[RTTY] Sound Cards

[Kok Chen]

On Jul 20, 2004, at 3:56 PM, N6OJ wrote:

> Can anybody tell me if there is a difference in demodulation quality
> between the $29.95 special sound card and the $500.00 super sound card

I have compared the built in A/D converters of a Macintosh (AGP G4 
model) against a 20-bit Griffin iMic and found that the external 20-bit 
converter has a noise floor that is about 10 dB better.  About 80-85 dB 
for the Mac and 90-95 dB for the 20-bit iMic.

Even the newer 16-bit Griffin iMic (what they are currently selling - a 
case of newer is not better) has a 5 dB better noise floor than the 
built-in A/D, the advantage of not being in the same Faraday cage as 
all the noise generators of a computer, I guess

An M-Audio Transit (24 bit converter, in the $100 price range) is very 
good, at least 5-10dB better than the iMic, and the M-Audio Quattro 
(forgot what I paid for it a few years ago, but somewhere north of the 
price of a KAM) is even better.  The Quattro is powered by its own 
brick supply and has balanced in/out, while the iMic and the Transit 
are USB bus powered with single ended in/out.

I have not tested the Transit with the optical TOS-lInk interface (the 
Transit comes with optical interface for both input and output, albeit 
at 16-bits). Only one of my computers right now has optical audio input 
and it is not in the shack.

So, moving the A/D converter outside your computer box can help.  So 
can more bits, BUT only if your operating system can make use the extra 
bits.  Another factor is the AGC range of your receiver and whether you 
are working with narrow filters.  If you are using narrow IF filters 
and only copying the largest signal in the passband and the rig has 
good AGC action, or you are willing to manually ride the receiver or 
A/D gain, then you would not need much dynamic range at the A/D 
converter.

Just my humble observations, based upon testing with a homebrewed 
floating-point based matched filter + ATC FSK demodulator program.

If the weakest detectable audio signal from the rig (post AGC) is at 
least 5% of the max audio from the rig (i.e., 26 dB range or a loss of 
4 bits full scale for the weak signal), I don't think you can ever see 
better copy with a better ("more expensive") converter.

But if the range of audio signals from the receiver is greater than 40 
or 50 dB, you probably will see an improvement with more bits in the 
converter if your operating system knows how to move more than 16 bits 
of sampled data.  Again, just MHO.

  73
Chen, W7AY