QuickSmith
Articles
Notes on impedance matching and on getting an RF design off the chart and onto a board. Where an article describes a circuit it ends with a link that opens that circuit in the program, so you can pull it apart rather than take our word for it. Every number in them goes through the solver before it goes into the prose.
Start here
How to match an antenna to 50 ohms on a Smith chart
From a VNA reading to two components. Why a series part and a shunt part move you along different circles, how to find every two element match at once, and how to choose between four networks that are identical at the design frequency but differ by a factor of three in bandwidth.
What the Smith chart is actually showing you
The impedance plane folded into a circle so that infinity fits on the page. Where the constant resistance and constant reactance circles come from, and why that geometry is the reason each kind of component moves you the way it does.
Insertion loss, return loss and mismatch loss are three different things
A perfectly lossless quarter wave line can show 0.695 dB of insertion loss without turning a single milliwatt into heat. What each of the three actually measures, and the rules of thumb worth carrying around.
Designing a network
L, Pi or T: choosing a matching network and what it costs in bandwidth
An L-network's Q is fixed by the impedance ratio and you get no say. Pi and T add a third component so you can choose it. What that buys, what it costs, and how to pick between Pi and T on practical grounds.
Q circles: why your match works at one frequency and nowhere else
Loaded Q is the highest ratio of reactance to resistance along your element path, and it is what sets the bandwidth. How to see it on the chart, and how to design to a bandwidth instead of discovering one afterwards.
Component Q: what a real inductor does to your match
Real parts do not just cost you power, they move your match. How much a given component Q costs, why a high Q network is punished three times harder for the same components, and what Q to assume for parts you can buy.
Single stub matching, worked end to end
Matching 10 - j15 ohms to 50 ohms at 1 GHz with a series line and an open stub. Why a line rotates rather than transforms, and how to convert published air lengths onto a real board.
Getting it onto a board
50 ohm trace width on FR-4, and where that number comes from
About 3.08 mm on 1.6 mm FR-4. Why the third decimal place is not real, what an etch tolerance actually costs you, and how to turn a matching network into a track width and a length.
Reading a nanoVNA measurement into a Smith chart
Import a Touchstone .s1p and use the measured curve as the load,
so you match a band rather than a point. A dipole taken from 3.88:1 down to
1.84:1 across 100 to 200 MHz with two components.
Amplifiers
Matching a 2N5642: the AN721 input and output networks
Motorola AN721 section 4.2 on a chart you can drag. Transforming a 1.94 + j1.1 ohm transistor input to 50 ohms at a loaded Q of 10, and working out what the collector needs to see before matching the output.
Stability circles and the gain versus noise trade-off
Three questions in order: will it oscillate, how much gain can I have, how quiet can it be at that gain. Rollett's K and |Delta| worked from real S-parameters, and why the gain and noise optima are never the same impedance.
About the tool
Share a working circuit instead of a screenshot
Copy Share Link puts the whole design in a web address, measured loads included. Nothing is uploaded and nothing expires, because the design is the string.
A 1993 Windows program, rebuilt for the browser
What it took to rebuild the solver, the chart and the tests without changing how the program is used, and the five real defects the regression suite found before anything was rewritten.
Elsewhere on the site
The worked examples go through seven designs from the textbooks. The FAQ covers how to drive the program itself.