Wire antennas, modelled for your space and your bands.
Quiet right now: a new job starts straight away.
0 job(s) running, 0 waiting, and 0 background job(s). When nobody's job is waiting, the servers work through a library of antennas in the background, so answers for common situations are ready sooner. Your job always goes first: the background work pauses for it.
Updates every 15 seconds.
Describe the space you have (two trees, the house and a fence post, one pole, or a few supports for a loop). The Lab tries every antenna that suits it on 80 m to 6 m, with and without a tuner, ranks what works and gives you a build sheet for each. It takes a few hours; it can email you when it's done.
Pick from 41 antenna types (dipoles, OCFs, end-feds, verticals, loops, wire beams) or paste your own NEC model. Tick what may change and choose your bands.
Tell it your lowest band, the gap between your supports and your tallest support, and it sizes the antenna to fit.
See a real result: a 150 ft delta loop worked through from the model to the balun to wind. Anyone can open it.
The antenna, your space and your bands.
Pick from 41 antenna types or paste your own NEC model. Not sure of the sizes? Start from what you have sizes it from your lowest band, the gap between your supports and your tallest support.
NEC-2 tries hundreds of versions over real ground.
Every band is checked edge to edge, and the balun ratio is worked out for you. It also shows what moves each band, and which other band sets the same wire can do.
Lengths to cut and the balun to wind.
Measurements in metres and %, how carefully to cut each one, and a check that adds random build errors to see if it still works. The balun turn by turn, with a test resistor. A build sheet to print.
Where your signal goes, and whether your tuner can match it.
Each band against a plain dipole at the same height. Your tuner checked with its real coil and capacitors. A table for your analyser readings, and a link to share with a mate.
A real job: a 150 ft (45.7 m) delta loop, bottom wire 3 m up, over poor ground, with a remote tuner at the feed.
Drawn to scale from this design's NEC model (each view has its own scale bar). Lengths are along each straight wire; the red dot is the feed point. The legs either side of the feed are measured from the middle of the feed point.
| Bands | 7 of 10: 40, 20, 17, 15, 12, 10, 6 m. Not: 160, 80, 30 m. Your ATU-100 7x7 (N7DDC design) can match 8: 40, 30, 20, 17, 15, 12, 10, 6 m. Not 160, 80 m. (30 m too: this check uses the tuner's real coil and capacitors, not the 10:1 rule of thumb.) |
|---|---|
Radiated?How much of your power leaves as radio waves. The rest warms the ground, the wire, coils or resistors. | 40 m 40% 20 m 45% 17 m 66% 15 m 61% 12 m 67% 10 m 62% 6 m 63% More than half your power gets out on 17, 15, 12, 10, 6 m; a fair share (20–50%) on 40, 20 m. The rest heats the ground, wire or coils. |
Against a dipole?Signal at low angles (good for long distance) compared with a plain half-wave dipole at the same height. Minus means weaker. | 40 m −4.7 20 m −9.1 17 m −3.6 15 m −8.1 12 m −5.1 10 m −2.7 6 m −1.9 dB at low (DX) angles, 6–24°, against a plain half-wave dipole at 16.2 m over the same ground: below it on every band shown. |
| Building it | Most careful with side length: Some room (46 cm shorter loses 40m; 46 cm longer is fine). |
Feeding it?The balun is a small transformer at the feedpoint that matches the antenna to your coax. The Lab works out the ratio and how to wind it. | A 6.89:1 balun: 21 turns tapped at 8 on an FT-140-43, plus a choke. Anything from 6.48:1 to 56.05:1 works just as well. With your tuner at the antenna (remote), matching up to 10:1. |
SWR across the whole band, for every band you tick. The dashed line is the limit (here the tuner's 10:1): 20 m passes, 30 m does not by the model, though the tuner check says this tuner can still match it. Fails are part of the answer, not hidden.
It started with an antenna I am building, the "May cage": a caged inverted-V with separate wires for 80, 40, 30, 20 and 10 m on one feed. Getting five bands to play together meant changing one wire, re-running the model, then finding another band had moved. Over and over. I wanted something to do that grind for me: try hundreds of combinations, show me which one works, and tell me what moves what.
In the model it found a design with all five bands under the SWR target on a 2.25:1 balun. Now it is being built for real, with 3D-printed parts where needed, and will go up on one of the towers here to see how close the model gets. The Lab grew from there.
Peter, VK2ADG
On top of the Lab there is an optional helper, Claude (an AI made by Anthropic). It can explain a finished job in plain words, suggest the next run, fill in the form from a description, suggest antennas for your block and run small experiments for a site. For testers it is switched off by default, because every request costs the Lab money and because an AI can be wrong. Peter can switch it on for you: ask in the suggestions box.
Claude only reads what the Lab sends it. The numbers always come from NEC-2, never from the AI, and everything Claude writes is marked as an AI reading. Testers get 3 free Claude requests to try it when it is switched on, and can buy credit for more (what Claude costs the Lab plus a small margin that pays for new testers' free tries).
Without it you can still have a chat with Claude yourself: Download for analysis on any job page gives one file with the whole run and a ready-made prompt to attach in your own chat.
Yes, and they are excellent; most of us learned on them. The Lab is for a different way of working:
The Lab is a hobby project by Peter May, VK2ADG, at Jerrawa in the Southern Tablelands of NSW. It is not a business and is not trying to make money. It does cost money to run (servers, cloud time when it is busy, and the AI), so if it ever asks for a contribution, that is only to cover those costs. It is still being built: more is coming, and what gets built next depends a lot on what people ask for below.
Credits: the simulations use NEC-2, written by Gerald J. Burke and Andrew J. Poggio at Lawrence Livermore National Laboratory, through nec2c, the C translation by Neoklis Kyriazis, 5B4AZ. Any mistakes in how the Lab uses NEC are ours, not NEC's.
By queuing a job you accept these terms and use the results entirely at your own risk.
Testing is open to a small group at a time (7 testing now, 0 waiting). Join the list and you will be emailed when a place is free. Log in is by an emailed link: no passwords. If you have a job link, it opens directly. Testers with a code can still use it on the job form.
Ideas, an antenna you would like added, or something that is not working. We will see what we can do.