Is there a proper way to connect multiple radios via coaxial cable so they can be tested without going over the air?

Question

Let me state up front that I am not an RF engineer, but I do have a EE background.

I'm working with a group that has a collection of radios, VHF and UHF (military) that is licensed to use a VHF and UHF frequency for over the air testing. They have also created a lab with a run of coaxial cable with taps at intervals connected to inline RF attenuators to which they connect these radios for isolated testing or for using alternate frequencies. They are typically using the same frequency (at a low power setting) and usually take turns transmitting (although this is not guaranteed). It seems to work, often quite well; however, it strikes me as potentially problematic. While these radios are naturally capable of voice communications, the bulk of their use involves digital data. I have observed situations where it is difficult to reliably exchange data and I struggle with the suspicion that it could be due to this connection configuration. So I'm seeking some opinions/guidance from those who probably far better versed that I.

First, is this even a legitimate thing to do? Second, is there a best way to do it?

I admit I know probably just enough to be dangerous, but my concerns when all of these transmitters are basically just connected to and operated over the same piece of wire:

• Impedance/termination
• vSWR
• Cable run length/tap spacing

I sometimes feel like when this does work, it's really by accident and when it doesn't it just confirms my suspicion that it is a dodgy implementation.

Answer 1

My company built a system like this, to allow real radios to be used in a flight simulator. It had fixed and digital variable attenuators, so it could simulate exactly the effect on communication of (say) rolling the plane, etc.

Important notes would be:

• high power 20 dB attenuators on each of the radios, first, to take care of the bulk of the power.
• then additional (variable) attenuators to adjust things as you wish
• double screened cables everywhere - RG400 is a good choice.
• Typical attenuation would be around 60 dB (approx 100 m range) to 110 dB (30 km). So you would need 30 to 50 dB of attenuation on every radio before running the cables to the coupler or next (attenuated) radio.

If you want to simulate weak signals, or flying out of range, the radios still need to be physically separated because everything leaks a small amount. 100 dB shielding is considered excellent, and you need >120 dB. Keep the first few attenuators close to the radios. Lots of physical distance. Metal screens with barrel connectors through them, ferrites on coax and power cables, etc.

It's simple to test whether you have enough shielding: disconnect and remove all the cables, leaving only the 40 dB attenuator on the radios, and see if they can communicate.

If you're connecting more than two radios, you can simply make an awful many-connector box, no need for special matching, as there are so many other attenuators in the system. You just have to measure through, to find out the final attenuation. It should definitely be a single-point box though, think strange things will happen if you run a long piece of cable and connect many radios to it.

Apart from just massive overload, or some sort of protection being triggered by very high VSWR, I can't think of any good reason they would fail to communicate. Multipath, or multiple echoes, would be no worse than in the real world.

Answer 2

A run of coaxial cable, with an attenuator, is a perfectly good way to simulate an “over the air” connection. It is done all the time in labs.

(However, note that it does not automatically mean your system is not radiating. Signal may escape from the circuits through the chassis, the power supply wiring, or any other wire connections to the transmitter or receiver — hopefully not enough to exceed regulatory limits, but certainly enough to be received by a nearby receiver not attached to the cable system.)

Addressing your specific concerns:

I have observed situations where it is difficult to reliably exchange data and I struggle with the suspicion that it could be due to this connection configuration.

The likely failure mode of a wired system is too much signal (causing distortion of the signal at the receiving side). This can be addressed by additional attenuation.

• Impedance/termination
• vSWR
• Cable run length/tap spacing

Normally, one does not “tap” a coaxial cable. If they are doing so, then that will indeed produce undesirable RF reflection and thereby present incorrect impedance to the transmitters (which could damage the transmitter if it is not protected against this, or merely cause it to reduce transmit power).

Reflections can also interfere with high data rate digital signals (they are artificial multipath, which is normal and must be tolerated for HF, but not necessarily for UHF). The way to think about this is: divide the speed of light by the length of cable, and compare that to the symbol rate of the digital modulation in use. If the reflections are much faster than the symbol rate then they will not be significant; if they are similar or slower then they will interfere with the signal.

The simplest way I can think of to reduce this problem is to insert attenuators on all sides of each tee in the circuit (or at least — one attenuator on each line between tees in addition to between the tee and the radio). This will damp the reflections, and bring the impedance closer to expectations. (The larger an attenuator's attenuation is, the more of its ports' impedance is made up of internal resistance rather than the line attached to the other port.) Perhaps there is also some kind of symmetrical power divider design that uses fewer total components and will provide all the desired coupling and impedance; I don't know.

It is also possible to design a tapped coaxial line with predictable characteristics. Famously, 10base5 Ethernet used this system. However, I am not familiar with it and cannot advise you on how to build one. Perhaps someone else can write a better answer about “bus” / “shared medium” design.

Also, at the ends of the line, there should be something with the proper impedance — which can be either a radio, or a terminator (dummy load) that absorbs all the incoming RF. This is more straightforward than branching.