Phase meter w/ 56002 EVM

[Andre Kesteloot]

Talbot Andrew wrote:

> >-
> >From: boffin1 [mailto:boffin1@compuserve.com]
> >Subject: LF: Phase meter?
> >I have belatedly seen your queery about phase.  In the  abstract their is
> >no answer to your question for phase reqires a reference i.e the phase of
> >something relative to ....???     (i.e another part of the wave).   In the
> >limit the phase must always close;  Nature abhors odd bits of phase
> >73, Roger,  G2AJV.
> ------------------------------------
> The original question, I believe, was for software that would show the phase
> of received signal / audio tone based on an internally generated reference.
> This would obviously have to be at a specified frequency, but could easily
> be generated by software using a Numerically Controlled Oscillator (a DDS in
> software).  The procedure is to generate COS and SIN waveforms at the
> specified frequency, which are jusst two sinewaves in software 90 degrees
> apart.   Multiplying each of these by the incoming signal gives two
> products, I and Q components respectively, these are usually then passed
> through a  (software) low pass filter.  When displayed on a graph such that
> the I compont is on the horizontal axis and the Q component on the vertical
> a good representation of the signal in vector space is seen.  Such a display
> is usually called a Vectorscope.
>
> A clean tone with no noise appears as a dot on the screen, whose distance
> from the origin is the amplitude.  A frequency error between the tone and
> internal reference causes the dot to rotate around a circle at the
> difference frequency - anticlockwise for a frequency too low and clockwise
> for too high.  So this is a very good way of measuring exact frequency.
> Noise appears as a amorphous circle around the dot.   The instantaneous
> phase (relative to the internal reference, Roger) of a signal can be
> determined immediately just by its position on the vectorscope
>
> Soundcards are usually too unstable for this - their internal clock sources
> are notoriously drifty and innacurate.  However, for those who understand
> Windoze programming (All Hail !) it is quite a simple piece of software to
> write.  It needs a user defined NCO which can be tuned in very fine steps -
> 0.01 Hz error only takes 100 seconds to complete a whole revolution.
> Amplitude needs to be controlled - you can have a log/dB display of
> amplitude versus radius. An alternative would be to use the left channel as
> the signal input and the right as a user supplied reference tone.  This
> would not have the versatility of an NCO appraoch, but could ensure absolute
> freqeuncy stability of the display.
>
> A more advanced and better solution would be to use a reference input tone
> to continuously determine the Soundcard sampling rate, then use this
> calculated figure in setting the NCO.
>
> I use the 56002EVM module with its clock locked to a freqeuncy standard to
> do the downconversion of a signal sampled at 8kHz.  The NCO is set in steps
> of 8000 / 2^24 Hz.  The zero frequency I/Q samples are filtered and
> decimated (reduced in sampling rate) to values of 800 right down to 1.95
> samples per second.  These are output to a PC via a serial link for further
> processing.  This basic tuning / decimation / filtering routine for the EVM
> forms the workhorse of just about everything I do in the DSP line these
> days, splitting the processing between two boxes really makes for easy
> writing of software using a 66 MHz 486 running in a DOS environment !