An occasion presented recently to implement an APRS station on Australia's national 2m frequency, 145.175MHz.

Initial results with equipment known to work on previous occasions was disappointing. Though the channel was quite busy during the morning drive time test, too busy for good performance, the only packets being decoded were from the Mt Ginini digi some 180km distant whilst packets from the nearby digi at 17km were not decoded at all despite being full scale on the TS2000 receiver. (The local digi has conflicting LAT/LON and location name, they are 10km or so apart.)

Previous experience is that AFSK transmitters are setup very poorly, not just home stations but network infrastructure like APRS digis and BBSs, and the TS2000 is unable to decode grossly distorted signals.

An Icom IC2200H was tried, with no success. FM mobiles usually have broader filters than the TS2000, and this radio is designed to work with a digital voice add in module, but unable to decode 1200bps AFSK from the local digi, though again, it could decode signals from Mt Ginini which were in the noise. An Icom IC208H was tried, but again unable to decode the local digi. An Icom IC7000 was tried, but again unable to decode the local digi.

Then a cheap Chinese Feidaxin hand held was tried, surely it should have a broader IF filter. Yes, that worked and allowed capture of the spectral distribution of the local digi.

Digging in the cupboard revealed another FM mobile, a Yaesu FT2800m which was tried, and that worked ok with the local digi.

Figure 1:

Fig 1 shows the spectral response of a burst from the local digi using the FT2800M receiver.

First observation is that the sync period runs for about 250ms after the rx mute has opened, on this occasion more than half the channel utilisation is for possibly excessive sync.

Next, and a strong hint of decoding problems, is the very high energy level below 1kHz. This is a signature of an overdriven AFSK transmitter.

Figure 1:

Fig 2 shows the spectral distribution during the data part of the packet. The high energy level below 1kHz is due to intermodulation distortion created by a non-linear audio path.

Figure 3:

More revealing is to look at the slope across the in-band part of the data spectrum. Fig 3 reveals a slope of -5dB per octave. This slope is a signature of an overdriven AFSK transmitter.

When an FM mobile is driven with excessive level AFSK signal, the limiter (if present, and they usually are) limits the peak signal level, and if sufficient AFSK signal is provided, the deviation due to MARK and SPACE frequencies will be approximately equal effectively defeating the pre-emphasis circuit which is usually prior to the limiter. De-emphasis circuit in the receiver will deliver a slope of approximately -6dB per octave on the receiver audio output.

A separate measurement of deviation using a Motorola 2009D shows peak deviation is a steady 4kHz, no sign of pre-emphasis.

This transmitter has been setup with a simplistic "all knobs to the right". It has excessive deviation, an over driven limiter, no pre-emphasis, and occupies more bandwidth than it should... hence the inability to decode it with receivers with better quality IF responses.

One of the problems in the APRS network is that 'infrastructure' stations such as this that are heard over a wide area are mistakenly taken by struggling hams as exemplar setups.

The observation that the Mt Ginini digi could be decoded on the TS2000 when the signal was barely discernable from the background noise whilst the local digi could not be decoded though it was a very strong signal is a patent example that correct setup will allow your own packets to be heard by more stations, and at weaker signal levels.

The test equipment used for the graphics above was an ordinary VHF mobile, a PC and audio spectrum analysis software.

Anecdotal evidence is that most AFSK transmitter lineup is done by ear, ie without measuring instruments of any kind, and off air monitoring reveals the outcome. In 25 years of use of AFSK on VHF for AX.25, nothing much has changed.



Version Date Description
1.01 04/01/2012 Initial.

© Copyright: Owen Duffy 1995, 2017. All rights reserved. Disclaimer.