The Capacitor Decides Your Magnetic Loop: Voltage, Arcing, and the 25 Watt Ceiling
A magnetic loop's power rating is a statement about its tuning capacitor, not about its loop. Here is the voltage math, why loops arc, and what to check before you buy.
When a magnetic loop is advertised at 25 watts, that number is not describing the loop. Copper tube does not care about 25 watts. The rating is describing the tuning capacitor, and specifically the voltage at which its plates flash over.
This article contains affiliate links. I may earn a small commission at no extra cost to you.
Once you know that, most of the odd behavior of small transmitting loops stops being odd. The power ceiling, the cracking noise on humid mornings, the retune every time the sun moves, the way the antenna shifts the moment you take your hand off the knob: all of it comes from one part.
What 100 watts actually looks like inside the loop
A magnetic loop is a series resonant circuit. The loop is the inductor, the capacitor tunes it to resonance, and at resonance the circulating current in that loop is enormous while the radiation resistance is measured in milliohms.
The voltage across the capacitor is that current multiplied by the reactance, and the reactance is in the hundreds of ohms. The result is kilovolts.
Numbers for a 1 meter diameter loop of 22 mm copper tube at 100 watts, computed from loop inductance, skin-effect conductor loss, and small-loop radiation resistance:
| Band | Loop reactance | Efficiency | Circulating current | Capacitor voltage |
|---|---|---|---|---|
| 80m (3.6 MHz) | 55 ohms | ~2% | ~66 A | ~3,700 V |
| 40m (7.1 MHz) | 109 ohms | ~16% | ~52 A | ~5,600 V |
| 20m (14.2 MHz) | 218 ohms | ~68% | ~27 A | ~5,800 V |
| 15m (21.2 MHz) | 326 ohms | ~90% | ~14 A | ~4,500 V |
| 10m (28.5 MHz) | 438 ohms | ~96% | ~8 A | ~3,400 V |
These assume a lossless capacitor and lossless joints, so treat them as the optimistic end. A real loop with a wiper contact and clamped joints runs lower Q and somewhat lower voltage. The kilovolt scale holds regardless.
Two things in that table are worth sitting with. Circulating current peaks at the bottom of the range, where efficiency is worst, because nearly all your power is being dissipated as heat in the conductor. Capacitor voltage peaks in the middle, around 40m and 20m. A loop that survives 10m at 100 watts can still arc on 40m at the same power.
Voltage follows the square root of power
This is the single most useful consequence, and it runs opposite to intuition.
Circulating current is the square root of power over resistance. Voltage tracks current. So quadrupling your transmit power only doubles the voltage across the capacitor.
A 25 watt loop and a 100 watt loop are not separated by a factor of four in anything physical. They are separated by a factor of two in plate spacing. Going the other way, backing a 100 watt loop down to 25 watts cuts the stress in half, which is why an arcing loop often behaves perfectly at QRP levels and why running low power is a real fix rather than a workaround. The same reasoning is behind the power ceiling described in why QRP is probably your ceiling indoors.
It also explains the market. The jump from a 25 watt loop to a 100 watt loop is not a bigger version of the same part. It is a different class of capacitor entirely, and the price step reflects that.
The three capacitor types, and what each one costs you
| Type | Breakdown | ESR and loss | Typical power | Cost |
|---|---|---|---|---|
| Air variable with wiper | ~1 kV per mm of gap in dry air | Wiper contact carries full circulating current | 10 to 25 W | $20 to $60 surplus |
| Butterfly or split stator | Same per-millimeter figure, two gaps in series | Rotor floats, no current through a rubbing contact | 100 W and up | $60 to $200 |
| Vacuum variable | 5 to 30 kV depending on part | Lowest loss available, no humidity sensitivity | 500 W to legal limit | $150 to $400 used |
The middle column is where homebuilt loops quietly fail. In an ordinary single-gang air variable, the rotor connects to the circuit through a bearing or a spring wiper. That contact sits in series with the loop, carrying every one of those 50 amps. A few tenths of an ohm of contact resistance is a rounding error in most circuits and a catastrophe here, because the total loop resistance you are competing with is measured in hundredths of an ohm. The contact becomes the dominant loss, efficiency collapses, and the part heats.
A butterfly capacitor removes the problem by geometry rather than by better materials. The rotor forms two capacitances in series with the stator halves and connects to nothing, so no current has to cross a moving joint. If you are building rather than buying, a butterfly variable capacitor is the part that makes 100 watts plausible, and a vacuum variable capacitor is what makes it comfortable.
Vacuum variables have one wear item that surplus buyers miss. The bellows that lets the shaft move while holding vacuum has a finite cycle life and can fail after decades on a shelf. A capacitor that has lost its vacuum looks identical from the outside and breaks down at roughly the same voltage as air.
Arcing: the failure that gets worse every time it happens
Arcing announces itself as a sharp crack or a buzz from the loop, a sudden SWR excursion, and ALC misbehavior on the rig. Some operators hear it as a raspy note on their own monitor.
The mechanism that matters is what happens afterward. An arc pits the plate surfaces and leaves microscopic points behind, and a point concentrates the electric field. The next flashover happens at a lower voltage than the last one. A loop that arced once at 100 watts on a damp morning will arc at 90, then at 70. The power ceiling ratchets downward and does not come back without dressing the plates.
Everything that triggers a first arc is environmental, which is why the same antenna behaves differently week to week:
Humidity and dew are the largest single factor. The 1 kV per millimeter figure for air is a dry-air, sea-level number, and it degrades sharply as moisture condenses on the plates. An outdoor loop at dawn is at its most vulnerable point of the day, exactly when 40m is good.
Dust, spider webs, and insects bridge the gap directly, and a web across a 2 mm gap conducts once it is damp. Attic loops accumulate this faster than anyone expects. Salt film does the same within a few miles of the coast, and it does not wash off with rain alone.
Tuning through resonance at full power is the self-inflicted one. Sweeping the capacitor while transmitting at 100 watts drags the loop through its highest-voltage condition with no warning. Find resonance at 5 to 10 watts, then bring the power up. That habit alone prevents most first arcs.
Enclosing the capacitor helps with all of the above, with one condition attached: a sealed enclosure that breathes through a temperature cycle condenses water inside itself. Vent it low, or accept that you have built a humidity trap around the highest-voltage point in your station.
Why the loop drifts, and why that is the same problem
High Q means narrow bandwidth, and the magnetic loop roundup covers what those bandwidths look like per band. The relevant number here is 40m, where a 1 meter loop has a few kilohertz of usable bandwidth.
A capacitor is metal plates held at a spacing by a metal frame. Warm the frame and the spacing changes. A shift of a few hundred parts per million in capacitance is mechanically trivial and electrically enough to walk a 5 kHz bandwidth clean off frequency, and the loop conductor expands at the same time, moving the inductance in the same direction. Sun reaching one side of the loop is sufficient. So is sunset. A motorized loop retunes faster but does not escape this.
Hand capacitance, and why remote tuning is not a luxury
Total tuning capacitance on a mag loop runs somewhere between 20 and 250 pF depending on band. Your body standing next to the capacitor adds a few picofarads in parallel. On a circuit where the whole value is 40 pF, three picofarads is a large fractional change, and the loop retunes the moment you step away. Manual tuning appears to work right up until you let go of the knob.
There is a second reason, and it is not about convenience. The capacitor gap is the highest electric field point anywhere in your station. Standing at it while transmitting at 100 watts is the worst position you could pick for near-field exposure, and it is the scenario the RF exposure evaluation is built to catch.
Two workable answers. A 6:1 vernier reduction drive on a long insulated shaft gives the resolution to land on frequency and puts distance between you and the gap. A stepper motor and controller run back to the operating position solves both problems, and is what every commercial base loop uses.
A loop's capacitor is not doing the job of the automatic tuners in antenna tuners for compromise antennas. A conventional tuner presents a 1:1 match to your rig while the antenna stays out of resonance. The loop capacitor brings the antenna itself to resonance, which is the only thing making it radiate.
What to check before you buy
Ask what the capacitor is. If a seller cannot answer, the answer is an air variable with a wiper.
Check whether the power rating is stated per mode. The Alpha Antenna EmComm loop publishes 100 W PEP SSB, 50 W CW, and 25 W digital, and that spread is honest engineering rather than hedging. SSB has a low duty cycle, so average heating sits far below the peak. FT8 is a continuous carrier for 13 seconds, and the capacitor sees that as a thermal load the SSB rating never tested. If digital modes are your plan, and in constrained spaces they usually are, size the loop on the digital number.
Confirm the tuning is remote if the loop will live in an attic or on a roof. The Alpha BASEMAGLOOP and the motorized Ciro Mazzoni Baby Loop both do this. A manually tuned loop in an inaccessible location is an antenna you will operate on one frequency.
Measure rather than trust. A NanoVNA sweeping the loop shows the actual resonant bandwidth, which is a direct read on total loss. A measured bandwidth much wider than the published figure means resistance somewhere in the loop is eating your power, and the wiper contact is the first place to look.
If you only need to receive, none of this applies. A broadband amplified receive loop has no high-voltage resonant capacitor at all, which is why those products cost a fraction of a transmitting loop and never arc. The decision tree covers when splitting transmit and receive is the better answer.
The recommendation
For a transmitting loop that will see digital modes at real power in a restricted space: butterfly or vacuum capacitor, 3 mm minimum plate gap, remote or geared tuning, and a vented enclosure over the capacitor. Buy on the digital-mode power number, not the SSB headline.
For an existing loop that has started arcing: drop to 25 watts, find resonance at low power before raising it, and inspect the plates for pitting and webs. The antenna is not finished. Its ceiling has moved, and that ceiling is a mechanical property you can measure and sometimes restore.