How do I measure the settling time of a frequency synthesizer after a frequency hop?
Synthesizer Settling Time Measurement
Settling time is one of the most important specifications for agile frequency synthesizers because it determines: the maximum hop rate (hops per second = 1/settling_time), the system throughput (time spent settling is time not spent communicating or sensing), and the vulnerability window (in military FHSS: the signal is detectable and jammable during the settling transient).
Typical Settling Times
- Integer-N PLL: 50-500 μs (limited by the loop bandwidth, which must be < reference_freq/10)
- Fractional-N PLL: 10-100 μs (wider loop bandwidth possible)
- DDS (Direct Digital Synthesis): < 1 μs (essentially instantaneous, limited only by the DAC settling)
- PLL + DDS hybrid: 1-10 μs (DDS for fine steps, PLL for coarse steps)
PLL settling: t_settle ≈ -ln(Δf_tolerance/Δf_step)/(ζ×ω_n)
Where ζ = damping factor, ω_n = natural frequency
For ζ=0.707, ω_n=2π×10kHz: t_settle ≈ 50 μs (to ±1 ppm)
DDS settling: t_settle ≈ 1/f_clock (pipeline delay)
Frequently Asked Questions
What equipment is best?
For fast synthesizer settling time measurement: real-time spectrum analyzer (Keysight UXA or MXA with RTSA option, R&S FSW with RTSA, Tektronix RSA5000): directly displays the frequency vs. time during the hop. Most intuitive and informative. Cost: $30,000-100,000. Oscilloscope + FM discriminator: a high-bandwidth oscilloscope (2+ GHz) with a wideband FM discriminator (commercial or built from a mixer + delay line). Cost: $10,000-50,000+ (mainly the scope). For simple measurements: a spectrum analyzer in zero-span mode (max hold at the target frequency) can approximate the settling time, but lacks the time resolution of the methods above.
How do I reduce settling time?
Design techniques for faster settling: increase the PLL loop bandwidth (faster settling but: more reference spur leakage and higher phase noise). Use a fractional-N PLL (wider loop bandwidth is possible because the reference frequency can be higher). Use frequency pre-steering (the VCO tuning voltage is pre-set to the expected value for the target frequency using a DAC; the PLL only corrects the residual error, settling much faster). Use a DDS (near-instantaneous frequency changes, but: limited in frequency range and spurious performance). Use a PLL + DDS hybrid (the DDS handles fine frequency steps within the PLL bandwidth; the PLL handles coarse steps).
What about spectral purity during the hop?
During the frequency transition: the synthesizer output sweeps through intermediate frequencies, generating wideband spectral content. This transient spectrum: can interfere with other channels (in a crowded spectrum environment), reduces the effective signal-to-noise ratio during the transition, and may violate spectral mask requirements. In military FHSS systems: the transient spectrum is called 'splatter' and must be minimized. Techniques: fast settling (shorter transient duration means less total splatter energy), output blanking (an RF switch turns off the output during the hop transient, eliminating splatter but creating a gap in the signal), and shaped hopping (controlling the VCO tuning voltage trajectory to minimize the transient spectral width).