Connecting two parts of a system becomes complicated when your signals are up in the gigahertz range. Let’s take a look at some new cabling options.

This article is not about a revolutionary ultra-high-speed processor, or an amazing new 24-bit ADC, or a highly integrated system-on-chip that is currently threatening to become self-aware and take over the world. No, in this article we’re dealing with something far less exotic and impressive.

But like so many other things in life, lack of glamor does not imply lack of importance. Years ago I worked on an RF communication system that consisted of two rather sophisticated data links. Despite the advanced technology involved, we would have gotten nowhere fast without our abundant supply of basic SMA hardware—cables, attenuators, 50 Ω terminators.

 

These new cables use 1.0 mm screw-type connectors. Image courtesy of Pasternack.

 

At the same time, though, I don’t want to downplay the careful design that goes into very-high-frequency cabling products. All kinds of complicated things happen when signal frequencies are way up in the gigahertz region, and the last thing we want is a meticulously designed RF system that starts misbehaving because the test cables are inadequate in one way or another.

 

Standing Waves

One way of characterizing a cable is by measuring the voltage standing wave ratio, or VSWR. You can also characterize an antenna this way. The basic idea is to determine how much energy is reflected from the cable (or antenna). More reflected energy is bad—we want our RF signal traveling through the cable to the load, not being reflected back to the source.

 

 

A VSWR of 1 is ideal; this indicates that no energy is reflected. Pasternack’s new RF cables have a maximum VSWR of 1.5. Is that good?

Obviously the dividing line between “good” and “bad” VSWR depends on the requirements of a particular system. Nevertheless, it is safe to say that a value of 1.5 is quite good. A VSWR of 1.5 corresponds to a reflection coefficient of 0.2, and if we take the square of the reflection coefficient, we find that only 4% of the power is reflected back to the source.

By the way, why did I say that the “maximum” VSWR is 1.5? Well, VSWR varies with frequency. The Pasternack cables are described as having a frequency range extending to 110 GHz, and the VSWR for a 110 GHz signal is 1.5. The spec is even better for lower frequencies (e.g., 1.3 at 50 GHz).

 

Flexure

One interesting characteristic noted in the product description is that these cables exhibit “excellent phase/amplitude stability under flexure.”

I mentioned above that life in the gigahertz world can be rather complicated. Often we would hardly think twice about how a system might be affected by the movement of cables between subsystems. You could design a robot with ribbon cables or what not connecting the various portions of the system, and as long as your connectors are solid, all kinds of bouncing and twisting would have no significant effect on, for example, motor-drive or I2C signals.

But RF signals are much more sensitive to the physical medium through which they are traveling. I assume that these effects can be analyzed and described in scientifically and mathematically robust ways, but that’s beyond my intellectual comfort zone. I prefer to evaluate the situation in the context of my hands-on experience, and consequently I will describe it thus: gigahertz-region RF has a mind of its own, and if you twist the cable the wrong way, or forget to tightly screw on the connector, or wear a brightly colored blazer, or perform an important test on the day after a full moon, the system will fail.

Exaggeration aside, it’s annoying when you are afraid to rearrange a device because the resulting cable flexure could seriously alter the performance that you are trying to observe or characterize. Maybe these Pasternack cables can help to mitigate this problem.

 

Capacitance

Another thing to keep in mind is that coaxial cables aren’t just conductors with low resistance. The proximity between the internal conductor and the surrounding shield results in nontrivial capacitance, and this capacitance is measured per unit length: a longer cable means more capacitance.

 

 

As usual, this parasitic capacitance results in a low-pass response, so less cable capacitance is desirable in high-frequency applications. Here is the spec for the Pasternack cables:

 

 


 

Feel free to leave a comment if you would like to share any RF-cabling tips or misadventures.      

 

Comments

0 Comments