[RTTY] Product Review: Computer Sound Cards for Amateur Radio

[Kok Chen]

On Apr 24, 2007, at 1:03 PM, Jim Reisert AD1C wrote:

> There is a product review in May 2007 QST, "Computer Sound Cards  
> for Amateur
> Radio".  ARRL members can read the article at:
>
>     http://www.arrl.org/members-only/prodrev/pdf/pr0705.pdf

For my own simple investigations (your mileage will, I am sure vary),  
here are my two cents...

If there is only a single RTTY signal present and there is no  
multipath present, and the rig has good AGC, you need very little  
dynamic range.  10 dB is sufficient.  If you are using FM techniques  
(hard clipping, PLL, etc) rather than linear demodulators, then you'd  
need even less dynamic range.  RTTY tones do not overlap.  You only  
need to confidently see a single tone at a time.

With a PSK31 signals, you will need more dynamic range than with RTTY  
since you can locally generate IMD with just a single PSK signal by  
itself.  Ditto the case of multipath RTTY when the mark and space  
signals overlap substantially.  Practically, however, not enough of a  
difference to worry about as long as you don't hard clip that PSK31  
signal on receive.

So...

... if you are doing RTTY or PSK31 and you are using an I.F. filter  
that only lets through a single signal at a time, there is no need to  
use a good A/D converter.

You also don't need extra dynamic range to handle signals of  
different strengths if the rig's AGC can put out a moderately  
constant audio levels.

However...

... if you are using wide and frequency agile interfaces (I use 2.4  
kc passbands not just for PSK31 but also for RTTY and CW) the story  
is different.

To take full advantage of your rig, you'd want the dynamic range of  
the A/D converter to be better than the dynamic range of your rig  
plus perhaps an extra few dB so that the weakest signal won't appear  
as a square wave -- although in the practical case, the weak signals  
will substantially be dithered by noise, and an A/D converter with  
the same dynamic range as the rig is perfectly adequate.

With wideband interfaces (e.g., if you are using a 2 kc wide  
waterfall or a spectrum to find and tune signals in) you will also  
want to choose an A/D converter whose second harmonic distortion (and  
perhaps even the third harmonic distortion) is low.

If the second harmonic of a loud 800 Hz tone is only 50 dB down for  
example, it can interfere with a weak signal whose tone is at 1600  
Hz.  This case is less of a problem since you can retune the VFO of  
the rig a little to move the harmonic of the loud station away from  
the weak signal.  But you need to know that there is a weak signal  
there in the first place, so you can go turn the VFO knob, so having  
low harmonic distortion is still a good idea :-).

If you are using a narrow I.F. filter, you also need not worry about  
harmonic distortion as long as the software has even a moderately  
good bandpass filter.

All this assumes that your software is written to make use of the  
extra dynamic range, otherwise having extra dynamic range in the A/D  
converter will not do you any good.

If the software is implementing filters with 16 bit integers, you  
won't benefit from a 24 bit converter.

I cheat and I use floating point in the software that I write, so I  
need to worry about this aspect.  The resident sound framework of the  
OS that I use hands the samples to me as floating point numbers.  So  
as long as the manufacturer has written a real 24 bit driver for the  
A/D converter, I don't have to worry about the dynamic range of  
numbers once it reaches my programs.

Even though MFSK16 is more complex, it behaves from the dynamic range  
viewpoint pretty much like an RTTY signal.  MFSK16, too only  
transmits one tone on at any one time.

Bottom line for me: dynamic range is cheaper to attain at the A/D  
converter and demodulating software that it is to attain in a  
commercial rig. You don't need it to be much larger than the dynamic  
range of your receiving chain, but it is not worth letting it be the  
limiting factor.

How many contacts will you lose?  Probably 0.01%.  But that one in  
ten thousand  contact could be a P5.

Speaking of which, many people lost the chance to work P5/4L4FN  
really because their modem was stuck at 45 baud when Ed was operating  
at 50 baud, as he often did.  Something simple and nothing to do with  
dynamic range :-).

73
Chen, W7AY