L
Admin
The Vivaldi, or exponentially tapered slot antenna, is the default choice when you need a multi-octave end-fire element on a PCB: UWB radar, EMC probes, wideband arrays and direction finding. It looks forgiving, but the usable bandwidth is usually limited by the feed transition, not by the taper.
1. How it radiates
A slotline is opened gradually along an exponential curve. Each frequency radiates from the region of the taper where the slot width is roughly half a wavelength, so the active region (and the phase centre) slides toward the throat as frequency rises. That sliding phase centre is fine for a single probe and annoying for an array or a time-domain system.
2. Sizing the taper
- Aperture width at the mouth of about lambda/2 at the lowest frequency is the usual starting point; go narrower and the low end rolls off quickly.
- Taper length of one wavelength or more at fmin gives smoother impedance and higher gain; short tapers ripple.
- The opening rate R in y = c1 * exp(R*x) + c2 trades beamwidth against match. Sweep it in EM rather than trusting a single formula.
3. The feed decides the bandwidth
The classic feed is a microstrip-to-slotline transition: the microstrip crosses the slot and ends in a radial stub, while the slot ends behind the crossing in a circular (or rectangular) cavity. The stub acts as a broadband RF short and the cavity as a broadband open. If either one resonates inside your band, you will see a notch in S11 that no amount of taper tuning removes. Check the transition on its own (back-to-back) before blaming the taper.
4. Antipodal and balanced variants
- Antipodal Vivaldi puts the two flares on opposite sides of the substrate and feeds directly from microstrip, avoiding the slotline balun. Bandwidth is excellent, but cross-polarisation rises at the high end because the flares are not coplanar.
- Balanced antipodal versions add a third metal layer to cancel that cross-pol at the cost of a stripline-style stack.
5. Practical tweaks
- Corrugations or slots on the outer edges lengthen current paths and improve low-frequency gain and front-to-back ratio.
- A lens or a dielectric director in front of the mouth tightens the beam at the high end.
- Keep the substrate thin and low-loss; a thick high-Dk board launches substrate modes that tilt the beam.
6. Bench checklist
- Measure S11 and realised gain versus frequency, not only S11; a good match with a wandering beam is common.
- Plot group delay or the time-domain impulse response if the antenna is for UWB or radar.
- Note the phase-centre shift if you are calibrating a range or an array.
Questions for the board: which feed transition has given you the cleanest low-frequency end, and how do you handle the phase-centre drift in arrays?
1. How it radiates
A slotline is opened gradually along an exponential curve. Each frequency radiates from the region of the taper where the slot width is roughly half a wavelength, so the active region (and the phase centre) slides toward the throat as frequency rises. That sliding phase centre is fine for a single probe and annoying for an array or a time-domain system.
2. Sizing the taper
- Aperture width at the mouth of about lambda/2 at the lowest frequency is the usual starting point; go narrower and the low end rolls off quickly.
- Taper length of one wavelength or more at fmin gives smoother impedance and higher gain; short tapers ripple.
- The opening rate R in y = c1 * exp(R*x) + c2 trades beamwidth against match. Sweep it in EM rather than trusting a single formula.
3. The feed decides the bandwidth
The classic feed is a microstrip-to-slotline transition: the microstrip crosses the slot and ends in a radial stub, while the slot ends behind the crossing in a circular (or rectangular) cavity. The stub acts as a broadband RF short and the cavity as a broadband open. If either one resonates inside your band, you will see a notch in S11 that no amount of taper tuning removes. Check the transition on its own (back-to-back) before blaming the taper.
4. Antipodal and balanced variants
- Antipodal Vivaldi puts the two flares on opposite sides of the substrate and feeds directly from microstrip, avoiding the slotline balun. Bandwidth is excellent, but cross-polarisation rises at the high end because the flares are not coplanar.
- Balanced antipodal versions add a third metal layer to cancel that cross-pol at the cost of a stripline-style stack.
5. Practical tweaks
- Corrugations or slots on the outer edges lengthen current paths and improve low-frequency gain and front-to-back ratio.
- A lens or a dielectric director in front of the mouth tightens the beam at the high end.
- Keep the substrate thin and low-loss; a thick high-Dk board launches substrate modes that tilt the beam.
6. Bench checklist
- Measure S11 and realised gain versus frequency, not only S11; a good match with a wandering beam is common.
- Plot group delay or the time-domain impulse response if the antenna is for UWB or radar.
- Note the phase-centre shift if you are calibrating a range or an array.
Questions for the board: which feed transition has given you the cleanest low-frequency end, and how do you handle the phase-centre drift in arrays?