[Lowfer] Re: NC overnight run

[Bill de Carle]

At 01:02 PM 3/1/2004 -0500, Jay Rusgrove wrote:
>Bill:
>
>I too had a screen full of question marks at times overnight. Is that
>something built into the program that is an indicator of possible
>interference?

We have seen this before on this particular frequency [186.050 Khz] and
when using MS100.  The 32-bit code in ET3 mode for question mark [?]
is 1FFF:FFFF - almost all 1's.  We're using differential-mark, which means
each of those 1's represents a phase inversion from one bit to the next.
The most likely scenario is an interfering carrier removed from our
nominal 800-Hz audio frequency by +/- 5n Hz where n is a small integer.
At MS100 the response of an integrate & dump filter (which we are using)
is a main lobe 20 Hz wide centered on 800 Hz, then diminishing sidelobes
(each 10 Hz wide) on either side of the center frequency.  So a carrier
15 Hz away, for instance, would hurt badly whereas ones at +/- 10 Hz, 20Hz,
etc. would be effectively notched out.  I have looked at the audio using
a FFT program and found there is a real nasty, ragged line at 850 Hz
[186.100 Khz] - some of which would get through because the spectral
energy is distributed over a range of frequencies near 850 Hz and would
get past the notch.  But what is probably causing the question marks is
another line at 845 Hz [186.095 Khz].  From one 100-msec bit time to the
next, a carrier at +/- 5n Hz displacement will produce the apparency of
a 180-degree phase change.  That is what causes all the 1's, which ET3
decodes as "?".

A good solution would be for Dex to move down, say 30 Hz to 186.020 Khz
to get further away from it (not too far down please because there is
another line below us).  Also, going to MS200 instead of MS100 would
place the comb notches every 5 Hz instead of every 10 Hz and one of
'em would fit right over that nasty little line at 186.095 Khz.  That
should take him out enough to enable good copy.

73 de Bill VE2IQ