[RTTY] Commercial Sound Cards: Internal vs. External & USB

[Kok Chen]

On Dec 12, 2012, at 6:43 PM, RLVZ at aol.com wrote:

> Others contend that the USB sound card is more likely to have  latency 
> problems, especially if a number of USB devices are plugged into  the computer, 
> and therefore the Internal sound card is  better.  

Buffer sizes and sampling rate defines the latency.

In Mac OS X, default sound buffer sizes are 512 samples.  At 48,000 samples per second, the buffer latency is 10 ms.  That is the duration of half of a 45.45 baud bit period!  I don't see how sound card latency can ever be a problem for RTTY decoding.

To get good ATC for selective fading, you typically want to add about 200 milliseconds of latency anyway to 45.45 baud RTTY so you can get a decent estimate of the envelopes of the Mark and Space signals.  The 10 ms is not even a blip.

> Would an internal Broadcast quality sound card like  the Digigram 
> VX-222 that has a S/N ratio of >-95dB, and very low distortion  such as 0.002%, 
> when installed in the average consumer  desktop computer, provide a good deal 
> better weak signal reception of RTTY  signals?  

If you are using a narrow filter ahead of the sound card that only lets through a single RTTY signal, the answer to you question is a definite no.  

If you use wideband approaches, or run a skimmer, then the answer is yes.  But probably not what you think the reason is.

The key to using a sound card, which lots of people miss, are:

1) you want the noise floor of the sound card to be 3 or more dB lower than the noise floor of the audio from your rig.  Considering that full scale of most sound cards are around 0 dBu (about 0.77 volt RMS), a well designed 16 bit sound card will have a noise floor of around 12 microvolts!  It is highly unlikely that noise floor of the line output of your receiver is that low.  Not too much below, since you will be wasting the dynamic range of the sound card.

2) once you have the noise floors set correctly, the only thing you need to remember is that the sound card must never, ever clip.  Never, ever.  Never, ever.  Never, ever.  This is what gets most people into trouble.  Turn the RF gain of the receiver down before the sound card clips.  Modems come with a VU meter for a reason.

If you are running narrow filters, it is highly unlikely that you even need 96 dB of dynamic range.  The only case I can think of is when there is serious flat fading that your receiver's AGC cannot handle (recall that selective fading don't require a large dynamic range, a good ATC will handle it).  Rigs like the Icoms don't even have that kind of dynamic range.  Remember that even the ST-8000 only has 75 to 85 dB of dynamic range.  

That being said, not all 16 bit sound cards are made equal; I have seen some pretty poor ones.  Both a 2004-era Griffin iMic and the microKeyer, for example have 16 bit codecs, but the microKeyer had 4 dB better dynamic range.  It has to do with the care taken to keep the noise floor low -- remember the 12 microvolts number above?   And many 16 bit sound cards are worse than the Griffin iMic.

Once you open up bandwidths, you will need more dynamic range since there will be the possibility of the sound card seeing a strong signal while the demodulator is trying to pick out a much weaker signal.  Even so, your receiver might still be the limiting factor -- there are probably only 5 or 6 rigs that can benefit from a sound card that has better than 96 dB of dynamic range; if you can afford a Hilbering, the cost of a sound card is not going to be a problem anyway :-).

When you widen the receive filter is when things like IMD come into play.  I.e., a strong signal at 1995 Hz and a 2125 Hz space signal causing third order IMD to land on top of the 2295 Hz Mark tone when the mark is not there.

Harmonic distortion is not a problem until you open up the bandwidth to over 1 kHz or so.  I.e., a strong signal with 800 Hz offset creates a second harmonic falling on top of a much weaker signal at 1600 Hz offset.  Because of this, if you run a skimmer, you need sound cards with low IMD and harmonic distortion.  I have measured a sound card which has 92 dB of blocking dynamic range, but a second harmonic distortion that is as poor as -70 dBc.

On top of it all, if you use a transformer between the radio and the sound card, the transformer probably has worse IMD and harmonic distortion than the sound card.  Combining a $100 sound card with a $0.50 transformer is just plain silly if you want to use the sound card in a skimmer -- but I have seen people do that.

IMHO, the big difference in weak signal copy is not with the sound card, since you can judiciously use AF attenuators, I.F. filters and the RF gain knob to make a mediocre sound card perform just as well as a $2000 sound card (you just have to pay close attention to signal levels).  The big difference between being able to copy or not copy a weak signal under poor propagation conditions is the modem.  Spend you money to buy some decent software modem instead, and it will do you much more good than using a good sound card with a crummy modem.  Just MHO.

73
Chen, W7AY