Radar RF Component Selection

“Radar RF components” is not a catalogue. It is a frequency plan with a timeline attached. The same mixer, filter and oscillator families that show up in test equipment show up here; the waveform decides which constraint binds first. This page is the mapping. It sits in applications, where the binding constraint comes from the mission rather than from the part; the device pages live under components and YIG technology, and the frequency-plan identities live in the signal chain section.

The LO: range, speed, purity

Waveform constraint	What binds	Candidate to validate
Multi-octave coverage, sweep or search	Tuning range and linearity	YIG-tuned oscillator
Pulse-to-pulse hop across a wide band, microsecond class	Hop time	Start with VCO / fast synthesizer; verify installed timing
Pulse-to-pulse hop across a wide band, millisecond class	Hop time still fits a coil	YIG-tuned oscillator
Fixed channel, long coherent dwell, Doppler	Close-in phase noise	DRO, often locked
Wide instantaneous bandwidth	Far-out floor, not the close-in number	Quiet fundamental source; watch multiplied chains

Investigation routes, not family-wide guarantees. Check carrier, offsets, reference and lock state on the installed assembly.

The front end as a constraint chain

A typical receive path is antenna → limiter → LNA → preselector → mixer ← LO → IF. A typical exciter path is the reverse translation: IF or a clean reference, upconverted to the radiating RF. Downconversion and upconversion are those two directions. They often share the LO.

Three numbers then decide the parts:

1. How far the RF must move, and how fast.
2. Which offset frequencies the waveform integrates — Doppler, clutter, a jammer, the image.
3. How much power sits on the transmit path and how much can arrive at the receive path.

Write those down before you open a datasheet. The rest of this page is those three numbers mapped onto device classes.

The comparison behind that table is YIG oscillator vs VCO vs DRO. A YTO is worth investigating for wide coverage when its coil and driver response fit. Coverage alone does not establish pulse-to-pulse agility; check the model’s documented transition conditions.

Close-in purity on a fixed channel is a DRO problem, for the same reason a high-performance clock uses a dielectric resonator rather than a wide-tuning MMIC VCO: the resonator Q is the specification, not the tuning range on the front of the brick.

Phase noise is a radar specification

L(f) without an offset is not a number. In radar the offset is chosen by the waveform:

• Close-in (choose offsets from the coherent-processing requirement) sets how long you can integrate coherently and how fine a Doppler bin you can believe.
• Mid-range is where a strong nearby return or a jammer reciprocal-mixes into the IF. A tracking preselector is aimed at this as much as at the classical image.
• Far-out sets the noise that a wide chirp or a wide digital IF will integrate.

The relevant offsets change with the waveform; compare plots at the specified carrier and lock state rather than adopting a universal close-in cutoff.

The preselector: move it, or machine it

If the LO walks, the image walks. The image is 2 × IF from the wanted RF, and an ordinary single-path IF filter cannot separate it once it overlaps the wanted response. A tracking YIG filter is the part that can walk with the LO and still present tens of megahertz of absolute width. A cavity is the part you want when the RF does not walk: lower loss, far higher power handling, no coil.

Two radar-specific catches:

Readiness follows the command timeline. Example, assumed inputs: the oscillator and filter each take 2 ms from their own command to enter and remain within their respective project-defined frequency-error and passband-alignment tolerances. If both start at t = 0, both are ready at t = 2 ms. If the filter starts only at t = 2 ms, both are ready at t = 4 ms. For partial overlap, use the later measured completion time. These illustrative times and tolerance conditions are assumptions, not device specifications.

Limiting is not a damage rating. Micro Lambda’s public bandpass-filter explanation, section 1 describes limited RF transfer through a YIG sphere. Without a model and operating condition, its example range is not retained here as a limit. Check the power at the actual filter input, including prior gain, separately from LNA protection and absolute maximum ratings.

Transmit filters are almost never YIG. Power handling and a known radiating frequency point at a cavity or a waveguide filter. A radar can honestly use both: YIG on the receive tune, cavity on the transmit harmonic.

Converters follow the plan

The receive converter is a downconverter. The exciter, if the waveform is generated at IF, is an upconverter. They are opposite directions of one identity, and they usually share the LO.

If the converter brick has an embedded oscillator, you have inherited that oscillator’s family — VCO, DRO or something less named — and the phase-noise plot on the brick is the one that matters. If you supply the LO, the converter is filters plus a mixer, and the source comparison stays yours.

Image rejection on the receive side is the preselector’s job unless the mixer is an image-reject topology. Do not assign it to the IF filter. The identity does not allow it.

A compact selection map

Job	Wideband / hopping	Fixed channel
LO	YIG-tuned oscillator	DRO or locked narrow source
Receive preselect	Tracking YIG or a bank meeting coverage and rejection	Bandpass cavity or ceramic centred on wanted RF
Transmit filter	Cavity / waveguide — still	Cavity / waveguide
Converter	External LO into a down/up mixer	Often a brick with an embedded quiet LO
What you measure	Later LO/filter completion; L(f) at required offsets	L(f) at required offsets; residual FM; isolation

‘Hopping’ here means hops the coil can finish. Microsecond-class hops leave this table on the VCO / fast-synthesizer row.

If the answers say “wide and clean, and we can wait milliseconds”, start with the YIG oscillator and YIG filter pages. If they say “one frequency, integrate a long time”, start with the DRO side of the source comparison and a cavity. The word radar does not pick for you.

Record the waveform requirements

Use one record for timing and spectral acceptance. Investigate frequency converters under this same plan; calculate images with the RF, IF and LO guide.

Input / check	Waveform requirement	Assembly evidence / result
RF coverage / IF / LO side	Fill: bands, channel or chirp width, image	Fill: usable source/filter coverage
Transition commands	Fill: source and filter command timestamps	Fill: simultaneous, sequential or measured overlap
Readiness tolerances	Fill: frequency error, passband alignment and recovery	Fill: later completion against the deadline
Noise conditions	Fill: carrier, offsets, reference and lock state	Fill: noise plot and waveform budget
Wanted signal / blocker	Fill: levels and offsets at each reference plane	Fill: overload, image and reciprocal-mixing results
Receive / transmit power	Fill: gain before filter, limiter and maximum levels	Fill: separate linearity, limiting and damage checks

Takeaway: Validate the source, filter and converter against one timeline and noise budget, with receive and transmit power recorded separately.