OpenBCM V1.07b12 (Linux)

Packet Radio Mailbox

IW8PGT

[Mendicino(CS)-Italy]

 Login: GUEST





  
W6KWF  > ALL      01.05.20 15:16l 46 Lines 2161 Bytes #999 (0) @ WW
BID : 25409_W1XSC
Read: GUEST
Subj: Re: TO TESTIFY YOUR SOUNDMODEM
Path: IW8PGT<LU4ECL<JE7YGF<JH4XSY<N3HYM<N7HPX<GB7YEW<VA3TOK<N6MEF<W1XSC
Sent: 200413/2123z @:W1XSC.#NCA.CA.USA.NOAM [Santa Clara Co] #:25411 $:25409_W1

The video link didn't load for me, but assuming this is one of the 
tracks from the infamous WA8LMF test CD, it has a few issues:

1) Uploading the track to a random video site will run it through
 an unknown audio compression codec, which was not optimized for AFSK
 modem noises, so the resulting audio you get from downloading this
 will be very unrepresentative of the actually received audio.
 This is why the original WA8LMF site offers the recording as an
 uncompressed audio CD. I have extracted it as uncompressed wav files
 before, but I'm not certain where those ended up...

2) It being a random recording off the air has several issues:
  a - No one really knows how many packets should be in there, so
      as modems start to play with things like error correction (please
      don't), they start pulling garbage out of the noise and counting
      it as packets
  b - Many of the original stations captured in that recording have
      conplained since they have been plauged with that 40 minute
      recording accidentally getting I-gated to the APRS-IS every few
      years and their old packets coming back to haunt them.
  c - It's not like you're hearing that many different modems. When
      many of the packets have been digipeated through the same
      badly configured digipeater, all you're hearing is the same modem
      digipeating many packets with different original source addresses.

3) During processing, the de-emphasis filter accidentally applied a
 notch filter to the audio, which badly distorted the audio.


If you want to test a TNC's modem, someone should produce a synthetic
test suite with a known number of packets and step it through several
different parameters such as:
 - Decrease the signal to noise ratio in steps from perfect audio to 
     entirely white noise.
 - Have different lengths of preamble and the different opinions on
     preamble (strictly following the X.25 spec vs sending a sane sync
     preamble of all 0s.)
 - Play with different inter-packet spacings. Many modem state-machines
     badly handle the technically allowed single inter-frame flag.
 - Etc.

Kenneth Finnegan, W6KWF





Read previous mail | Read next mail


 23.12.2024 00:55:42lGo back Go up