[OZAPRS] TNC capture range

[Owen Duffy]

Hi,

On 11/10/2014 18:51, Darryl Smith wrote:
> Owen commented:
>> Isn't that entirely about VHF APRS using AFSK and Ham202 (ham adaptation
>> of Bell202) on FM LMR?
>>
>> It does not address the frequency accuracy issue that arises with FSK on
>> HF.
>>
> Kenneth has done more, some of which was in his presentation at the DCC,
> as opposed just to his paper. What he found was that there is that no two
> implementations of the Bell 202 are the same, with how they want the
> amplitudes of the tones, and their performance in noise, at base band.
Most modulator / demodulator pairs assume a flat channel, flat amplitude
and group delay response.

Group delay response is a bigger issue at band edges, and for AFSK, we
operate well inside the band edges of the radio's response. My own view
is that we have bigger problems with the amplitude frequency response.

FM LMR have preemphasis of nominally 6dB/octave from the bottom of the
voice band (say 200Hz break point), and complementary de-emphasis. These
characteristics are not perfect, and there is typically some residual
slope across the voice band, but not nearly 6dB/octave.

The original concept of AFSK AX.25 on these type of radios is that it
used an unmodified radio, ie audio in the mic and out the speaker
jacks... or equivalent. I don't know that this was ever expressed in a
standards document, even if it was, hams do not tend to pay attention to
such things.

So, apart from some hams thinking they knew a better way an bypassing
either or both of pre and de-emphasis... things worked ok. Failure to
use matching pre and de-emphasis at both ends of the link does not
prevent the link working, it just degrades performance (ie needs higher
S/N for same error rate).

Some might argue that a channel without pre AND de-emphasis is flatter,
and that is probably true, but a mixed channel is VERY poor. Now hams
being hams have devised recipes for mixed channels, you will hear people
say "we run our digi with no demphasis on the rx and we pre-emphasis the
tx"... but I have never seen credible quantitative tests to show that is
actually better.

So, as I said, most of the community had converged on a consistent /
compatible use of pre/de-emphasis regime and it was very convenient
because you could use the existing external jacks on a mobile or hand
held to do AFSK AX.25.

Then Kenwood apparently saw no reason to conform, all Kenwoods that I
have tested do NOT use pre/de-emphasis. They are not the only ones, the
digis that use Argent T3-135 Alinco combination do NOT use pre/de-emphasis.

Darryl, the track you laid through the Southern Highlands a few days ago
was excellent compared with some weeks ago when the area was flooded
with iGate traffic from Orange via Canberra. I took the trouble to
listen to the radio as you passed through and I decoded you well if you
were 40% or more on the S meter. By contrast, I recieve hundreds of
packets a day from VK1RGI as low as 20% S meter reading, even though
VK1RGI has only 50% of the audio drive that it should. A likely
explanation of why I note that Kenwood radios need to be stronger is
that my rx uses de-emphasis.

So, it all works, but it works better if the channel is flat, either by
not using pre/de-emphasis at both ends, or using matching
pre/de-emphasis at both ends.

Is there specs on this parameter? Well modem chip specs often
incorporate a "twist" parameter, and the maximim twist permitted is
typically 6dB for a Bell202 modem. So failure to use matching
pre/de-emphasis puts the system within a dB of its tolerance limit for
that factor alone.

But, it works... hams prove it every day.

================Picture removed, too large for the mail list ... it is
Figure 2 from Kenneth's paper.==================

Kenneth gives the figure above, but does not mention that it is not a
model signal, it is a classic result of de-emphasis after no
pre-emphasis. Why would he show a defective signal as representing the
modulation method?

>
> If there is a lack of data about what the best relative amplitude of the
> tones going into the modem decoder is for the best SNR, you have little
> hope of working out what to use.
It wouldn't matter, we hams will do what we like.

A fundamental question is whether the whole community should swing to no
pre/de-emphasis to follow the lead of Kenwood, because Kenwood is
unlikely to change?

We will not rid the system of incompatible stations... it is part of the
environment, it always existed with more general AX.25 networks to some
extent, but now to a greater extent.

We can't achieve consistent deviation level, and in an application where
excess audio drive is much worse than being short, the tendency is to
excess... all knobs to the right and some prominent digis lead by example.
>  At least that was what I got out of his
> presentation, and then spending too many hours in the bar that night
> talking to him about this stuff
That is good.
> Darryl
>
> ---------
> Darryl Smith, VK2TDS POBox 169 Ingleburn NSW 2565 Australia
> Mobile Number 0412 929 634 [+61 4 12 929 634 Int] - 02 9618 6459

73
Owen