[EMAIL PROTECTED] wrote: >The current cal scheme is unlikely to ever work well for most users. The >issue >is the error in the oscillator in everyone's sound card. Without taking this >into >account, most of us can achieve very good accuracy at one WWV frequency >but will find it is way off on another. My sound card was several hundred Hz >off (at 24 MHz). Unlike the DDS error correction, it does not scale with >tuned >frequency. I think this is on Bob's list of things to look at. > >Mike W3IP > > I'm not sure I follow why the sound card error would result in accurate results at one RF frequency but not another. Error in the sound card clock is like an error in the last local oscillator -- it generates a constant offset that doesn't change with frequency.
For what it's worth, the clock in the Delta 44 card (at least mine) seems to be very good. I have used a Delta 44 to drive a spectrum analyzer program under Linux for the last three ARRL frequency measuring tests, and I do frequency comparisons (between a pilot frequency and the unknown signal) with milliHertz resolution. While I haven't done phase noise tests, I've measured a precise reference tone, and the frequency accuracy over periods of tens of seconds to tens of minutes is very good -- at 1kHz, down in the 0.00x Hz range (i.e., parts in 10e-6, about what you'd expect from a good quality crystal). Since that's a direct offset that's not multiplied, the sound card's contribution to the RF frequency error is pretty minimal. Despite those results, my goal is still to have the whole frequency chain stabilized. One thing I've thought about is hardware hacking the Delta to use an external clock. I haven't dug into that possibility yet. Several of the other M-Audio allow an external "word clock" which the pro audio world uses to lock the sampling rates of multiple cards. It's a shame that feature isn't on the Delta 44 as it would allow an easy way to use an external reference. John