/frequency detector (PFD)

    The PFD compares the reference edge with the divided feedback edge and emits two pulses, UP and DOWN, whose widths encode the sign and size of the phase error. Unlike a simple XOR detector, a tri-state PFD also senses frequency difference, so it pulls the loop into lock even from a cold start without relying on a lucky initial phase.

    PFD timing: REF leads FB, so UP pulsesREFFBUPDOWNUP width = REF-to-FB gapDOWN stays lowCourseTron · coursetron.com
    With the reference edge leading the feedback edge, the PFD asserts UP for the gap between them.

    Charge pump

    The charge pump converts those digital UP/DOWN pulses into charge delivered to the loop filter. UP sources current, DOWN sinks it. In steady-state lock the two pulses are nearly equal and the net charge per cycle is zero, holding the control voltage constant. Mismatch between the UP and DOWN currents is a leading cause of reference spurs, so current matching is a core concern in pll design.

    Loop filter

    The loop filter integrates the charge-pump current into a control voltage and sets the loop dynamics. A second-order passive filter — a series resistor and capacitor in parallel with a smaller shunt capacitor — provides an integrator (for zero steady-state phase error), a stabilizing zero (for phase margin), and a high-frequency pole (to suppress ripple). Its component values directly set loop bandwidth and damping.

    Voltage-controlled oscillator (VCO)

    The VCO turns control voltage into output frequency with a gain KVCO in Hz/V (or rad/s/V). It is the loop's integrator from frequency to phase and usually the dominant jitter source at high offset frequencies. Its tuning range must cover the target band across process, voltage, and temperature, which is why real designs use banks of switched capacitors plus a continuous varactor.

    Feedback divider

    The ÷N divider scales the fast output down to the reference rate for comparison. In an integer-N synthesizer N is a whole number; in a fractional-N design a delta-sigma modulator dithers N to synthesize fine frequency steps, trading quantization noise for resolution.

    How Lock Happens

    When power is applied the output frequency is wrong, so the PFD produces lopsided UP/DOWN pulses. The charge pump drives the control voltage up or down, the VCO frequency slews toward target, and the phase error shrinks. The watch-the-packet view below shows a correction signal circulating until the error vanishes.

    Reaching lock: the correction circulatesPFD +Charge pumpLoopfilterVCODivider÷ Ncontrol voltage settles, phase error goes to 0lock progressCourseTron · coursetron.com
    A correction travels around the loop each reference cycle; the bar tracks the loop approaching lock.

    Lock is reached when the divided feedback edge lines up with the reference edge and the control voltage stops moving. The time this takes is the settling time, set mainly by loop bandwidth. The loop stays locked by continuously correcting tiny phase slips caused by noise and supply disturbance.

    The Core Trade-Offs

    • Loop bandwidth. Wider bandwidth locks faster and suppresses VCO phase noise, but passes more reference noise and spurs. A common rule of thumb keeps bandwidth below about one-tenth of the reference frequency for stability.
    • Phase margin. The loop-filter zero sets damping; 45° to 60° of phase margin is the usual target to avoid ringing and overshoot during lock.
    • Reference spurs. Charge-pump current mismatch, leakage, and filter ripple modulate the VCO at the reference rate, creating spurs offset by fref. Current matching and good filtering are the defenses.
    • In-band vs out-of-band noise. Inside the loop bandwidth, reference, PFD, charge pump, and divider noise dominate (multiplied by N). Outside it, the VCO dominates. The bandwidth choice decides which noise floor wins.
    Learn this hands-on

    Real projects and industry tool flows, with certification.

    Explore CourseTron Courses →

    pll design — FAQ

    What is loop bandwidth and why does it matter?

    Loop bandwidth is the frequency up to which the PLL actively corrects phase error. It governs settling speed and the crossover between reference-dominated and VCO-dominated noise. Choosing it is the single most consequential decision in pll design.

    What causes reference spurs in a PLL?

    Spurs appear at multiples of the reference frequency when periodic disturbances reach the VCO — chiefly charge-pump UP/DOWN current mismatch, control-line leakage, and residual ripple on the loop filter. Matching the pump currents and adding filter poles reduces them.

    What is the difference between integer-N and fractional-N?

    Integer-N multiplies the reference by a whole number, so frequency resolution equals the reference — forcing a low reference for fine steps, which hurts noise. Fractional-N uses a delta-sigma modulator to average a non-integer divide, giving fine resolution with a high reference, at the cost of quantization noise that must be filtered.

    Why use a charge-pump PLL instead of a simple phase detector?

    A tri-state PFD plus charge pump gives an infinite pull-in range, zero steady-state phase error (thanks to the integrator), and clean separation of UP/DOWN logic from analog filtering. That combination makes it the default for frequency synthesis.

    Related: related topic · Glossary · all courses.

    Advertisement