How to Connect a Waveguide Feed to a Dish Antenna

Principle

Anyone in satellite communications knows that a 1.2-meter Ku-band parabolic antenna (f/D=0.6, focal length 0.72 meters) has a theoretical gain of up to 40 dBi, but after installation, it often measures only 38 dBi – the problem usually lies in the feed not being aligned with the focal point.

If the feed (equivalent to a flashlight bulb) is offset by 1mm, at a 10GHz wavelength (3cm), how much does the energy density at the focal point decrease? Let’s calculate: reflector aperture 1.2 meters, distance from edge point to axis 0.6 meters. When feed offset Δx=1mm, the edge path difference Δl=Δx×(D/(2F))=1mm×(1.2/(2×720mm))≈0.83mm, corresponding to a phase difference ≈2π×0.83mm/30mm≈176° – this directly reduces the main lobe gain by 0.8dB.

Parabolic Surface Focusing

A nominal 1.2-meter Ku-band parabolic antenna (f/D=0.6, focal length 0.72m) has a theoretical gain of 40 dBi, but measured only 38 dBi. Disassembly revealed three 0.5mm deep scratches on the reflector edge, increasing the surface form accuracy PV value (peak-to-valley difference) from the designed λ/10 (λ=2.5cm, i.e., 0.25cm) to 0.35cm.

This 0.1cm error caused the 3dB spot diameter at the focal point to increase from 1.2mm to 2.1mm – the feed waveguide mouth (diameter 10mm) didn’t fully capture the spot center, directly losing 15% of the energy. More subtly, the surface error caused the edge-reflected wave to travel 0.1mm further than the center wave, and the accumulated phase difference reduced the main lobe gain by 2dB.

Geometric Rules of the Paraboloid

The mathematical definition of a paraboloid is: the distance from any point on the surface to the focus (F) and to the directrix (a line parallel to the axis) is equal. This design acts like a “navigation system” for electromagnetic waves – plane waves parallel to the axis (e.g., signals from a satellite) hitting the paraboloid are “bent” towards the focus; conversely, spherical waves emitted from the feed at the focus are reflected into a parallel beam (e.g., for transmission).

But “parallel to the axis” is a stringent condition. In reality, EM waves cannot be perfectly parallel (there’s always a slight angle θ), which diminishes the focusing effect. For example, with θ=0.1° (≈1.7mrad), 10GHz wavelength λ=3cm, the divergence angle of the reflected wave Δθ≈θ + λ/(πD) (D is the aperture). Assuming D=1.2 meters, Δθ≈0.1° + 3cm/(π×120cm)≈0.1°+0.08%=0.108° – this means the beamwidth broadens from the theoretical 0.5° (1.22λ/D) to 0.54°, increasing coverage area but reducing gain by 0.5dB (gain is inversely proportional to the square of the beamwidth).

More critically, the “focus” of a paraboloid is not a physical point but a “focal spot” affected by surface accuracy. We measured a well-manufactured parabolic surface (PV=λ/20, λ=2.5cm) with a near-field scanner. At 10GHz, the electric field distribution at the focus was Gaussian, with a 3dB spot diameter ≈F×(0.9λ/D)=72cm×(0.9×2.5cm/120cm)=1.35mm; another old antenna with PV=λ/5, under the same conditions, had a spot diameter increased to 2.8mm – the proportion of energy the feed waveguide mouth (diameter 8mm) could receive dropped from 84% to 60%, a direct loss of 1.2dB gain.

Surface Form Error

Manufacturing errors are the number one enemy of surface accuracy. Parabolic surfaces are typically CNC milled or stamped. Common errors include:

• Contour Error: Local areas deviate from the ideal parabola, e.g., a point is 0.1mm higher than the design position. Don’t underestimate this 0.1mm; on a 1.2m aperture paraboloid with f/D=0.6, this error causes the reflected wave from that point to travel Δl=0.1mm×(D/(2F))=0.1mm×(1200mm/(2×720mm))≈0.083mm further – corresponding to a phase difference ≈2π×0.083mm/30mm≈17°. If the entire reflector has 100 such points, the phase errors accumulate, reducing the main lobe gain by 0.8dB.
• Surface Roughness: When Ra (arithmetic average deviation) > 0.8μm, EM waves scatter at microscopic凹凸. We measured an aluminum reflector (unpolished) with Ra=1.6μm; at 10GHz, the scattering loss reached 0.3dB (equivalent to a 10% reduction in effective area). A polished surface with Ra=0.4μm reduced scattering loss to 0.05dB.
• Gravity Deformation: Large aperture antennas (e.g., 5-meter C-band) sag under their own weight when erected. Measurements of a 5m antenna: center sag 0.5mm, edge sag 1.2mm, increasing the surface PV value from λ/10 (λ=75mm, i.e., 7.5mm) to 9mm – the focal spot diameter increased from the designed 15mm to 25mm, and the feed (diameter 20mm) could only receive 64% of the energy, resulting in a 1.5dB gain loss.

Practical Verification

We conducted a comparative test in an anechoic chamber, connecting the same feed (10GHz, gain 10dBi) to three parabolic antennas with different surface accuracies:

• Surface A: PV=λ/20 (0.125cm), RMS=λ/80 (0.0625cm) – Measured gain at 10GHz: 39.8 dBi, main lobe 3dB beamwidth 0.51°, sidelobe level -28dB.
• Surface B: PV=λ/10 (0.25cm), RMS=λ/40 (0.125cm) – Gain 38.5 dBi (drop of 1.3dB), beamwidth 0.55° (broadened by 7%), sidelobe level -25dB (increased by 3dB).
• Surface C: PV=λ/5 (0.5cm), RMS=λ/20 (0.25cm) – Gain 36.2 dBi (drop of 3.6dB), beamwidth 0.63° (broadened by 24%), sidelobe level -18dB (increased by 10dB).  in satellite signal reception tests: Surface A could receive weak signals at -100dBm (SNR 15dB), Surface B required -97dBm (SNR 12dB), while Surface C couldn’t lock at all (SNR <10dB). This is the “signal capturing ability” directly determined by surface accuracy.

Feed Phase Center

Using a 1.8-meter Ku-band antenna (f=11.3GHz, λ≈2.68cm), the feed’s nominal phase center was 1.5λ inside the horn mouth (≈4mm). But measurements revealed a 0.8mm deviation between the feed’s physical center (waveguide mouth geometric center) and the phase center – this 0.8mm reduced the received signal’s SNR from 22dB to 19dB, equivalent to a loss of 15% effective power.

The manufacturer’s manual only stated “feed length 50mm”, without specifying the phase center location. This is not an isolated case: we disassembled 12 commercial feeds; 7 had a phase center deviation from the physical center exceeding 0.5λ (≈1.5mm at 10GHz).

The phase center is not the “screw hole position”; it’s the “origin” of electromagnetic wave energy synchronization – an offset of 1mm, and the energy waves are “out of phase”.

What Exactly is the Phase Center?

The feed’s function is to convert high-frequency current into electromagnetic waves, or vice versa. But EM waves don’t “jump out from a single point” – for example, in a waveguide feed, the EM wave propagates inside the waveguide and gradually diffuses into a spherical wave at the horn mouth. The center of the sphere defined by the equiphase surface of this spherical wave is the phase center. An analogy: the phase center is like a “light bulb”, emitting light (EM waves) as concentric spherical waves; if the bulb is tilted, the light spot isn’t round, edge intensity weakens, and center intensity is insufficient.

Specifically for different feed types, the phase center location varies greatly:

• Rectangular Waveguide Horn Feed: Dominated by TE10 mode, phase center is inside the horn mouth, 1-2 wavelengths from the aperture (at λ=3cm, ≈3-6cm). We measured a BJ220 waveguide feed (aperture 58.4×29.2mm) with a near-field scanner; the phase center was 4.2cm forward from the horn aperture – offset by 1.8cm (≈0.6λ) from the physical center (waveguide mouth center).
• Microstrip Patch Feed: Common in small antennas, phase center is behind the patch, 0.5-1 wavelength (at λ=2.4GHz, ≈6-12mm). Measurement of a 5G microstrip feed showed the phase center 8mm behind the patch, while the patch itself was only 5mm×5mm.
• Helical Feed: Used for circular polarization, phase center is behind the helix center, 2-3 wavelengths (at λ=10cm, ≈20-30cm). We disassembled a satellite communication helical feed; physical length 40cm, but the phase center was at the 32cm mark – an 8cm difference (≈0.8λ).

Phase Center Offset by 1mm

The most direct impact of phase center deviation from the focus is reduced main lobe gain and increased sidelobes. We conducted a comparative test in an anechoic chamber: the same 1.2m Ku-band antenna (f=10.7GHz, λ≈2.8cm, focal length 0.8m), using three feeds with different phase center deviations:

• Ideal Case: Phase center strictly aligned with focus (Δx=0) – Main lobe gain 41.2 dBi, 3dB beamwidth 0.45°, sidelobe level -27dB.
• Deviation 0.5mm (≈0.05λ): Gain 40.7 dBi (drop 0.5dB), beamwidth 0.46° (broaden 2%), sidelobe -26dB (increase 1dB).
• Deviation 1.2mm (≈0.1λ): Gain 39.8 dBi (drop 1.4dB), beamwidth 0.48° (broaden 7%), sidelobe -24dB (increase 3dB).
• Deviation 3mm (≈0.3λ): Gain 38.5 dBi (drop 2.7dB), beamwidth 0.55° (broaden 22%), sidelobe -19dB (increase 8dB). More隐蔽ly, the noise temperature: phase center offset causes reflected waves to add out of phase, reducing the antenna’s received noise floor (G/T value) by 0.3dB/K – during rain fade (e.g., 10dB fade), this 0.3dB can cause the SNR to drop below the threshold, resulting in satellite signal “stuttering”.

How to Find the Phase Center?

Manufacturer manuals rarely specify the phase center location; you have to measure it yourself. We commonly use two methods:

Method 1: Near-Field Scanner – Finding the “Zero Gradient Point” via Electric Field Phase

Fix the feed on a rotary stage, place a near-field scanning accuracy ±0.1mm  in front. Scan the electric field distribution 5λ (≈14cm at 10GHz) in front of the feed, recording the phase value at each point.

The point where the “slope” (gradient) of the phase change with respect to position is zero is the phase center. We measured a waveguide feed; scanning revealed the phase center was 1.2cm forward of the physical center – a deviation of 0.3cm (≈0.1λ) from the design value.

Method 2: Laser “Simulation” of EM Waves – The Phase Center is Where the Spot is Stable

Use a Helium-Neon laser to shine into the feed waveguide mouth (laser wavelength 632nm, simulating the EM wave path). Adjust the feed position so the reflected spot falls on the cross mark at the paraboloid’s focal point. If the spot wobbles, the phase center isn’t aligned; when the spot remains stable, the feed’s position corresponds to the physical location of the phase center. We calibrated 5 antennas using this method, with errors controlled within ±0.2mm (≈0.02λ), 10 times more accurate than visual estimation.

Temperature and Vibration Affect It

The phase center is not fixed – temperature changes cause feed material expansion, vibration causes structural deformation, both leading to phase center shift. We conducted environmental tests:

• Temperature Effect: An aluminum alloy feed (thermal expansion coefficient 23×10⁻⁶/°C), heated from 25°C to 65°C (ΔT=40°C), length expansion ΔL=23e-6×40×500mm (feed length)≈0.46mm. If the phase center is at the feed’s midpoint (250mm), this 0.46mm expansion shifts the phase center by 0.23mm (≈0.025λ at 10GHz) – corresponding to a gain reduction of 0.1dB.
• Vibration Effect: Simulating a vehicle environment (5-20Hz, 0.5g vibration), a plastic feed’s phase center shifted back and forth by ±0.3mm during vibration – the received signal’s bit error rate surged from 1e-6 to 1e-4, equivalent to “occasional disconnections”. Therefore, high-precision systems (like radio telescopes) add temperature compensation structures to the feed (e.g., Invar Bracket, thermal expansion coefficient 1.2×10⁻⁶/°C), or periodically calibrate the phase center – otherwise, when temperature changes or the antenna shakes, the phase center shifts, and gain “secretly drops”.

Steps

A 1mm error can cause the signal to collapse. For example, in a Ku-band 11.7GHz system, wavelength is only 0.0256 meters; if the feed deviates from the focus by more than 0.1 wavelength (2.56mm), the VSWR can jump from 1.1 to 1.5, and the gain directly drops by 0.5dB. For Ka-band (30GHz, λ=0.01m), the error tolerance shrinks to 0.02mm; a slight shake can cause the satellite SNR to jump from 1e-6 to 1e-4, with bit error rates exceeding standards and signal link breaks being common occurrences.

Theoretical Focus

Precautions

A 1 millimeter focal length error can cause a signal gain drop of over 1 dB in the 30 GHz Ka-band, equivalent to losing over 20% of the received power.

Mechanical Alignment

A 2 millimeter focal error on a 2-meter aperture antenna is enough to reduce the gain by over 0.8 dB for a 14 GHz Ku-band signal, equivalent to wasting over 15% of the transmit power. it can cause beam pointing deviation, potentially up to 0.1 degrees, which on a long-distance satellite link is enough to make you completely lose the signal.

1. Get the Focal Length Right

The core of focal alignment is placing the feed’s “phase center,” not its physical aperture, at the paraboloid’s focal point. This point is a theoretical equivalent point, generally located inside the waveguide opening, about λ/π (wavelength divided by π) from the aperture. For example, a 12 GHz signal has a wavelength of 25 mm, so the phase center is roughly 8 mm inside the aperture.

Many feed manuals directly specify this location, with an error possibly within ±0.5 mm. Before installation, you must check the manual and mark this point on the feed housing with a vernier caliper; this is the benchmark for all your measurements. Don’t use the waveguide mouth as the benchmark directly, as that might introduce a systematic error of 3 to 5 mm from the start.

2. The String Method is the Rudimentary but Most Reliable Verification Tool

Stretch cross strings on the reflector; the intersection point is the paraboloid vertex. Use a thin cotton or nylon string, fix one end at the vertex, and pull it taut; its length should strictly equal the focal length F from the manual. The point where the string end points should be your marked feed phase center.

You need to repeat this operation from at least four symmetrical points on the reflector (e.g., 0°, 90°, 180°, 270° directions). The string ends pulled from different directions should converge in space within a spherical region with a diameter not exceeding 2 mm. If you find that strings pulled from different directions point to locations deviating by more than 1 mm, it indicates that your reflector itself might have a surface accuracy problem of 1-2 mm, or the support rod has bent imperceptibly to the eye.

3. Axis Alignment is Crooked

Ensure the feed’s electrical axis (i.e., its maximum radiation direction) is precisely aligned with the center of the paraboloid and perpendicular to the reflector. One check method: stand behind the reflector and look through the gap in the feed support; the feed waveguide mouth should be exactly in the center of the reflector, and the exposed border width around it must be completely uniform.

The precise method is to use a calibrated theodolite to first measure the reference plane of the parabolic aperture, then aim the telescope crosshair at specific mechanical feature points of the feed waveguide mouth (e.g., screw holes), measuring its concentricity deviation, which should be controlled within 0.3 degrees. An axis deviation of 1 degree at 30 GHz can directly cause a 3 dB gain loss.

4. Insufficient Stiffness of the Support Structure

The feed bracket’s first-order resonance frequency must be far away from possible environmental vibration frequencies. An unqualified single-arm bracket made of 15 mm diameter aluminum tube might have a resonance frequency of only 15 Hz; a breeze of 3 meters per second (about Force 3 wind) can cause it to vibrate with an amplitude of 0.5 mm at low frequency, leading to continuous signal fluctuation.

You should choose a triangular or quadrangular support structure made of carbon steel or stainless steel pipe with a diameter not less than 25 mm and a wall thickness not less than 2 mm, designed with a resonance frequency above 50 Hz.

5. Don’t Forget Temperature Makes Metal Expand and Contract

In areas with a daily temperature range of 30°C, a 500 mm long aluminum support rod will change length by ΔL = α * L * ΔT = 23.1e-6 * 500 * 30 = 0.35 mm. This means the antenna you adjusted at dusk at 20°C will have its focal point drifted by 0.35 mm by noon at 50°C.

For Ka-band systems (e.g., 30 GHz), this change is not negligible. The solution is either to use materials with a lower thermal expansion coefficient like Invar alloy or carbon fiber for key components, or to design the system to allow motor-driven fine adjustment of the focal point by ±1 mm for compensation, calibrated at different ambient temperatures.

Electrical Connections

A waveguide flange with a tiny 0.1 mm air gap can easily introduce 0.5 dB of insertion loss at 18 GHz, equivalent to losing over 10% of the power. worsening the system’s VSWR from an ideal below 1.1 to over 1.5, not only drastically reducing efficiency but potentially burning out expensive amplifiers. 90% of system performance issues ultimately trace back to these seemingly insignificant connection points.

1. Sequence is More Important Than Force

Metal-to-metal contact of waveguide flanges must be absolutely uniform; any微小 deformation will leak power and introduce phase error. During installation, always use a torque wrench and strictly follow a diagonal sequence in multiple rounds. For example, for a WR-75 flange, use a torque value of 1.2 N·m. First, hand-tighten all screws until snug. Then, using the torque wrench, first round tighten all screws to 0.6 N·m (50% of target value), second round tighten all to 1.2 N·m. Prohibit tightening in a clockwise or counterclockwise order in one go, as that will warp the flange like a pot lid, potentially causing a flatness error exceeding 0.05 mm. After tightening, check the gap around the flange interface with a feeler gauge; it should be less than 0.03 mm.

2. Poor Waterproof Sealing Makes the Signal Disappear on Its Own in Six Months

Moisture is the killer of microwaves. An N-type connector in an environment with 85% humidity, if simply wrapped with a few layers of electrical tape, will have its internal pin oxidize and turn black within 6 months, causing the contact resistance to surge from <5 mΩ to >500 mΩ, increasing insertion loss by over 0.3 dB. Reliable sealing is a three-layer structure: First, wrap 2 layers of high-quality electrical tape over the tightened connector threads as a base. Second layer, apply waterproof mastic with a thickness not less than 3 mm, knead it firmly to completely envelop the connector metal part, extending 5 cm onto the cable at both ends, ensuring no air bubbles. Third layer, slide on a dual-wall heat shrink tube with hot melt adhesive, use a 1200W heat gun to heat evenly from the middle towards the ends until about 1 cm of adhesive uniformly oozes out from both ends. The completed seal should withstand immersion in 1 meter deep water for 30 minutes without leakage.

3. Don’t Let the Waveguide-to-Coax Transition Become the Bottleneck

The waveguide-to-coax transition is a critical component in the system. Its VSWR bandwidth must cover your entire operating band. For example, a transition rated for 10.7-12.75 GHz Ku-band should have a VSWR less than 1.15 across the band. During installation, always connect the transition to the waveguide port first, then connect the coaxial cable to the transition, avoiding torsional stress on the transition’s coaxial interface (usually N-type or SMA). The transition’s own weight needs additional support;  When testing with a VNA, the insertion loss introduced by the transition itself should be below 0.2 dB; if you find a sudden increase in loss of 0.5 dB at a certain frequency, the internal probe is likely damaged.

4. Insufficient Cable Bend Radius Leaks Signal Power on the Way

The minimum bend radius of a coaxial cable is 5 to 10 times its outer diameter. For example, a feeder cable with an outer diameter of 10 mm should have a minimum static bend radius not less than 10 cm. A sharp bend will irreversibly alter the internal dielectric structure of the cable. At 18 GHz, a bend tighter than 5 cm can cause the local VSWR at that point to rise to 2.0, creating signal reflection. When routing cables, any 90-degree turn should use a curved transition with an arc radius of at least 15 cm, and be secured with cable ties to prevent the cable’s own weight or wind sway from causing gradual bending.

5. Incorrect Grounding Makes Noise Louder Than the Signal

Grounding for an antenna system provides a low-impedance path for electrostatic discharge and lightning-induced currents, not for a “working neutral.” The outer conductor of the coaxial cable must be grounded at both ends: at the antenna side and at the equipment entry point into the building. Ground wires should be “short, straight, thick,” less than 1 meter long, with a cross-section not less than 6 mm². Using copper or stainless steel braid is better than round wire due to its lower high-frequency impedance. Measure the grounding resistance with a ground resistance tester; the ideal value is less than 4 ohms.

Debugging and Optimization

Debugging without instrument assistance is like shooting blindfolded. At 18 GHz Ku-band, the beamwidth might be only about 1 degree; an antenna pointing error of 0.2 degrees is enough to reduce signal strength by 3 dB, losing half the power. A polarization angle error of 10 degrees loses another 1 dB. A feed axial position error of 2 mm loses another 0.5 dB. Stacked together, the total system gain might be 4-5 dB lower than the design value, a full 70% loss of effective power.

1. First Find the Satellite Signal to Get the Rough Direction Right

If you’re doing satellite communication, don’t touch精密 instruments first. Use a satellite spectrum analyzer or a modem with signal quality display, and roughly point the antenna towards the satellite’s orbital position. Slowly rotate the antenna, adjusting azimuth by 1 degree and elevation by 0.2 degrees each time, observing the signal strength bar. Once you see a signal peak on the spectrum analyzer, first adjust the azimuth and elevation to where the peak is maximum; the signal strength might only be 60% of the theoretical value at this point.

2. Using a Vector Network Analyzer to Tune VSWR is the Professional Approach

After roughly adjusting with the satellite signal, bring out the Vector Network Analyzer (VNA). Connect the VNA’s two ports to the antenna system using calibrated cables, one port transmitting, one port receiving the reflection. Set the VNA’s sweep range 20% wider than your used band; for example, for C-band uplink (5.8-6.5 GHz), sweep from 5.5 to 6.8 GHz. The screen will display an S11 parameter curve, representing how much signal is reflected back. The optimization goal is to press this curve as low as possible across the entire operating band, ideally <-15 dB (corresponding to VSWR <1.5).

3. Adjust Feed Depth to Minimize Reflection

Now start微调. One person watches the VNA screen, another extremely slowly moves the feed back and forth, each movement amplitude controlled to 0.5 mm. You’ll see the valley (resonant point) of the S11 curve change with the depth. The goal is to find the position where the valley is deepest and the curve is flattest across the entire band. For example, moving 3 mm might optimize S11 from -12 dB to -20 dB at 6 GHz.

4. Rotate the Feed to Match the Polarization Angle Precisely

Polarization mismatch is a hidden killer. Keeping the feed depth constant, slowly rotate the feed body in steps of 2-3 degrees each time. For linearly polarized signals, observe S21 (transmission parameter, if the system is transceiver-integrated) on the VNA or go back to the satellite spectrum analyzer to watch the downlink signal strength. You’ll find the signal power changes significantly with the rotation angle. Find the angle that gives the strongest signal; it might deviate by 15 degrees from the initial angle. For circular polarization, this rotation adjusts the dielectric片 in the polarizer, ensuring it is at a 45-degree angle to the waveguide mouth, with an error controlled within ±2 degrees.

5. Sweep a 3D Pattern to Verify Final Performance

After everything is tuned, if conditions allow, you should measure the antenna’s 3D radiation pattern. Use an antenna measurement system to scan the antenna in azimuth and elevation with 0.1-degree steps. The software will generate a color cloud map, allowing you to clearly see the main beam width and sidelobe levels. A qualified 1.8m C-band antenna should have a -3 dB beamwidth of about 1.8 degrees, and the first sidelobe level should be below -14 dB. If the pattern appears skewed or sidelobes are too high, it indicates the reflector可能有 over 1 mm of deformation or installation stress.

6. Re-test at Different Temperatures to Ensure Year-Round Stability

Metal expands and contracts with heat. Re-test the antenna’s S11 parameter and signal strength at noon (30°C) and early morning (10°C). Because the support rod thermally expands/contracts, the focal point will drift, potentially causing S11 to worsen from -20 dB at low temperature to -16 dB at high temperature. If the change exceeds 2 dB, it indicates insufficient thermal stability of the mechanical structure.