Audio Signal Processing

Audio Signal Processing.

Much debate in technical journals and magazine concern the subject of signal processing for audio signal processing. The subject and its implementation has trans-versed from RC filtering to switched capacitors filters, now onto Digital Signal Processing (D.S.P.), but without good constant audio gain circuitry, D.S.P. processing can be driven into signal distortion due a too higher a signal input. There also a minimum signal has also a problem, in other words the signal dynamic range of the D.S.P. circuitry.

R.F. signal processing of the voice audio for the transmitter section of the radio is an art form that that has many of its followers. However a D.S.P. function here can be quite useful. The input audio can be processed using a mathematical form, i.e. to use a square root function on the TX audio before modulation would be useful, or to even use a log₁₀ function would also reduce the high peek of the human voice. Too return the received audio back into its original form, an x² or antilog₁₀ mathematical functions could be used.

The shift and width control of the I.F. filtering is a useful facility, something that unless achieved by analogue I.F. stage, can otherwise only be done by D.S.P. function. The shift and width control once understood is a very useful function to have for a radio receiving section.

A D.S.P. circuit should really not cost the earth. Derived from perhaps an audio baseband, a 16bit A/D convertor for Hifi use can be utilised then perhaps a CPU of perhaps of “Raspberry PI” type fame could be used. Finally a 16bit D/A chip can be used. The point is that the modern D.S.P. circuitry costs are very high, or the profit margin is great. Essentially an I.F. stage is synchronously demodulated to a zero I.F., here the fourier principle of calculation is used upon the RF band plan to pick out the intended signal. However, the fourier equation can be substituted by a digital oscillator at the intended signal frequency, here by replacing the fourier calculation. After this point, an audio derived D.S.P. unit say for intended HiFi application, can now be used for the audio filtering requirements, plus any other signal processing. The specification needs for audio processing are far different as for a Spartan derived radio equivalent, I assume.

The demodulation could be achieved by using a pair of D.B.M each fed with either an “I” or “Q” BFO signal. This would allow the demodulation of all analogue voice modulation, AM, NBFM, WBFM, SSB and also the data signal of CW, RTTY, Slow scan TV, FAX and PSK31.

Now this is a point, as once the digital oscillator has completed it's work to bring the intended signal to a zero I.F., the sampling rated for the audio signal need only be twice the maximum audio bandwidth (Nyquist rate), say 4KHz and a 12KHz sampling rate. The Spartan alternative uses a sampling clock of some 300MHz, assumingly I am correct, while the circuitry costs a far greater than a 12KHz HiFi circuit alternative.

However even here alternative D.S.P. approaches can be made. The D.S.P. 12KHz alternative can sample at the RF level, say the 9MHz I.F. stage. The digital oscillator output need only pulsed using a 12KHz sample clock thereby gating the R.F. output of the digital oscillator in a RF bursts cycles gated by a 12KHz clock as a frequency counter would. The 12KHz gated R.F. burst digital oscillator would in effect sample the intended signal at the R.F. point say 9MHz I.F. or even say 14.060MHz carrier frequency, to make a point, and in doing so bring the R.F. signal directly down to the 12KHz sampling clock I.F., ready for D.S.P. processing with D.S.P. audio circuitry. Remember that the gating clock, the 12KHz sampling clock, even gating the digital oscillator sampling intended R.F. signal, need only be twice the information rate, namely the audio upon the R.F. carrier.

The microphone connection has to be as simple as possible. I would suggest a stereo microphone connection as the stereo connection that would normally be used can be separated into mono microphone and PTT switch. Perhaps microphone of the radio transmitter should isolated by an audio isolation transformer, separating the grounding path. The output speaker may be wise to tackle in a similar circuit manner.

For CW use, the use a paddle key and straight key should be facilitated. For the paddle key, an Iambic Morse circuitry should be included.

Some MP3 players use small scale D.S.P., the MP3 player cost in the order of £50, so the low sampling clock version with the gated digital oscillator alternative may see the day of light.