|
G'day Wolf,
I have been using this technique for some time now
with no problems. The difference is I having been using the 1000Hz
time pips from a local time and frequency station (VNG - 2.5MHz).
As long as you stick to the "standard" sample rates
of 11025Hz, 22050Hz and 44100Hz then the sample rate will be pretty much as
accurate as normal xtal oscillator (usually +/-100ppm). This
means there is no ambiguity (trying to decide which peak is the correct one) as
+/-100ppm is +/-0.1Hz @ 1000Hz. There is more than one peak
because the pips vary from a 50mS burst to a 500mS burst producing varying
relative amplitudes of the 1 second sidebands. The result is a
peak at 1000Hz, 1001Hz, 1002Hz.... (and going down 999Hz. 998Hz,
997Hz...). But as long as your card is within +/-500ppm then
you just pick the nearest one to 1000Hz. I have found that on
many signals the "biggest" one may not be the correct one, so be careful with
trying to sort out the correct one using Loran lines as they appear to be spaced
by 0.1Hz.
Doesn't WWV or other time frequency stations have
pips (maybe not 1000Hz).
8000Hz is a "semi-standard" frequency so it may
follow the same behaviour.
BTW, for those doing programming, avoid any
"non-standard" sampling frequencies (I can't say one way or another about 8000Hz
- simply that the error on my laptop at 11025Hz is +53ppm, while the error at
8000Hz is +300ppm) as I have found that, for example, using 5500Hz can be a
disaster for anything other than a 100.000% soundblaster soundcard.
On my VFSKCW program I have had reports ranging from mild inaccuracies to a
factor of 6 !!!! in output frequency. It seems to be fairly
common on laptops because they don't have soundblaster cards.
You might be interested in some measurements I did
using this technique a couple of years back. They are still relevant
to me as I still have the same gear :-)
Go to: -
Containing:-
ULF, ELF, VLF & LF
Experimentation
InfraSonic Experimentation
Laser Comms DX
Amateur Radio
Astronomy
=============================================
| 