Specifying a coaxial cable looks like one of the simplest tasks in RF engineering, which is exactly why it is so often done badly. An engineer who would never sign off a link budget on a guess will happily write “RG58” on a drawing, order a reel, and move on, only to find at commissioning that the loss is too high, the connectors do not mate, the jacket has perished in the sun, or the assembly is generating intermodulation on a multi-carrier site.
The root of the problem is treating a cable as a single object rather than as a specification with several independent parameters. A coaxial assembly is really a combination of decisions: the impedance, the loss the system can tolerate at the operating frequency, the shielding, the power handling, the physical and environmental constraints, and, separately, a connector chosen for each end. Get any one of those wrong and the cable can be useless even when every other choice is correct. This article walks through those parameters the way you should think about them on the job, and points to the two tools that turn the judgement calls into numbers.
The most useful habit when specifying a cable is to work backwards from the two things it has to connect. Before any thought about cable family or loss, write down what sits at each end: the equipment, its connector interface and gender, the frequency and bandwidth passing through, the power level, and the environment the cable has to survive. Almost every specification mistake is really a failure to pin down one of those before choosing the cable.
Once the two ends and the path between them are defined, the rest of the specification falls out of them in a fairly mechanical way.
The first decision is impedance, and it is binary in practice: 50 ohm for almost all radio, transmit, and test applications, and 75 ohm for video, broadcast distribution, and CATV or DOCSIS systems. They are not interchangeable. Putting a 75 ohm cable into a 50 ohm system introduces a mismatch that shows up as return loss and ripple, and the reverse is just as true. This sounds obvious, yet 75 ohm cable and connectors are common enough in mixed installations that they find their way into 50 ohm jumpers more often than they should.
Attenuation is where most cables are actually won or lost, and the critical point is that loss is a function of frequency. A cable that is perfectly adequate at 150 MHz can be hopeless at 2.4 GHz, because attenuation per metre rises with frequency. Specifying loss therefore always means specifying it at the frequency you will actually run, not in the abstract.
This is the parameter to put a number on rather than feel your way through, and it is exactly what the Cable Loss Calculator is for: enter the cable type, length, and frequency, and it returns the loss so you can check it against the budget the system can afford. A receive feeder that adds a couple of decibels in front of a low noise amplifier is one conversation; the same loss on a transmit feeder is throwing away power you paid for at the radio.
Cable diameter is the lever that trades loss against practicality. Larger cable has lower loss, because a bigger centre conductor and dielectric mean less resistive and dielectric loss per metre, but it is heavier, more expensive, harder to route, and far less willing to bend. This is why a long antenna feeder up a tower is often half-inch or seven-eighths corrugated hardline, while the short jumper at the top is a flexible cable that can actually be dressed around a connector.
In practice most systems end up mixing families: a low loss main run to carry the distance, and a short flexible jumper at each end where bend radius matters more than the fraction of a decibel. The skill is putting the low loss cable where the length is and accepting the flexible cable only over the short hops.
Shielding decides how much of the outside world gets into the cable and how much of the cable’s signal leaks out. A single braid is cheap and flexible but leaks; a foil-plus-braid or double braid is far tighter; and solid or corrugated outer conductors approach a fully closed system. On a quiet bench it rarely matters, but on a crowded site with strong nearby transmitters, poor shielding is a direct path for interference, and it is worth specifying explicitly rather than accepting whatever the cheapest cable happens to offer.
For any transmit path, the cable has to carry the power without overheating the dielectric, and power handling falls as frequency rises and as the cable gets smaller and hotter. A thin jumper that is fine for a handheld will not survive a high power base station feeder. Power rating is one of the parameters that is genuinely dangerous to leave implicit, because the failure mode is a cooked cable rather than a quiet underperformance.
Wherever several carriers share a transmit path, passive intermodulation (PIM) becomes a specification in its own right. Poor connectors, contaminated interfaces, and low grade assemblies generate intermodulation products that fall back into receive bands and quietly raise the noise floor. On a single-carrier link PIM is usually irrelevant; on a shared cellular or land mobile site it can be the parameter that decides whether the site works, and it is a reason to specify low-PIM assemblies and the connector series that support them.
Most of the time the electrical length of a cable does not matter, but in phasing harnesses, antenna arrays, and some measurement setups it matters a great deal. The velocity factor sets the electrical length for a given physical length, and where two paths have to be matched in phase, the cable type and even the temperature stability of its dielectric become part of the specification rather than an afterthought.
Finally, the cable has to survive where it lives. A jacket rated for indoor use will craze and split after a season of Australian sun, so outdoor runs need a UV-stable jacket; buried runs need a direct-burial or conduit-rated construction; cabling in air-handling spaces may need a plenum or low-smoke zero-halogen jacket to meet fire rules; and exposed or rodent-prone runs may need armour. None of this affects the RF performance, but all of it affects whether the cable is still doing its job in two years.
Here is the point that trips up more engineers than any other, and it is worth stating plainly: a cable assembly does not have to have the same connector on both ends. The two ends are specified independently, by series and by gender, according to what each one has to mate with.
This is completely normal practice. A pigtail might be SMA-female at the radio and N-female at a bulkhead. A tower jumper might be N-male at one end and 4.3-10 male at the other. A test lead is routinely N-male to SMA-male. The cable in the middle is the same cable; the ends are whatever the equipment demands. Once you see an assembly as “cable type, plus connector A, plus connector B”, the whole thing becomes much harder to get wrong, because you are forced to look at what is actually on each piece of equipment instead of assuming symmetry.
Two things have to be nailed down for each end. The first is the series, which has to be both mechanically and electrically suitable: the right frequency range, the right power and PIM class, and the right size for the cable. The second is the gender, which is set entirely by the mating part, since a male connector mates to a female and nothing else.
A short reference of the common 50 ohm series:
| Series | Typical use | Notes |
|---|---|---|
| BNC | Test, low power, lower frequency | Quick bayonet coupling, not for high power or high frequency |
| TNC | Threaded BNC equivalent | Better at frequency and vibration than BNC |
| N-type | Antenna systems, feeders, general RF | The workhorse outdoor connector, weatherproof when assembled correctly |
| SMA | Small equipment, modules, test | Compact, common on radios and modules, easy to overtighten and damage |
| 7/16 DIN | High power, multi-carrier base sites | Robust, low PIM, large |
| 4.3-10 | Modern high power, low PIM | Smaller and lighter than 7/16 DIN, increasingly the default on cellular sites |
The reason this matters beyond simply mating is that every unnecessary adapter you add because the ends were specified wrong is another interface that adds loss, another mechanical weak point, and on a transmit site another potential PIM source. Specifying the correct connector on each end in the first place is almost always better than fixing it later with a bag of adapters.
A complete coaxial specification, then, is not a part number but a short list: impedance, cable family sized for the loss budget at the operating frequency, shielding appropriate to the RF environment, power and PIM class for the application, a jacket suited to where it runs, and a connector chosen independently for each end by series and gender. Work through those in order, from the two ends inward, and the cable more or less specifies itself.
Two of these are judgement calls best made with numbers rather than instinct. Use the Cable Loss Calculator to confirm the loss at your real frequency and length sits inside the budget, and use the Coaxial Cable Selector to compare candidate cable families against the impedance, frequency, loss, and power constraints you have written down, so the choice is a real comparison rather than a default.
The errors that recur are nearly always a parameter left implicit. Quoting loss without a frequency, and being surprised when a cable that was fine on the radio bench falls apart at microwave. Running a flexible jumper for a long feeder to save a fiddly install, and paying for it in lost decibels for the life of the site. Specifying an indoor-jacketed cable outdoors and replacing it a year later. Assuming a cable is symmetric and ordering the wrong gender at one end, then bridging the gap with adapters that add loss and PIM. None of these is exotic, and all of them disappear once the cable is treated as a specification with two independently chosen ends rather than a single line item.
At noIM₃, the coaxial tools are built to make this specification process fast and defensible rather than a matter of habit. The Cable Loss Calculator turns cable type, length, and frequency into a real loss figure you can hold against a budget, and the Coaxial Cable Selector lets you compare cable families against the constraints that actually matter for the job. Used together, they take the two hardest parts of the decision, the loss and the cable choice, out of the realm of guesswork, and leave you to do what an engineer should be doing: deciding what the system needs and specifying a cable, end to end, that meets it.
Related Insights
There is no single best frequency for long range communication. The right choice depends on distance, terrain, antenna size, licensing, noise floor, and required availability. Here is how to pick.
Apr 27, 2026
A practical refresher on free space path loss, the log distance model, Fresnel zones, and the two ray ground reflection effect: the four numbers every field engineer should be able to estimate before climbing a tower.
Apr 22, 2026
RF interference is becoming more common as wireless systems multiply and spectrum becomes increasingly congested. Learn what causes it, why it is worsening, and what engineers can do to mitigate it.
Mar 25, 2026
