On Oct 22, 2006, at 9:23 AM, Bill Turner wrote:
> As a rough guess, in a
> contest where you make 1000 QSOs, you might lose one or two by not
> using one of the two listed above. Not a big deal, unless you lose
> first place by one or two QSOs. :-)
I agree with Bill's statement. But from the viewpoint of a DX chaser
and not a contester, I have found good modems to be far better on the
nerves than bad modems ("did the DX really respond to me?"). How
often have you seen other people's call signs go pristinely through
-- until the DX finally calls you and the characters are gabled? :-)
I have also found software modems to potentially be better in terms
of dynamic range than the ST-8000. It is very easy to get DSP to do
the right thing.
On top of that, you can use more sophisticated multi-path
countermeasures than the circuit in the ST-8000 to reduce mark and
space overlaps -- software can mimic that with two lines of code.
Basically, except for the analog AGC, it is easy to mimic everything
in an ST-8000 with today's fast computers if that is what you want to
do. And with a well designed implementation, you can beat the
ST-8000's 85 dB of dynamic range without using any AGC.
When conditions are close to the white Gaussian noise case, you can
also get that fraction of a dB in SNR by using a true matched filters
in a software modem -- granted, the advantage of a matched filter is
minimized when the channel deviates from the AWGN case -- for
example, when there is flutter.
In the days when I was using hardware modems I have found the
"goodness" of modems that I have used to be in about the following
order: MFJ 1278 - KAM Plus - PK900 - Modified KAM Plus - HAL DXP-38
- HAL DSP4100 - Timewave DSP 599zx - HAL ST-8000. And a good
software modem to be at the top or better in the list. Your milage
may vary.
In any case, I'll wager that a contest score will vary by more than
just a few points if you are copying with an old TNC versus an
ST-8000. Just miss decoding that really weak and fluttery signal,
and you could lose a mult.
The modified KAM Plus I used had some mods made to the switched
capacitor filters to tighten them up from the stupid Q that
Kantronics chose for Pactor baud rates (RTTY'ers are always an after
thought :-). NI6T has back then done the same thing to a PK-232,
but the PK-232 used filters built from op-amps rather than the switch
capacitor filters used in the KAM Plus.
The DXP-38 was a unit HAL had graciously sent me for evaluation which
I used as a second modem during an RTTY Roundup.
The DSP-4100 was a unit I had owned for a long time since it was a
standalone modem from way back that had worked with Macs (back then
HAL's amateur modems were PC bus driven).
I had picked up the ST-8000 in used condition (but it had gone back
to the factory for recalibration a few months before I bought it) --
it was bought with the sole purpose of using it as a "golden
standard" to compare software implementations against. The 599zx is
pretty much on par with the ST-8000 once you make sure the audio
chain is properly set up -- each seem to miss a character the other
one prints at about the same rate.
All modems will miss characters and even the best will miss a
character now and then when a lousy modem gets it correctly. But the
frequency of hits and misses is what distinguish the good ones from
the less good ones.
I think this is reason why many people use multiple modems. With
cocoaModem, I had implemented 9 separate FSK demodulators to watch a
single signal -- each demodulator has a slightly different relative
mark/space delay and a different ATC (automatic threshold correction
in the slicer) time constant. The nine outputs then goes through a
soft decision logic to print a single character to the screen --
basically, it is like a human watching multiple modems to decide
which character to use when they are different, but it is not as
smart as a human, it does not do context sensitive stuff that a human
brain is really good at; or the 123456... to QWERTY conversion in
Baudot. A good adaptive equalizer will be better than using multiple
demodulators, but I am not smart enough to design a good adaptive
equalizer for HF band conditions.
Now, with all that said, using a good modem can often be very
frustrating -- you can print more signals that cannot print you!
With a DXpedition to an uninhabited island, that is not usually a
problem. They have better SNR than you do at your own QTH and that
is where a good modem at our own end can be reassuring.
Still, when you donate a modem to a DXpedition, don't give them a
cast off. Give them the best modem money can buy. Whether they can
print your signal may depend on that HI. With software modems, a
good laptop is cheaper than an ST-8000! Send them a decent laptop
and A/D converter :-).
I have always contended (just my own biases) that when people report
inferior results from a software modem, it is often not because of
the implementation, but of how it is set up in the shack. Unless
your rig has a narrow I.F. filter which passes a single RTTY signal,
it is essential that the audio signal does not clip the A/D
converter. Even if you do use a narrow I.F. filter, clipping the
signal will put the demodulator into "FM limiter" mode, which is
often not what you want -- there are better ways for good modems to
handle QSB than using a limiter. (This, by the way is what the KAM
Plus calls "FM" mode -- it just clips the signal when it comes to the
box. The AM/FM button is really just a linear/limited signal
selection switch.)
On the other hand, you also don't want to reduce the audio gain of a
software modem to the point where you are working with just the least
significant 8 or 9 bits of your A/D converter - you will not beat an
ST-8000's 85 dB of dynamic range that way. The ST-8000 and DSP 599zx
for example, both have analog AGC -- these boxes are riding the gain
for you. So, although the 599zx is a digital box, it is probably
easier to set up than other digital boxes; it also has a built in AC
millivoltmeter so you can actually adjust proper signal levels -- but
all good software modems should have instructions for how you can do
that.
Just my two cents. Your mileage will definitely vary.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|