Setting up an amateur station is mostly a chain of decisions about loss: where you put the antenna, what you feed it with, how you match it to the transmitter, and how much power you actually need at the far end. The gear in the middle — the radio — is the easiest part and the least interesting. Almost every station that disappoints its owner does so because the antenna is too low, the feedline is too long or too lossy, or the operator assumed that the legal power limit would cover for a bad radiator. It won't. Get the antenna and the match right and a modest transceiver will work the world; get them wrong and no amount of power fixes it.
What the rules actually constrain, and what they don't
US amateur stations may run up to 1.5 kW peak envelope power on most bands, with two notable exceptions. In certain HF segments, when the control operator holds a Novice or Technician class license, the ceiling drops to 200 W PEP. And spread spectrum emission types are limited to 10 W PEP — a real constraint if you're experimenting with weak-signal digital modes that use spreading. At the top of the spectrum, stations in the 76–81 GHz band are capped at 316 W EIRP rather than a transmitter power figure, because at those wavelengths antenna gain is so large that transmitter power alone tells you almost nothing about what's actually radiated.
That last point generalises. PEP is defined as the average power supplied to the antenna transmission line during one RF cycle at the crest of the modulation envelope. Note what it measures: power at the transmission line, not power radiated. Your coax eats some of it, your tuner eats a little more, and the antenna's gain then multiplies whatever survives. EIRP — equivalent isotropically radiated power — is the other end of the same accounting: it combines transmitter output and antenna gain into a single number representing what an isotropic radiator would have to put out to match your signal in its strongest direction. A 100 W radio into a beam with meaningful gain can produce more EIRP than a 1.5 kW amplifier into a dipole, and it will do so on a fraction of the mains current. This is why experienced operators spend money on aluminium before they spend it on transistors.
The limits are also, in practice, not the binding constraint for most people. If you live in a city lot, deed restrictions, HOA covenants, landlord rules, and the sheer impossibility of a 40 m beam in a backyard will decide your antenna long before Part 97 does. The 1.5 kW figure matters mostly as a planning number: it tells you what your feedline, tuner, and connectors must survive if you ever intend to run it. Building a station around 100 W and then discovering your coax is only rated for a few hundred watts at your operating frequency is an annoying way to learn about velocity factor and loss.
One more framing point, because it explains why the rules are written the way they are. The Amateur Radio Service is defined in federal regulation as a radiocommunication service for self-training, intercommunication, and technical investigations, carried out by duly authorised persons with a personal aim and without pecuniary interest. That phrase does real work. You cannot sell your services, you cannot run it as a business, and you cannot accept payment for operating. What you can do is build, measure, fail, and rebuild — which is exactly what antenna work is.
Choosing bands before choosing hardware
Pick your bands from your property, not from a catalogue. The single most consequential fact about any HF antenna is its height above ground expressed in wavelengths, and that ratio is fixed by physics while your lot is fixed by your mortgage. A 20 m half-wave dipole wants roughly 10 m of height to behave like a dipole rather than a cloud warmer; a 40 m dipole at the same height is less than a quarter wave up and will radiate most of its energy steeply upward — fine for regional contacts a few hundred miles out, useless for DX. If your best available height is modest, plan around 20 m, 17 m, 15 m, and 10 m and use a vertical or a low inverted-V for the lower bands, accepting the compromise explicitly rather than discovering it after your first contest.
Low-band work has a different cost structure. On 80 m and 160 m, the antenna is almost always electrically short for the space available, which means loading coils, capacity hats, or a vertical with a good radial field. Radials are the part everyone under-builds: a vertical with a handful of short radials is a dummy load with ambitions, and the difference between that and a proper radial field is far more audible at the receiving end than any change of radio. The trade-off is labour and lawn disruption; the alternative is a magnetic loop, which is small and quiet and narrow-band and expensive per unit of efficiency.
VHF and UHF are a separate decision. A 2 m or 70 cm vertical on the roof plus a modest beam for weak-signal work will get you local repeaters, packet, and — when the troposphere cooperates — contacts well beyond line of sight. The cost here is mostly height and feedline: at VHF, every extra metre of cheap coax costs you real signal, and the numbers get ugly fast compared with HF. If your interest is satellites or EME, you are building a rotator system and a low-noise preamp at the masthead, and the budget conversation changes completely.
The full US allocation spans 1800 kHz to 250 GHz, which is a useful thing to remember when someone tells you amateur radio is one hobby. It is several, sharing a licence. Nobody builds a station that covers all of it. Build for the two or three bands you will actually use this year, and leave room to add.
Feedline, connectors, and the loss you can't get back
Buy the best coax you can afford and keep the run as short as you can make it. Loss in feedline is not a rounding error you can amplify away: it happens before the antenna, it costs you both transmitted power and received signal, and on receive it degrades your noise figure in a way that no preamp placed at the radio can recover. A long run of thin coax at 70 cm can lose more than half your power; the same run with a proper low-loss cable and quality connectors loses a fraction of that. The rule of thumb worth internalising is that feedline loss is symmetric and permanent — you pay it twice, once going out and once coming in.
Connectors are where good cable goes to die. Use the correct connector for the cable type rather than adapting, weatherproof every outdoor joint with self-amalgamating tape plus a layer over it, and make sure the shield connection is genuinely bonded rather than relying on a crimp that looked fine indoors. Water ingress into a connector shows up as gradually rising SWR over weeks, usually during a rainy spell, and by the time you notice it the connector is scrap. The cheap fix is a drip loop and proper sealing; the expensive fix is a new run of cable.
Balance is the other half of the feedline story. Coax fed to a balanced antenna without a balun puts RF on the outside of the shield, which means your feedline radiates, your shack becomes part of the antenna, and you get RF in the audio, touch-sensitive equipment, and unpredictable patterns. A 1:1 current balun at the feedpoint costs little and solves this. For open-wire or ladder line feeding a doublet, you get very low loss and multiband operation from a single wire, at the cost of needing an antenna tuner that handles balanced line and routing the line away from metal — a real constraint in a small garden.
Matching, SWR, and what a tuner does not do
An antenna tuner does not tune the antenna. It transforms the impedance it sees at the shack end of the feedline into something the transmitter is happy with, and in doing so it protects the finals and keeps the radio from folding back power. What it cannot do is reduce the loss already incurred in a mismatched feedline. If your antenna presents a high SWR at the shack, the tuner will make the radio happy while the feedline quietly dissipates a chunk of your power as heat — and the higher the SWR, the larger that chunk. This is why a tuner is a fine tool for multiband operation on a well-designed antenna and a poor substitute for one that isn't.
SWR itself is a crude diagnostic. A 1:1 reading tells you the load matches the line, not that the antenna radiates well — a dummy load is a perfect 1:1 and radiates nothing. A high reading tells you something is wrong, but not whether it's the antenna, the balun, or water in a connector. The useful habit is to log your baseline SWR per band when the station is new and dry, then watch for drift. A slow rise across weeks is almost always moisture; a sudden jump after wind is a broken element or a chafed feedline; a change after you moved furniture is coupling to something nearby.
If you're building a permanent multiband installation, an antenna analyser is the single most useful instrument you can buy after the radio. Sweeping an antenna in place tells you its resonant frequency, its bandwidth, and whether the transformer or balun is doing what you think. It also lets you tune a vertical or a wire without transmitting into a mismatch, which is kinder to the amplifier you may eventually add. The trade-off is cost and the temptation to over-optimise a number on a screen instead of making contacts.
Grounding, bonding, and lightning
Grounding serves three separate purposes and they are often conflated: electrical safety, RF return, and lightning protection. Bond your station ground, your AC service ground, and your antenna mast to a single common point at the building entrance — separate grounds at different potentials are worse than one good one, because a strike or a fault will find the difference. Inside the shack, a copper bus bar with short, wide straps to each piece of equipment keeps RF where it belongs. Outside, a properly installed ground rod array at the mast, bonded back to the service entrance, gives a strike a low-inductance path that isn't through your radio.
None of this makes you strike-proof, and that matters. The only reliable protection is disconnecting the feedline and the power when storms are forecast and leaving them disconnected until the weather has passed. Surge arrestors on the coax and the mains are worth fitting, but they are a second line, not a first one. The cost of a good grounding installation is a weekend and some copper; the cost of skipping it is a fire or a dead station, and insurers have opinions about unpermitted antenna structures. Check local codes before you sink a rod — the electrical inspector's requirements and the antenna's requirements are usually compatible but not identical.
RF grounding is a different problem from safety grounding and deserves separate thought. A station that gives you RF bites on the key, distorted audio, or a computer that reboots when you transmit on 40 m has a common-mode current problem, not a bad earth. Chokes on the feedline, ferrite on the microphone and control cables, and a proper balun at the antenna usually fix it. Adding a ground rod will not.
Power, amplifiers, and knowing when to stop
Run the least power that makes the contact. This is not a moral position, it's a practical one: lower power means less heat, less RF exposure, less strain on every connector and tuner in the chain, and a station that is easier to diagnose when something goes wrong. The jump from 100 W to 1.5 kW is a bit over 11 dB — under two S-units at the other end. On a quiet band with a decent antenna that is rarely the difference between a contact and no contact; on a crowded band with a poor antenna it might be, and that is the honest case for an amplifier. The honest case against is that an amplifier into a bad antenna mostly heats the feedline and annoys your neighbours.
If you do add one, the rest of the station has to be rated for it. That means feedline and connectors specified for the power at your operating frequency, a tuner with adequate headroom, and an antenna that can take the voltage at the feedpoint — high power into a high-SWR antenna produces very high voltages across baluns and capacitor hats, and things arc. Plan the whole chain at the legal limit even if you only ever run a fraction of it, and you will never be the person diagnosing a melted balun in the rain.
RF exposure is part of the same calculation and is easy to forget. Higher power, higher gain, and a closer antenna all raise the field strength where people can stand. Do the assessment for your actual installation rather than assuming the rules exempt you, particularly if the antenna is near a property boundary or a rooftop you can walk on. The remedy — moving the antenna, lowering power, or restricting access — is usually cheaper than the alternative.
The last decision is the one nobody puts in a build plan: what you will do when it doesn't work. Most first antennas are too low, most first feedlines are too thin, and most first matches are worse than the owner believes. Log the baseline, change one thing at a time, and measure before you buy. The station that gets built in a weekend and never touched again is usually the one that gets abandoned; the one that gets adjusted for a year is the one that ends up working everything.


