[aprssig] More efficient use of channel capicity through shorter packets

[Ron Stordahl *]

There has been recent discussion concerning the use of higher baud 
rates, shortening packets through shorter TXDelay and a shorter payload 
through the use of Mic-E and eliminating extraneous added text.

I would be very much interested in knowing exactly how long (time) APRS 
packets are and with that information I could better understand exactly 
what could be gained from this approach.

I realize there is both a TXD (at the head of the transmission) and a 
TXTail (at the end).  Between these is data as required by the 
specification.  Part of that data is of fixed length and other variable 
length.  For example as a packet is repeated the path information builds 
up (and it does not appear to be insignificant in size).  If I 
understand the ax.25 spec the path is part of a variable length field.  
Then there is the 'payload', which can be large, but could be small if 
everyone used Mic-E without extraneous extra text.

Experimentally I have determined that for MFJ1270C's and Motorola 
Micor's that between them packets can be reliably decoded with a TXD as 
small as 90 milliseconds.  This is for relatively strong signals.  Below 
90 milliseconds reliability drops off rapidly.  Since relaying by high 
digi's does constitute a major share of the traffic, reducing their TXD 
(and TXTail if possible) would seem worthwhile.

With UIDIGI firmware there is no user option to set TXTail, and I do not 
actually know what it is hard coded as.  Ill take a guess that it's 
around 100 milliseconds, but could easily be wrong.  Without a way to 
control this I have no way to tell experimentally how it's length 
effects reliable decoding.

The Motorola Micor's are crystal controlled and work very well.  I do 
not know if one could run a modern synthesized mobile radio with TDX as 
low as 100 milliseconds...my recollection..and it is from many years 
ago..was that such radios required a longer TXD for their frequency to 
settle down after key down.  This may no longer be true.

Or it could be that the TNC itself requires a certain minimum TXD and 
TXTail. 

I expect someone here who is more current in the technology can explain 
the limiting factors in TXD and TXTail and offer numbers which could be 
used reliably with current hardware.

Ideally I would like to have a formula into which I could plug TXD, 
TXTail as well as the variable length elements in the packet and get the 
total channel time for 1200 baud packets.  I may be asking for the 
nearly impossible here..looking at the AX.25 spec makes my head spin.  
But with that it would be possible at least estimate what improvement 
could be anticipated by minimizing the elements over which we have some 
control.

Ron, N5IN