[aprssig] Clock/timing accuracy when generating an APRS signal

[Stephen H. Smith]

On 8/31/2011 8:28 AM, Richard Corfield wrote:
> I've found in experiments the opposite. I need an accurate clock source in my 
> generator for it to be received - at least by the AGWTracker software.
>
>

Are you using a soundcard softmodem like AGW Packet Engine with Tracker?

If so, there will be wild variations in timing if you are running it on the 
"brain-dead" parts-reduced basic sound system embedded in the motherboard of 
most modern PCs.

Classic add-on sound cards did most of the heavy lifting of decoding and 
synthesizing audio with dedicated hardware controlled by an accurate 
crystal-controlled timebase.

The cheap motherboard-based systems have replaced most of this with a simple 
A/D converter and massive software stacks running on the machine's main CPU 
instead.   The sound card sampling processes derive their timing from 
interrupts asserted on the CPU.  Depending on how many other processes are 
running at the same time competing for attention by the CPU and the number of 
pending interrupts, the latency in acknowledging interrupts can vary wildly.    
This contributes to random phase jitter in the decoding/tone generation process.

Further, depending on how many other programs (including background apps like 
firewalls, virus scanners, instant messengers, printer status monitors, etc) 
are running at a given time, the sample rate timing in sound card apps can 
actually change by hundreds of parts per million.     If you go through the 
sample rate calibration drill on a sound card app like mmSSTV or MixW, and then 
start up power-hog programs like MS Outlook, Photoshop, Firefox, etc, you will 
actually see the sample rate in the sound card app, as shown by their 
calibration utilities, change by 50-100 ppm.

[This el-cheapo approach has become nearly universal on PCs since CPU 
throughput went through the stratosphere with the first Pentium One MMX cpus.  
The CPU has so much throughput, most unused, it's cheaper to burn up CPU clock 
cycles and interrupts simulating a sound card in software than spending $5 (and 
the necessary PC board real-estate) on dedicated sound card chips. This same 
"do-away-with 
hardware-and-dump-it-on-a-fast-CPU-with-huge-software-drivers-instead" design 
approach is used with modern cost-reduced Windows GDI printers, and the final 
generation of dial-up 56K "WinModems". ]

Substituting a "classic" hardware-based sound system, either internal like a 
Soundblaster PCI card or an external USB-connected one,   will make a MAJOR 
difference in the performance and reliability of sound-card softmodems.    Note 
that you don't need a fancy card with Dolby 5.1 surround, "SRS 3-D audio", 
etc.  Just a basic $20 two-channel PCI sound card.

External sound systems   (including the Tigertronics SignalLink USB which is 
actually an external sound system combined with a ham interface in a single 
box)    have another advantage:  The analog part of the sound system is removed 
from the firestorm of digital noise, switching power-supply hash and AC 
ground-loop hum present on most motherboards.     The noise floor of the analog 
input on an external device is often 10-20 dB lower, which makes a noticeable 
difference in real-world  weak-signal copy, by not adding noise to a signal 
already degraded by less than full quieting.

-------------------------------------------------------------------------------

--

Stephen H. Smith    wa8lmf (at) aol.com
=== Now relocated from Pasadena, CA back to 8-land (East Lansing, MI) ===
Skype:        WA8LMF
Home Page:          http://wa8lmf.net

=====  Vista & Win7 Install Issues for UI-View and Precision Mapping =====
     http://wa8lmf.net/aprs/UIview_Notes.htm#VistaWin7

*** HF APRS over PSK63 ***
    http://wa8lmf.net/APRS_PSK63/index.htm

"APRS 101"  Explanation of APRS Path Selection & Digipeating
   http://wa8lmf.net/DigiPaths