Companders and keys – What you need to know about analogue radiomics

Companders and keys – What you need to know about analogue radiomics

By Pyers Easton

Anyone who has ever used an analogue radiomic has probably heard of a ‘compander’. Over the years I have heard them called various thing such as ‘compounder’ or ‘compoonder’, but it is actually a portmanteau of the words ‘compressor’ and ‘expander’.

As many of you will know an audio compressor is designed to make quiet sounds louder and loud sounds quieter. This has the effect of bringing quiet sounds above the noise floor of the wireless system, and reduces the potential overload. The downside is the it destroys the dynamics of the sound , making it sound unnatural due to the compression applied.

The solution to the above issue is to use an expander in the receiver. An expander is basically a compressor in reverse, it makes quiet sounds quieter and loud sounds louder. It therefore stands to reason the if you put a compressor in the transmitter and expander in the receiver then you have a linear transfer function from input to output, therefore a natural sound. Both compressors and expanders have a number of parameters that need setting, such as ‘threshold’, ‘ratio’ and attack/release times. It is critically important that the compressor and expander be exactly matched in these parameters to assure transparent pass through of audio.

The use of DSP for the companding function can help to exactly match the compander with the expander as its accuracy is only limited by the accuracy of the programming and the clock frequency.

One thing to remember is that a compander only reduces the perceived signal to noise ratio because it reduces the gain in the receiver when there is low or no audio present. When there is sound present then the expander increases the gain so you can sometimes hear noise ‘pumping’ or ‘breathing’ if you have noise on the channel or the signal level is low. One way to test for this is to clap into the mic and listen for a noisy trail on the end, this is the release time for the expander allowing you to hear the actual signal to noise of the signal before the expander drops the level.

The primary method to reduce the impact of this pumping is to use pre-emphasis and de-emphasis on the audio. This involves boosting the high frequency content of the audio driving the TX and cutting it in the RX. One of the issues this produces if unchecked is that ultrasonic frequencies can modulate the compander and overload the audio stages. One of the classic compander tests is the ‘key test’ which involves jangling keys in front of the mic and listening to the result. From my own tests I have found that jangling keys produce maximum SPL at around 35kHz, which is way beyond the limits of human hearing. Microphone amps in many wireless transmitters on the other hand have a gain bandwidth that meets or exceeds the 20kHz limit of human hearing (for young people, mine stops at about 15kHz these days).

The impact of this is that with some systems the extremely high frequencies produced by jangling keys can overload the amplifier stages, relative to 400Hz for instance the gain for 50uS pre-emphasis at 35kHz is around 21dB, enough to put amplifiers and compressors into chronic clipping distortion, hence the terrible sound when you do the key test.

One way around this is to use less pre-emphasis or a low pass filter to shelve the pre-emphasis outside the audible range. Both of these solutions produce good results, but can be clumsy to introduce in the signal path in an analogue TX without DSP, remember that DSP is difficult to implement in a TX without significantly reducing battery life.

As a result the easiest way to reduce the ‘key effect’ is to reduce the pre-emphasis, thus reducing the gain at ultrasonic frequencies, so this has become the solution, high fidelity compander solutions use lower pre-emphasis curves such as 10uS to alleviate the effect of the high gain at ultrasonic frequencies disturbing the sound.

The only ill effect of such a solution is that the higher frequencies in the background noise are less attenuated due to the reduced de-emphasis, therefore the noise pumping may be more noticeable when the signal is weak, but this only happens right at the edge of range, or if you have on channel interference. But, when you have a strong signal there’s really not much difference to be perceived.

Written by Pyers Easton, MD of Raycom Ltd Evesham.

0 likes (You must login to like)

Leave a comment