RTTY
[Top] [All Lists]

Re: [RTTY] Checking Sound Card performance on RTTY with Audio Analyzer

To: RTTY Reflector <rtty@contesting.com>
Subject: Re: [RTTY] Checking Sound Card performance on RTTY with Audio Analyzer
From: Kok Chen <chen@mac.com>
Date: Thu, 02 Jun 2011 22:22:18 -0700
List-post: <rtty@contesting.com">mailto:rtty@contesting.com>
On Jun 2, 2011, at 6:32 PM, Joe Subich, W4TV wrote:

>> b) How much could I likely improve the S/N ratio with a better
>> quality sound card?
> 
> None ... you're not close to using the real dynamic range of even
> a poor sound card since most transceivers have an audio output range
> of only 30 to 40 dB.  Note - the microHAM interfaces have a dynamic
> range (internal noise floor to clipping) greater than 96 dB.


Be careful when comparing that "96 dB" number with an analog modem's dynamic 
range, such as the 85 dB of dynamic range of an ST-8000.  They are not the same 
-- you need to also account for quantization noise when dealing with sound 
cards.

A weak digitized signal needs to be a few bits above the sound card's least 
significant bit, otherwise it will be affected by quantization noise.

A signal that is 3 to 4 bits (18 dB to 24 dB) above the noise floor might be 
the minimum quantization tolerable because you need at least 10 to 15 dB of SNR 
to copy 45.45 baud RTTY through CCIR Flutter conditions.  That will set the 
limit of the real minimum signal that can be copied through a sound card.

The 10 to 15 dB number also varies with the software modem -- you can go take a 
look at Alex VE3NEA's plots.

The SNR for CCIR Flutter conditions is always measured against Gaussian noise 
which is uncorrelated with the RTTY signal.  Quantization noise is correlated 
with the RTTY signal, so who knows whether it is better or worse than this 10 
to 15 dB range.  I have never bothered to measure modem performance against 
quantization noise, since it is easy to buy a better sound card if you need 
more dynamic range :-).

The need for this extra 3 to 4 bits of headroom reduces the dynamic range to a 
hypothetical "96 dB" sound card by providing only ( 96-24 ) dB to ( 96-18 ) dB 
of dynamic range between the smallest signal that you can copy to the largest 
signal that you can copy.  Depending on band conditions, even good 16 bit sound 
cards will only provide 72 dB to 78 dB worth of usable signal strength range.

Something like the ST-8000 does not have quantization noise, so its 85 dB of 
dynamic range is fully usable.  Pretty darn good for 1980s technology, eh?

Having said all that, Joe is absolutely right.  Even 72 dB is still plenty of 
dynamic range for the audio output from most rigs (especially when used with a 
narrow roofing filter and with AGC engaged).  Your rig is most likely to be the 
weak link, not the sound card.

50 dB of sound card dynamic range is probably adequate if you are using a 
narrow roofing filter that lets through only one RTTY signal, and if you have 
adjusted the audio levels properly. 

My measurement of the Elecraft K3 for example shows that while the Headphones 
output has about 80 dB of IM dynamic range, the Line Output (what most people 
use for digital modes) of the K3 gives only 68 dB of IM dynamic range.  See

http://homepage.mac.com/chen/Technical/K3IMD/IMD/k3imd.html

The reason for the poor line output performance is probably due the audio 
transformers in the line output of the K3.  I have since replaced the original 
K3 transformers with the Bourns LM-NP-1001, but have not gone through to make 
another set of measurements.  Yet :-)  

The original K3 design was even worse that what I measured above.  I had 
already applied K8ZOA's modifications prior to making those measurements.  

If you are home brewing your own digital interface, it is probably worth paying 
careful attention to any isolation transformer that you use.  In many cases, 
you are better off with no transformer at all between the rig and the sound 
card, and instead, carefully bonding the ground of the sound card to the ground 
of the rig (see that first photo in the web page above, where I had used the 
balanced input of the E-MU sound card, and grounding the E-MU to the rig with a 
short piece of wire).  An inexpensive transformer is almost bound to have a 
worse IMD than the IMD from your sound card -- sometimes it is better to remove 
ground loops by starting with no ground induced noise in the first place :-).

In short, the sound card is very often not the weakest link in your audio 
chain.  Fix everything else first, and then, come back to the sound card.

73 (back to counting decibels :-)
Chen, W7AY

_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty

<Prev in Thread] Current Thread [Next in Thread>