[RFI] Different Ways of Looking at Data

Jim Brown jim at audiosystemsgroup.com
Wed Mar 24 17:09:08 EDT 2021


On 3/24/2021 12:21 PM, Don Kirk wrote:
> I just went back and looked at the audio on Garys recording using a real 
> benchtop oscilloscope instead of a software based audio scope.

Back in the early '80s, Richard Heyser, a brilliant engineer/physicist 
at JPL invented an instrumentation system called Time Delay 
Spectrometry. That invention, and his extensive teaching, revolutionized 
pro audio. One of his teachings was to use different ways of looking at 
systems and data from measurements. For example, FFT of time data 
transforms it to the frequency domain, giving us a different way of 
looking at a single measurement. And conversely, an inverse FFT of a 
frequency sweep gives us a broadband look at the time response. The 
inverse FFT of a very wide frequency sweep of a transmission line yields 
a high resolution TDR.

I used Heyser's "different way of looking" principal to study the effect 
of height on horizontal and vertical antennas and their relationship to 
soil conductivity to produce these tutorial app notes. My work was all 
done in NEC, and when presented at several ham clubs, some very sharp 
engineers in the audience gave me what I consider to be "peer review" 
thumbs up.

http://k9yc.com/AntennaPlanning.pdf

For this work, my "different way of looking" was to present vertical 
patterns for varying conditions (soil, mounting height) on the same 
graph, as opposed to the ARRL standard (and NEC default) of setting the 
amplitude scale to 0 dB. When you do this, the differences from one 
condition to the others stand out NOT as a launch angle, but as 
differences both in field strength at various vertical angles of 
interest and overall efficiency. And with multiple plots on the screen, 
the cursor can be moved from one curve to the next, and when on a 
"recalled" curve, the readout will show the difference between that 
curve and the reference curve.

Then another way of looking at the same data was to pick off data points 
from each curve at elevation angles of 5, 10, 15, 20, and 70 degrees, 
write them down on paper, and plot them on a new graph where the 
horizontal axis was mounting height and the vertical axis was gain in 
dBi (figs 33 and 36). These plots clearly show the value in dB for 
another 10 ft of mounting height.

One of the lessons of Fig 36, for example, is to disprove the long held 
myth that an antenna must be very low to be useful for NVIS. I used 70 
degrees for Fig 36 with the assumption that the shorter distances 
covered by 80-90 degree elevation angles are covered by ground wave. Fig 
36 shows that 1/8 wave height is near optimum for 70 degrees, and we 
lose only about 1 dB at 3/8 wave height.

Getting back to your measurement issue -- several things affect some of 
the measurement methods used by that computer-based system, one of which 
is resolution bandwidth, which in turn is related to sampling frequency 
and sweep rate of the measurement system. That's a possible reason that 
it didn't see the 60 Hz component clearly, where your traditional scope did.

There's a principle of measurement that, because time and frequency are 
the inverse of each other, our choice of measurement parameters to favor 
time detail proportionally reduces frequency detail, and vice-versa. A 
traditional scope is frequency blind, while a sweep is time blind.

Another technique that greatly reduces the strength of random frequency 
data in swept measurements is averaging a lot of sweeps. I use the 
highest setting for averaging in my P3 to bring out weak signals.

Around 1986, I used TDS to embed 64 sweeps in music program material at 
a level so low you had to know it was there to hear it to measure the 
dynamic response of a then-popular FM modulator that did extensive audio 
signal processing to maximize loudness. http://k9yc.com/AESPaper-TDS.pdf 
Software that Heyser had written allowed either scalar (amplitude only) 
or complex (magnitude and phase) average of up to 64 sweeps, which 
averaged out the music and display the amplitude and phase response. 
Scalar averaging yields 3 dB/doubling, complex yields 6 dB/doubling.

Techron, an industrial division of Crown (the power amp company then 
based in Elkhart, IN) built a dedicated TDS instrument revolutionized 
pro audio. I was one of the earliest owners (it cost $12K, delaying the 
down payment for my first house by five years), and what I learned from 
it (and from Dick Heyser) made my professional career. Gerald Stanley, 
who brought the design of Crown's first power amp along with him as a 
summer intern, designed that analyzer. Up to then they only made tape 
recorders.

73, Jim K9YC



More information about the RFI mailing list