Setup of FM tx for broadcast relay from a digital audio source
This article proposes a procedure for setup of an FM transmitter for relay of the WIA News from a digital audio source.
The intention is to help relay stations reliably set the audio level of their relay to a standard, eliminating relays of the quality of this off-air recording of a studio recorded segment of the RD opening this year. As an Amateur, are you proud of this relay?
The WIA News is distributed as a digital sound file. A digital sound file in simple form is created by sampling an analogue audio waveform at points in time, and storing a digital representation of those samples. More complex formats include compression, either by way of a non-linear digital code, or compression of the digitised stream.
The format of the WIA National News distribution is 32 Kb/s 44kHz sampled mono mp3. This format includes digital compression.
This article will refer to the sound level on a digital sound file in decibels relative to the maximum sample value. For example, 16 bit PCM audio samples are a signed 16 bit value with 32,768 being the largest value, and 1 being the smallest non-zero value. The convention is that the maximum sample value 32,768 is the reference level (0dB or 0dBFS), 1 is 20*log(1/32768) or -90.3dB.
The practice is to record the source files in at least 16bit 32kHz sampled mono Wave format to provide adequate quality for further editing without serious degradation. The source files should be recorded without clipping, ie the peak program level must not exceed 0dB.
In putting the broadcast to air, a compromise between audio quality and readability under poor signal to noise conditions is necessary.
Typical FM transceivers used by amateurs are designed specifically for voice, and have an audio limiter and pre-emphasis network ahead of a modulator to achieve FM with a 6dB/octave pre-emphasis across the entire voice band (which makes it effectively PM). There are a range of designs for the audio processing, ranging from simple clippers to more complex equaliser / limiter / compressor designs. The limiter can be relied upon to lift the average volume level to a certain extent, but should not be grossly overdriven because of the degradation in quality of recovered audio.
Because of the pre-emphasis which is typically added after the limiter, the limiter does not level deviation, but rather the input to the pre-emphasis network. The limiter is typically set to limit peak deviation to 5kHz when modulated by a high amplitude 1kHz tone.
Analysis of anchorman Graham Kemp's segments on several broadcast files indicates that the RMS value is from -15dB to -20dB. It would seem that a good target RMS level for recorded material is -18dB.
As a result of a series of experiments on broadcast content using both mobile and fixed stations, it seems that a reasonable compromise between lifting the average volume level for mobile listeners and acceptable quality for home station listeners is to overdrive the audio input by up to 6dB on peaks. This is equivalent to sufficient audio to cause 10kHz deviation (were it not for the limiter) with a 0dB tone at 1kHz. This has the effect of lifting the RMS level to around -12dB.
The broadcast audio must be fed into a transmitter audio input that has normal voice pre-emphasis, eg the microphone or packet interface set up for 1200 Bd audio. Do not use a transmitter audio input intended for high speed packet as it is unlikely to be pre-emphasised. The line output of a sound card is likely to be around 100 mV, whereas the microphone input is likely to be closer to 10 mV, so a suitable smoothly adjustable attenuator network that is flat across the passband and is DC isolated from the transmitter microphone input (some MIC inputs include DC bias for an electret microphone, some use DC signals on the MIC lead for PTT) is required.
If the operator has access to a deviation meter, she could line up the audio levels by applying a known test tone well below the limiter threshold and adjusting audio levels for the appropriate deviation measured with a deviation meter. For example, a 1kHz test tone at -20dB under this 6dB overdrive regime should cause a peak deviation of 5*10^((-20+6)/20) kHz, or 1kHz.
Most hams will not have access to a deviation meter. An alternative approach is to set the modulator by observing the Bessel first carrier zero. This method is easier to perform than it is to describe, so don't be daunted by the description. There is also a link to an audio demonstration of the technique to assist.
This method exploits a characteristic of the spectral distribution of energy in an FM wave. As the level of a sine wave modulating signal at 1kHz is increased from zero, the power in the carrier component (which can be observed with an SSB receiver) will decrease to zero and then increase again. At the point of this first carrier null, the modulation index is 2.4, and with the 1kHz modulating signal, this corresponds to 2.4kHz of peak deviation.
So for line-up, if a test tone at 1kHz at at a level that should result in 2.4kHz deviation is injected from the sound source, then the modulation can be adjusted up from zero to the point where the first carrier zero is observed. So, under this 6dB overdrive regime, the test tone level for the first carrier zero is 20*log(2.4/5)-6dB or -12.4dB.
This is a procedure to set the tx level using an SSB receiver to detect the first carrier zero.
The modulation index is now 2.4, and therefore the deviation is 2.4kHz. Peak Program Level will attempt to drive the transmitter to 10kHz deviation, but will be limited by the transmitter audio limiter.
The technique is very sensitive, be careful to not pass straight over the first carrier zero. The limiter may (should) prevent sufficient deviation to observe the second carrier zero at m=5.5. Not only is the technique very sensitive, it is also very accurate when the frequency of the tone is accurate (as from the digitally synthesised test file supplied below), and suitable for calibration of instruments.
You have read about the procedure, click to listen to a demonstration. This demonstration uses an SSB receiver with a 3.5kHz IF bandwidth, but I have used the technique with receivers with a 10kHz IF bandwidth, you just hear more of the sidebands, but need to concentrate on the carrier beat and null it out. The test receiver could be a high quality communications receiver or a scanner with a BFO or LSB/USB mode. You could sample the modulated signal at the carrier frequency, or by sniffing some signal from the IF of a super-heterodyne receiver.
Note: This procedure depends on a single sinusoidal modulating signal, CTCSS tx must be disabled temporarily to perform the lineup. If you use CTCSS on the broadcast channel, you could do the lineup on another non-CTCSS channel on a the same band. You will also need step-less control of the audio level (eg a potentiometer), so do not try to use the software controls of sound level which act in discrete steps any may hop right over the null point.
Unable to find a clear standard for setup of an IRLP node other than the model audiotest.wav file, the audiotest.wav file was analysed for peak level and RMS level to provide clues for a standard setup. The peak level was -1dB and the RMS level was -18dB. This is sufficiently similar to the broadcast content discussed above as to indicate that the same setup procedure is appropriate (ie -12.4dB 1kHz tone to cause 2.4kHz deviation which coincides with the first carrier null).
Unable to find a clear standard for setup of an Echolink node other than the model ECHOTEST message, the ECHOTEST message was analysed for peak level and RMS level to provide clues for a standard setup. The peak level was -3dB and the RMS level was -18dB. This is sufficiently similar to the broadcast content discussed above as to indicate that the same setup procedure is appropriate (ie -12.4dB 1kHz tone to cause 2.4kHz deviation which coincides with the first carrier null).
In addition to the test tone file proposed above, Echolink has its own internal test tone source which could be used for setup. To correctly set the levels for an Echolink node using its own test tone, use the Echolink menus to send a 1kHz test tone at -12dB to the transmitter and adjust the tx audio for 2.4kHz deviation using a deviation meter or first carrier null as described above.
A digitally synthesised test tone file is useful in setup of the FM side, it should be played from the same source as the broadcast sound file. The test tone file contains the following:
For operators with access to a deviation meter, the 1kHz test tone at -20dB should be used to set 1kHz peak deviation. Having set the audio level, the passband flatness can be checked using the other tones, the 500 Hz -20dB tone should cause 500 Hz deviation and the 2kHz tone should cause 2kHz deviation.
Download the test tone file in mp3 format here.
A checklist to improve chances of success:
V1.06: 12 March 2009 13:31
© Copyright: Owen Duffy 1995, 2020. All rights reserved. Disclaimer.