I agree with Chen to the extent that software solutions can rival fixed
hardware boxes... and in certain case with a lot of 'tweaking' exceed
them.
Unfortunately reality is a different teacher. A quick tune across the
band will verify that many software solution users are not set up
properly. This leads to all sorts of on-the-air problems, as we all know.
That's not to say the HAL, KAM, and PTC boxes are not all properly
installed and set up, either. Rather that TNCs are easier to set up
because they come with instructions while software stuff requires a lot
of patience or a dedicated Elmer to get it right.
Any comments?
Phil Sussman
Clayton, Ohio
------------
Quoting Kok Chen <chen@mac.com>:
> On Jan 28, 2005, at 5:40 PM, w4gkm@juno.com wrote:
> > I sure would like to get some input as to what everyone thinks about
> > the HAL ST-8000 vs. sound cards. If you have used the HAL and the
> > sound card what if any difference to you see? I have used them side
> > by and side and see little if any difference. At times one will print
> > when the other won't.
>
> I have a very old ST-8000 (S/N in the low 100's), so the following
> comments may or may not apply to anyone else. Also, I have only
> compared it against the only software modem I can run (cocoaModem), and
> the following are just my own opinions. I had bought this particular
> ST-8000 a little after the original owner had it checked out by HAL in
> January 2002 just so I have a gold standard to compare software
> algorithms against, but I didn't start writing a software modem until
> about the middle of 2004.
>
> The first thing to be mindful of is that ST-8000 comes with a very good
> audio AGC circuit which contributes to its 85 dB dynamic range. If you
> have a 16-bit sound card, you may have to "ride the gain" of the A/D
> converter between very strong and very weak stations to match that. If
> your A/D converter is capable of greater than 100 dB of dynamic range
> (noise floor to clipping), then you should have better headroom than
> the ST-8000.
>
> Be mindful, though that you will only get this maximum dynamic range in
> the digital world if you had carefully adjusted the audio gain from the
> rig so that the loudest possible signal _barely_ does not clip the A/D
> converter of the "sound card." Otherwise, just use the "armstrong"
> method and install a gain pot in the audio line.
>
> Many of the gripes you hear about software modems not working well is
> likely to be because the audio level is not properly adjusted or there
> is something else wrong with the audio chain (hum and harmonics of
> 50/60 Hz, for example). After connecting the audio chain up, use a
> software spectrum analyser to look at the rig's output to make sure
> there are no spurs or harmonics of the power line that is stronger than
> -80 dB below the strongest signal. Most "sound cards" do not come with
> balanced audio input (which both the HAL ST-8000, and a HAL DSP-4100
> which I also own, come with), so you may have to play around a bit to
> fashion a low noise audio chain.
>
> A software modem should be able to out-shoot the ST-8000's rather
> primitive multipath circuitry, assuming they'd made the effort to do
> so. The circuit in my old ST-8000 does not do much more than make sure
> that the mark and space demodulators outputs do not overlap. There is
> no true equalizer in the line.
>
> cocoaModem uses 9 separate demodulators, each with a slightly different
> delay line between the mark and space signals and also slightly
> different automatic threshold control (ATC, or adaptive slicer) time
> constants. The former is to counter the case where mark and space
> signals arrive at different times and are overlapping, and the latter
> is to lessen the harm done to auto thresholding by very rapid (of the
> order of 25 ms) QSB/flutter signals.
>
> The outputs of these demodulators then enter a soft decoder which does
> a majority vote on the bit patterns and the result goes to a character
> sync algorithm (just looking for the start bit is quite error prone)
> and then to the Baudot-to-ASCII translator to be printed. The behavior
> of the multiple modems is similar to watching the output of two modems
> and sometimes one would print correctly and sometime the other would
> print correctly -- even a KAM Plus occasionally prints a character that
> is misprinted by the ST-8000. In the case of cocoaModem, I have 9
> modems with slightly different characteristics watching the same signal
> and then doing a majority vote so you only see a single character
> stream coming out of it. (The "squelch" in cocoaModem is not based on
> signal strength, but is a threshold on how many of the different modems
> agree with one another -- maximum squelch threshold means that all 9
> modems have to agree perfectly to print, minimum squelch threshold
> prints everything.)
>
> > The reason for this is that I am thinking of going with the sound
> > cards with a SO2R to cut down on some of the equipment in the radio
> > room and thought that would be a good place to start but didn't want
> > to comprise the printing.
>
> I can't vouch for the software that you are using, but I myself have
> stopped using my ST-8000 and Timewave 599zx and have been exclusively
> using only software modems for half a year now.
>
> In addition to all of the above, there can be other differences in our
> set ups -- I use a couple of 24-bit A/D converters (M-Audio Transit and
> M-Audio Quattro) together with the MacOS's floating point sound
> library; all my filters are also implemented in floating point. I only
> use recursive filters in places that don't impact demodulation; the
> rest are linear phase FIR filters.
>
> Bottom line: your mileage may vary :-). But from my one person's
> experience, the potential of the software modem is definitely there.
>
> 73
> Chen, W7AY
>
> _______________________________________________
> RTTY mailing list
> RTTY@contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
>
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|