Design a resistive attenuator pad (T or π) that reduces a signal by a fixed number of dB while keeping the system matched to its characteristic impedance Z₀. Change the attenuation, Z₀, input power and topology to see the three resistor values, output power, dissipated power and voltage ratio update in real time.
Parameters
Attenuation
dB
Signal-level reduction (power ratio = 10^(dB/10))
Characteristic impedance Z₀
System reference impedance — 50 Ω is the RF standard
Topology
T-pad (series-shunt-series) or π-pad (shunt-series-shunt)
Input power P_in
W
Effective power incident at the input port
Source impedance R_s
Ω
Source-side output impedance (reference value)
Results
—
Voltage ratio α (×)
—
R series (Ω)
—
R shunt (Ω)
—
Output power P_out (W)
—
Dissipated power P_diss (W)
—
Output voltage ratio (%)
—
Circuit diagram (T / π) — signal animation
A two-port pad built from three resistors. Signal enters from the left and is attenuated through the selected topology (T or π) with the computed resistor values. The fading colour shows the falling signal amplitude.
α is the voltage ratio (α = 10^(dB/20)). A T-pad has two equal series arms R1 = R3 and a single shunt R2. R1 and R2 depend only on α and Z₀, and the input impedance is exactly Z₀ when the output is terminated in Z₀.
A π-pad has two equal shunt arms R1 = R3 and a single series R2. T and π are equivalent in performance; choose between them on resistor value availability and heat sharing.
The power ratio is 10^(dB/10). A 30 dB pad turns 99.9% of the input power into heat in the three resistors.
What is a resistive attenuator pad?
🙋
"Attenuator pad" shows up everywhere in RF books — it's basically a lump of resistors that weakens a signal, right? Why three resistors? Couldn't I just put one series resistor in line?
🎓
Good question. A single series resistor would reduce the signal, but it would also change the input and output impedance dramatically. Drop a 100 Ω resistor into a 50 Ω line and the source now sees 150 Ω and the load sees 150 Ω — at RF that produces reflections and the attenuation itself varies with frequency. The whole point of a pad is to attenuate while preserving the Z₀ match, and you need a minimum of three resistors (T or π) to do that.
🙋
Got it — three resistors for the matching. So which one should I pick, T or π? Is there a mathematical difference?
🎓
This is the fun part: T and π are completely equivalent in performance. For the same Z₀ and the same dB you get the same attenuation and the same impedance match either way. What differs is the resistor value range. At 10 dB / 50 Ω the T-pad needs about 26 Ω, 35 Ω, 26 Ω while the π-pad needs about 96 Ω, 71 Ω, 96 Ω. In practice you pick "whichever lands closest to the E96 series", "whichever spreads the heat better", or "whichever fits the PCB layout". Switch the topology selector above and you can see the difference in the bar chart.
🙋
What attenuation values do people typically use? Anything goes?
🎓
In the lab you see the "round numbers" over and over: 3, 6, 10, 20, 30 dB. 3 dB halves the power, 6 dB quarters it, 10 dB drops it by ten, 20 dB by a hundred. 10 dB is the most common "single step". For an arbitrary value you cascade standard pads and the dB simply add — 10+10+3 = 23 dB, for example. Typical roles: 10 dB to protect a spectrum-analyser input, 20 or 30 dB to bring a strong source down, 3 or 6 dB to level-match a mixer input.
🙋
The energy that gets "attenuated" turns into heat in the resistors, right? At high dB it must get really hot…
🎓
Yes, and that is the critical point. A 30 dB pad converts 99.9% of the input power into heat in the three resistors. Feed 10 W into a 30 dB pad and you need to dump nearly 10 W of heat — no ordinary chip resistor will survive, you need a flange-mount unit rated for tens of watts with a heat sink. Look at the "Dissipated power vs attenuation" chart below: as dB grows, the dissipation does not just keep climbing — it saturates toward P_in. Conversely a 1 or 3 dB pad dissipates very little and a tiny SMD resistor is fine. Pick the wrong power rating and the resistor burns out on the first pulse.
🙋
Last question — how high in frequency does this work? RF means hundreds of MHz, even GHz…
🎓
Pure resistors are theoretically flat from DC to infinity, but real resistors have parasitic inductance and capacitance that bite at high frequency. Leaded resistors start losing their flat response around a few hundred MHz. An 0603 SMD chip resistor is good to several GHz, and dedicated thin-film chip resistors or stripline pads reach tens of GHz. Commercial coaxial pads come as DC–18 GHz or DC–40 GHz parts. Whether a given pad reaches your frequency is always answered by the VSWR-vs-frequency graph on its datasheet.
Frequently Asked Questions
A resistive attenuator pad is a two-port network built from three resistors arranged in either a T or a π configuration. It presents an input impedance equal to the characteristic impedance Z₀ (typically 50 Ω) and reduces the signal by a known, fixed amount in dB. A capacitor or inductor cannot do the job because its impedance is frequency-dependent — the attenuation would vary with frequency and the reactive part would create reflections (impedance mismatch). With pure resistors the attenuation stays flat from DC to tens of GHz and the impedance match is essentially perfect, which is why resistive pads are the standard tool in RF measurement and signal distribution.
T and π configurations are mathematically equivalent: for any attenuation and Z₀ either topology delivers the same dB and the same impedance match. They differ in the resistor value ranges and in how heat is shared. The π form has lower series resistance and higher shunt resistance; the T form has higher series resistance and lower shunt resistance. In practice you choose based on (1) which values are closer to the E96 standard series, (2) which layout removes heat better, and (3) which is less sensitive to parasitic L/C. For example a 10 dB / 50 Ω pad needs 26 Ω + 35 Ω + 26 Ω as a T-pad or 96 Ω + 71 Ω + 96 Ω as a π-pad.
The values used over and over in RF labs are 3 dB, 6 dB, 10 dB, 20 dB, 30 dB and 40 dB. 3 dB halves the power (voltage ≈ 0.707×), 6 dB quarters it (voltage ÷2), 10 dB cuts power by ten (voltage ≈ 0.316×) and 20 dB cuts by a hundred (voltage ÷10). 10 dB is the most common "one-step" unit, used for spectrum-analyser input protection, for trimming high-level sources, and for level matching at mixer inputs. Because dB values add when pads are cascaded, you can build arbitrary attenuation from standard parts — 3+10+10 = 23 dB, 20+6 = 26 dB and so on.
All the attenuated power is converted to heat in the three resistors. A 10 dB pad fed with 1 W must dissipate 0.9 W; a 30 dB pad dissipates 99.9% of the input. Feed 10 W into a 30 dB pad and you need to get rid of about 10 W of heat, which usually means a flange-mount resistor on a heat sink rated for tens of watts. T and π topologies share the heat differently — typically the middle resistor carries 40–60% of the total — so the safe rule is to rate that one resistor at over half the input power. For RF pads always check three things on the datasheet: power rating, case-temperature derating, and the upper frequency limit (set by parasitic inductance).
Real-World Applications
RF instrument input protection: Spectrum analysers and network analysers have a front-end mixer or LNA that an over-driven source would destroy in a single shot. A 10 dB or 20 dB fixed pad at the input lets the test signal pass while keeping the absolute level safely within the instrument's damage limit. "When you suspect the signal might exceed full-scale, drop a pad in first" is one of the iron rules of RF measurement. Use the tool here to look up the resistor values and the dissipated power for the dB you need.
Source-level matching: Signal generators have built-in step attenuators, but their step size (0.1 dB or 1 dB minimum) may be coarser than you need, or the source level may simply be too high for the device under test. A series fixed pad trims the level and at the same time makes the source impedance look almost ideally 50 Ω, which improves the stability of mixers and other non-linear loads fed by the source.
Building arbitrary attenuation by cascading: An RF engineer's toolkit always contains 3, 6, 10, 20 dB pads ready to be combined in series. The dB simply add: 10+10+3 = 23 dB, 20+6 = 26 dB, 30+10+3 = 43 dB. Without a programmable step attenuator you can still cover a wide range of levels using just a few standard pads.
Isolation in splitters and combiners: When you combine two signal sources into one device, connecting them directly causes inter-source interference and reverse coupling. A 6 dB or 10 dB pad after each source attenuates any leakage in dB twice, so the mutual influence drops sharply. "An attenuator is the cheapest and most effective isolator you have" is a saying in the RF community for a reason.
Common Misconceptions and Pitfalls
The most common mistake is "if the pad's total power rating exceeds the input power, it is safe". In reality the three resistors do not share heat equally — typically one of them (often the middle resistor) carries 40 to 60 percent of the total dissipation. For a 30 dB / 1 W T-pad almost 0.5 W lands on the single centre shunt resistor. A naive design using five 0.25 W parts in parallel may look fine on the spec sheet, but the one centre part still burns out first. Use the resistor values shown in the tool together with the current through each resistor to estimate the dissipation in each part individually.
Next, "a 50 Ω pad in a 50 Ω system is always correctly matched". The pad presents an input impedance equal to Z₀ only when the output port is terminated in Z₀. If you leave the output open, short it, or connect a load far from Z₀, the input impedance and the attenuation both move. When you measure a stand-alone attenuator, always put a proper Z₀ load on the output port.
Finally, "dB adds linearly so I can cascade pads forever". The dB do add — but only when each pad is well matched to Z₀. Cheap or parasitic-heavy pads in long cascades show residual reflections at each interface that subtract from the predicted attenuation. In the GHz region the SMA connectors and short coax sections you stack between pads also contribute their own mismatch. When you cascade many pads, check VSWR and return loss of each stage, not just the dB.
How to Use
Enter attenuation in dB (e.g., 20 dB for 10:1 voltage reduction).
Specify input power in watts and source impedance in ohms (typically 50 Ω for RF systems).
Select T-pad or π-pad topology; simulator calculates series and shunt resistor values to maintain impedance matching.
Read output power, dissipated power, and voltage ratio from results.
Worked Example
Design a 6 dB attenuator pad for a 50 Ω RF transmission line with 1 W input power. Using π-pad topology: attenuation factor α = 10^(6/20) = 1.995. Series resistor R_s = 16.7 Ω, shunt resistor R_shunt = 133 Ω. Output power P_out = 0.251 W, dissipated power P_diss = 0.749 W. Voltage ratio = 50.1%, confirming impedance matching on both ports.
Practical Notes
T-pad excels for high attenuation (>15 dB); π-pad preferred for low attenuation (<10 dB) due to lower shunt impedance loading.
Match resistor tolerances to ±1% for systems above 1 GHz to maintain return loss better than -20 dB.
Dissipated power becomes thermal concern above 10 dB attenuation; use thin-film resistor arrays rated for instantaneous power spikes in pulsed radar applications.
Verify DC blocking capacitors on signal paths when driving high-impedance loads to prevent DC bias shift.