Attenuator

Attenuator is a device that dissipate (waste) energy into environment and as a result only reduced amount of energy gets out of the output port. Conceptually it is almost like a resistor in electronic circuit.

The diagram below shows the idea. Power enters at port p1 on the left. Part of it turns into heat, drawn as the arrows leaving the body, and the reduced power leaves at port p2.

Attenuator dissipating part of the input power at p1 and passing reduced power to p2

Figure 1. An attenuator between two ports. The power lost between p1 and p2 leaves as heat, so the device must be rated for the power it absorbs.

This page first puts numbers on the power that an attenuator removes. Then it explains why a system needs one, which kinds are on the market, and how a fixed attenuator is built from three resistors.

How much power does an attenuator remove ?

Before choosing an attenuator, you need a way to state how much it takes away. RF engineers state it in dB, because a ratio in dB turns multiplication into addition. The attenuation A is 10 log₁₀(P1/P2), where P1 is the power into p1 and P2 is the power out of p2.

In dB, the rule for levels is simple. If the input level is in dBm, the output level is P2 = P1 - A, also in dBm. For example, 1 W is 30 dBm. A 20 dB attenuator turns it into 10 dBm, which is 10 mW. The other 990 mW becomes heat inside the attenuator.

The table below lists the power ratio for common values. In a matched 50 ohm system, the voltage ratio in dB is the same number. The reason is that power goes with the square of voltage, so the voltage ratio in dB is 20 log₁₀(V1/V2).

Attenuation	Output power over input power	Output for a 30 dBm input
3 dB	0.50	27 dBm, 501 mW
6 dB	0.25	24 dBm, 251 mW
10 dB	0.1	20 dBm, 100 mW
20 dB	0.01	10 dBm, 10 mW
30 dB	0.001	0 dBm, 1 mW

Attenuators in series add in dB : a 10 dB and a 20 dB attenuator in series give 30 dB, which is a power ratio of 1/1000.
3 dB is half the power, not half the voltage : half the voltage is 6 dB.
An attenuator also sets a noise figure : a passive attenuator at room temperature has a noise figure equal to its attenuation. So a 10 dB pad in front of a receiver adds 10 dB to the noise figure of the chain.

Why we need an Attenuator ?

Probably an Attenuator would be one of the most frequently used RF device. The purpose of the device is to reduce the power level. Why do we want to reduce (waste) power level ? It is mainly to protect the input path of a system or reduce the input power level to fit into the dynamic range of the input port of a system (or a device).

Every device or every system has its own range of input power that the system (or device) can tolerate. Therefore, if the signal power you want to connect to a system (a device) is too high, you need to reduce the signal power so that it would not burn out the input channel of the system.

On the contrary, if the signal level is too low to be detected by the system (or device) you want to connect, you need to use an amplifier to increase the power level.

Let's check both jobs with a numerical example. Suppose a transmitter puts out +43 dBm, which is about 20 W. Suppose also that the input port of the analyzer is rated for +30 dBm. A 30 dB attenuator brings the signal down to +13 dBm, which is safely inside the rating. The attenuator itself absorbs almost all of the 20 W, so it must be a power attenuator rated for at least 20 W.

An attenuator has one more benefit that is easy to miss. A reflection from a poorly matched input passes through the attenuator twice. It is attenuated once on the way in and once on the way back. So a 10 dB pad improves the return loss seen at its input by 20 dB. A port with a 6 dB return loss looks like a port with a 26 dB return loss through the pad.

Protect the input first : take the highest level the source can produce and subtract the attenuation. Then compare the result with the damage level of the input.
Rate the attenuator for the power it absorbs : with a large attenuation, the attenuator absorbs almost all of the input power.
Use a pad to clean up a mismatch : the return loss improves by twice the attenuation, at the cost of signal level and noise figure.

Types of Attenuators

Attenuators differ in three ways that matter when you pick one. They differ in how much power they can absorb, in the frequency range and connector they use, and in whether the value is fixed or variable.

There are so many different types of Attenuator you may get from the market. A small set of examples are shown below.

(A) would be one of the most common attenuators that are used in various RF application. (B) would be similar to (A) but these are usually used for high power application with cooling fins shown in the picture. (C),(D),(E) are cavity /waveguid type which would be used for very high frequency / high power application. (A)~(E) are all the attenuator with a fixed value (i.e, you cannot change the attenuation level arbitrarily), but with (F), (G) you can change the level of the attenuation arbitrarily. These are called variable attenuator. For (F), you can change the attenuation level by manually rotating the knobs. For (G), you can change the attenuation level programatically.

Examples of attenuators A to G from coaxial and high power types to waveguide, manual step and programmable attenuators

Figure 2. Examples of attenuators. The size grows with the power to be absorbed, and the connector or flange changes with the frequency range.

A - coaxial fixed attenuators : these are small in-line units with connectors on both ends. The picture shows larger silver N and BNC types and small gold SMA types.
B - high power attenuators : the black fins are a heatsink. They carry away the power that the attenuator absorbs.
C, D and E - waveguide attenuators : these connect with waveguide flanges instead of coaxial connectors. D carries micrometer heads, and E is a large assembly for high power.
F - manual step attenuator : the knobs select the attenuation in fixed steps, and the dB values are printed around them.
G - programmable attenuator : a rack unit with many SMA ports lets a controller set the attenuation of each path. Test systems use this kind to emulate path loss and fading between several devices.

How is a fixed attenuator built inside ?

A fixed attenuator in a 50 ohm system has two jobs at once. It must reduce the level by A dB, and it must still look like 50 ohm from both ports. A single series resistor can do the first job but not the second. Three resistors can do both, in a T shape or in a Pi shape.

The diagram below shows both networks. Each is symmetrical, so it gives the same attenuation and the same match in both directions.

Figure 3. T pad and Pi pad attenuators. Two resistor values set both the attenuation and the 50 ohm match at p1 and p2.

Let K = 10^A/20 and Z₀ = 50 ohm. The T pad uses R1 = Z₀(K - 1)/(K + 1) in series and R2 = 2Z₀K/(K² - 1) in shunt. The Pi pad uses Rsh = Z₀(K + 1)/(K - 1) in shunt and Rse = Z₀(K² - 1)/(2K) in series. The table below lists the values for common attenuations. Each row was checked by computing the input impedance with a 50 ohm load, which comes out at 50.0 ohm.

Attenuation	T pad R1	T pad R2	Pi pad Rsh	Pi pad Rse
3 dB	8.5 ohm	141.9 ohm	292.4 ohm	17.6 ohm
6 dB	16.6 ohm	66.9 ohm	150.5 ohm	37.4 ohm
10 dB	26.0 ohm	35.1 ohm	96.2 ohm	71.2 ohm
20 dB	40.9 ohm	10.1 ohm	61.1 ohm	247.5 ohm
30 dB	46.9 ohm	3.2 ohm	53.3 ohm	789.8 ohm

A large attenuation pushes the values to extremes : at 30 dB, the T pad needs a 3.2 ohm shunt resistor. The Pi pad needs a 789.8 ohm series resistor. Parasitic inductance and capacitance then matter more, so high values are often built as two pads in series.
The pad is also the model for a resistor-based attenuator chip : digital step attenuators switch small T or Pi sections in and out to make each step.
A pad works at any frequency where the resistors stay resistive : at microwave frequencies, the layout and the resistor size decide how flat the attenuation stays.