Why is the crossover frequency usually at the -3 dB point?

If you plot the response of a diplexer on a graph, the low-pass filter's curve rolls down while the high-pass filter's curve rolls up. They must cross somewhere. If they cross at the 0 dB line (which is physically impossible), it would mean 100% of the power is going to both ports simultaneously, creating energy out of nowhere. By conservation of energy, if the input power splits equally between the two ports at that exact transition frequency, each port receives exactly half the power. A 50% power reduction is mathematically equal to -3 dB. Therefore, a perfectly matched diplexer inherently crosses at -3 dB.

What happens to a signal located exactly on the crossover frequency?

It gets destroyed by the split. Half of the signal's power is routed out the low-pass port, and the other half is routed out the high-pass port. Because neither port receives the full signal, and both ports suffer a massive 3 dB insertion loss, communication at that exact frequency is generally impossible. System designers must ensure that the crossover frequency is placed in an unused 'guard band' between the actual operating channels.

Can you design a crossover with a sharper drop-off?

Yes. You can increase the 'order' of the filters (adding more inductors and capacitors). This makes the filter walls steeper, narrowing the transition band around the crossover frequency. However, higher-order filters have more components, which increases the baseline insertion loss (heat) and causes massive phase distortion (group delay) near the crossover point, ruining high-speed digital signals. It is a trade-off between isolation and signal integrity.

Passive Components

Crossover Frequency

A network engineer installs a diplexer to split a single coaxial cable into two paths: one for low-frequency cable TV (10–800 MHz) and one for high-frequency satellite data (950–2150 MHz). To guarantee the satellite receiver doesn't blow out the TV tuner, the diplexer must cleanly separate the bands. The exact mathematical point where the device stops acting like a low-pass filter and starts acting like a high-pass filter is the Crossover Frequency. In this case, the designer sets the crossover frequency to 875 MHz. If you feed an 875 MHz test tone into the diplexer, you will find that the power splits perfectly in half—50% goes to the TV port, and 50% goes to the satellite port. Because half the power is lost to the other port, the signal suffers a brutal 3 dB insertion loss. To prevent dropped connections, engineers always place the crossover frequency in a "dead zone" (guard band) where no critical data is transmitted.

Category: Passive Components

Location: The intersection of LP and HP response curves

Typical Amplitude: -3 dB (50% power split)

Filter Transition Zones

Zone	Frequency Range	Filter Action	Signal Integrity
Passband	Far away from Crossover	Minimal Insertion Loss (< 0.5 dB)	Excellent (Data flows cleanly)
Transition Band	Approaching Crossover	Increasing Loss (1 dB to 2 dB)	Degraded (Phase distortion begins)
Crossover Frequency	The Exact Intersection Point	Massive Loss (-3 dB)	Destroyed (Power splits equally)
Stopband	Past the Crossover	High Isolation (> 40 dB)	Blocked entirely

Conservation of Power at Crossover:
For a perfectly lossless (ideal) diplexer, the total power leaving the two output ports must equal the input power.
|S₂₁|² + |S₃₁|² = 1
At the exact crossover frequency (f_c), the power delivered to Port 2 equals the power delivered to Port 3.
|S₂₁|² = 0.5 (Which equals -3.01 dB)
|S₃₁|² = 0.5 (Which equals -3.01 dB)

The Diplexer Gap (Guard Band):
Because the transition from passband to stopband isn't instantaneous, you cannot place a 900 MHz channel and a 901 MHz channel on opposite sides of a crossover. You must leave a sufficient gap (e.g., 850 MHz to 950 MHz) to allow the filter curves to physically roll off.

Common Questions

Frequently Asked Questions

Can a crossover frequency be at -6 dB?

Yes, if the diplexer is designed differently (e.g., a "non-contiguous" diplexer). If the system requires extreme isolation between the two bands, the designer might force a massive gap between them. In this case, the two filter curves don't intersect at the standard 3 dB point; they intersect much deeper in the stopband, causing the crossover point to exhibit 6 dB, 10 dB, or even 20 dB of loss. This is perfectly fine, as long as no data is transmitted at that frequency.

What is a Contiguous Diplexer?

A diplexer where the passband of the low-pass filter immediately touches the passband of the high-pass filter, crossing exactly at the -3 dB point. This is the most mathematically elegant design (often using a singly-terminated Butterworth or Chebyshev prototype), providing a perfectly flat 50-ohm input impedance across the entire frequency spectrum, including right at the crossover frequency.

How does group delay behave near the crossover?

Terribly. Group delay measures how much a filter slows down different frequencies. Deep in the passband, the delay is flat, so a digital pulse passes through intact. As the signal approaches the crossover frequency, the inductive and capacitive elements of the filter cause massive phase shifts, causing the group delay to spike violently. If a digital signal is placed too close to the crossover, this delay spike will smear the 1s and 0s together, causing fatal Inter-Symbol Interference (ISI).

Related Terms

← Cross-Polarization Discrimination Crosstalk →

Passive Design

Diplexer Transition Synthesizer

Input your lower and upper frequency bands and desired filter order. Visualize the intersection of the S21 and S31 S-parameters, locate the exact -3 dB crossover frequency, and calculate the resulting group delay spike in the transition band.

Synthesize Diplexer Curves