RF Test Connectors: SMA, 3.5 mm, 2.92 mm and N Compared

A connector is the first and last part of every RF measurement. It sets the highest frequency you can trust, decides what you can plug into what, and, through wear, decides how repeatable your results are. This article compares the four common families and gives a rule for choosing one.

The four families at a glance

Connector	Upper frequency	Mates with
SMA	specified to 18 GHz	3.5 mm, 2.92 mm
3.5 mm	specified to 26.5 GHz	SMA, 2.92 mm
2.92 mm (K type)	used to 40 GHz	SMA, 3.5 mm
Type N (50 ohm)	commonly lower microwave	N, or via an adapter

The first three form one compatible group. The 3.5 mm connector is compatible with SMA. The 2.92 mm connector mates with both SMA and 3.5 mm. You can therefore build a chain from mixed parts without adapters, which is convenient and also a trap, as the worked example below shows.

Type N is larger and rugged. The 50 ohm version is commonly used at lower microwave frequencies. Its size means it does not plug directly into the SMA group, so crossing between the two needs an adapter.

Why the frequency rating matters

Every connector is a small discontinuity in the line. At low frequency it is electrically tiny and its reflection is negligible. As frequency rises, the same physical discontinuity becomes a larger fraction of a wavelength, and the reflection grows. The upper frequency rating is the point up to which the manufacturer specifies that reflection stays acceptable.

Using a connector above its rating does not stop the signal. It adds reflections that your instrument reads as part of the device under test. The relation between return loss and the reflected fraction of the voltage is:

A larger return loss (RL, in dB) means a smaller |Γ|, so a quieter echo. See Return Loss vs Insertion Loss for how to read these numbers.

Mating compatibility is not the same as matching rating

Because SMA, 3.5 mm and 2.92 mm mate with each other, it is easy to join a 40 GHz-class part to an 18 GHz part. The joint will close and the signal will pass. But the rating of the pair is that of the weaker part:

The same rule applies to a whole chain: cable, adapter, attenuator, fixture and device port. One SMA part anywhere in the path limits the whole path to 18 GHz.

Worked example: a mixed chain

A test chain has a 2.92 mm cable, a 3.5 mm adapter and an SMA attenuator. The intended sweep goes to 26.5 GHz.

1. Ratings: 2.92 mm is used to 40 GHz, 3.5 mm is specified to 26.5 GHz, SMA is specified to 18 GHz.
2. Chain limit: min(40, 26.5, 18) = 18 GHz.
3. Shortfall against the sweep: 26.5 − 18 = 8.5 GHz of the sweep lies above the weakest rating.
4. Fraction of the 2.92 mm capability used: 18 / 40 = 45 %.

The fix is to replace the SMA attenuator with a 3.5 mm or 2.92 mm one. The chain limit then becomes 26.5 GHz, which just covers the sweep. If the sweep went to 40 GHz, the 3.5 mm adapter would also have to change to 2.92 mm.

Care: wear, pin depth and torque

Connectors are intended for a limited number of mating cycles, and wear degrades repeatability. A worn connector still passes signal, but each reconnection gives a slightly different result. This matters most for calibration and for any comparison between two measurements.

Three habits protect both the connector and the data:

• Check pin depth before mating. A pin depth out of tolerance can damage the mating connector. A pin that protrudes too far can push into its partner. Check suspect parts with a gauge before they touch a good port.
• Use a torque wrench. It gives repeatable connection force, which improves the repeatability of the measurement. Hand-tightening varies from person to person and from day to day.
• Keep a sacrificial adapter on the instrument port. Mate and unmate the adapter, not the instrument. When the adapter wears, replace it.

Which connector for which test

• Up to 18 GHz, general bench work: SMA is enough, and parts are plentiful.
• Up to 26.5 GHz: use 3.5 mm throughout. It still mates with SMA, so existing SMA parts can join at the edges of the setup, but then the sweep must stay at or below 18 GHz.
• Up to 40 GHz: use 2.92 mm throughout.
• Lower microwave, rugged and frequently mated: Type N, with an adapter where it meets the SMA group.

When in doubt, choose the rating by the highest frequency you will sweep, then verify every part in the path against it.

In plain words · for students and beginners

Connectors in plain words

An everyday analogy

Think of a water hose with fittings. Every fitting must match the hose or water leaks at the joint. A connector is a fitting for a signal. At low frequency a small gap hardly matters. At high frequency the same gap leaks a lot.

The analogy breaks in one place. Water leaks out and is lost. A badly joined RF connector sends part of the signal back, like an echo, and your instrument hears it.

A bigger return loss means a quieter echo, which is what you want.

The same example in simple steps

You have three parts in a row. Each has a top frequency it is made for.

1. The cable is 2.92 mm, good to 40 GHz.
2. The adapter is 3.5 mm, good to 26.5 GHz.
3. The attenuator (a part that lowers signal level) is SMA, good to 18 GHz.

A chain is like a team where the slowest runner sets the pace. The slowest part here is the SMA one, so the whole chain is good to 18 GHz. If you want to test up to 26.5 GHz, swap the SMA part for a 3.5 mm one. Then the slowest is 26.5 GHz and you are covered.

Mistakes beginners make

• Assuming "it screws together" means "it works at that frequency". Mating and rating are different things.
• Tightening by hand. A torque wrench, a tool that stops at a set force, makes every connection the same.
• Never checking the pin. The centre pin sticks out a set distance. If it is too long it can damage the other connector.
• Plugging and unplugging the instrument port all day. Each mating wears it a little.

Remember

• The slowest connector in the chain sets the frequency limit: SMA 18 GHz, 3.5 mm 26.5 GHz, 2.92 mm 40 GHz.
• SMA, 3.5 mm and 2.92 mm mate with each other; Type N is bigger and needs an adapter.
• Check pin depth and use a torque wrench, because wear and loose joints ruin repeatability.

Related pages on RF View

• Return Loss vs Insertion Loss: What Each Number Tells You