Correlogram

Expand Messages
• Hi, I have been experimenting with the correlogram function in Spectrum Lab. The signals in a band width of 2 kHz at ~383 kHz were recorded at Sydney (top) and
Message 1 of 3 , Jul 14, 2013
Hi,

I have been experimenting with the correlogram function in Spectrum Lab.

The signals in a band width of 2 kHz at ~383 kHz were recorded at Sydney (top) and Moonah (bottom), Australia.

The recordings were started at approximately the same time, 9:30 PM.

A signal from non directional beacon WU Wanganui NZL has been passed through a band pass filter.

The process works well.

1. How is the amplitude of the correlogram adjusted?

i.e. the range shown as -0.50 to 1.50.

2. In the correlogram waterfall display where do the horizontal bars come from?
Do they represent the coinciding of the signal cycles.
The blue horizontal bar is when the signal cycles match
and the spaces when the signal cycles do not match?

3. When working with the three panes what is the preferred method to select one to change the settings?

4. Are there any functions to return the values from the correlogram so they can be plotted?

Regards, Edgar
Moonah, Tasmania.
• Hello Edgar, ... The amplitude unit is quite arbitrary. Two sinewaves with 0 dBfs and same frequency will correlate with amplitude 1.0 at all time lags where
Message 2 of 3 , Jul 16, 2013
Hello Edgar,

you wrote::
Hi,

I have been experimenting with the correlogram function in Spectrum Lab.

The signals in a band width of 2 kHz at ~383 kHz were recorded at Sydney (top) and Moonah (bottom), Australia.

The recordings were started at approximately the same time, 9:30 PM.

A signal from non directional beacon WU Wanganui NZL has been passed through a band pass filter.

The process works well.

1. How is the amplitude of the correlogram adjusted?

i.e. the range shown as -0.50 to 1.50.
The amplitude unit is quite arbitrary. Two sinewaves with 0 dBfs and same frequency will correlate with amplitude 1.0 at all time lags where they coincide. The unit is linear. I didn't care much about calibration when I tried the FFT-based correlator algorithm; so the amplitude units may be wrong. Consider them to be 'quite arbitrary' :)

2. In the correlogram waterfall display where do the horizontal bars come from?
Do they represent the coinciding of the signal cycles.
The blue horizontal bar is when the signal cycles match
and the spaces when the signal cycles do not match?
Most likely, yes. Periodic signals give peaks in the correlation at time lags which are multiples of the signal period. This is one of the reason why the correlation is often useless - it works best (i.e. gives a single strong peak) when both inputs are fed with a noise-like signal (random or pseudo-random noise), delayed by a certain amount.

3. When working with the three panes what is the preferred method to select one to change the settings?

Click into the main 'frequency' scale (which in fact is a time scale for the correlogram). The sliders on the left will be selected for that part of the display then.
4. Are there any functions to return the values from the correlogram so they can be plotted?
None .. the correlogram was more or less an experiment which was rarely used (at least afaik..)
The most impressive experiment which I used it for was to measure the distance between the walls in the living room (!), by filling the room with white noise from a loudspeaker, and feeding the signal from a microphone into both channels of the correlator (which effectively turned it into an autocorrelator). The peaks in the correlogram appeared at the distances between the walls, divided by the speed of sound. Oh well, the distance could have been measured much easier :o)

All the best,
Wolf .
• Hi Wolf, Thank you for the information on the correlogram. Regards, Edgar Moonah, Tasmania.
Message 3 of 3 , Jul 17, 2013
Hi Wolf,

Thank you for the information on the correlogram.

Regards, Edgar
Moonah, Tasmania.
Your message has been successfully submitted and would be delivered to recipients shortly.