MEASUREMENT OF THE DYNAMIC TRANSFER 2410 (E-3)

CHARACTERISTICS OF MULTIBAND SIGNAL

PROCESSING SYSTEMS BY TIME DELAY SPECTROMETRY

James W. Brown
Sound Engineering Associates
Chicago, Illinois

Presented at , u DIO
the 81st Convention
1986November 12-16
Los Angeles, California ®
Thispreprint has been reproduced from the author's advance
manuscript, without editing, correct/ohs or cons/derationby
the Review Board. The AES takes no responsibilityfor the
contents.

Additional preprints may be obtained by sending request

and remittance to the Audio EngineeringSociety, 60 East
42nd Street, New York, New York 10165USA.

Al/rights reserved. Reproduction of thispreprint, or any

portion, thereof, is not permitted without directpermission
from the Journal of the Audio EngineeringSociety.

AN AUDIO ENGINEERING SOCIETY PREPRINT

MEASIREMENT OF THE DYNAMIC TRANSFER CHARACTERISTICS OF MULTIBAND SIGNAL

PROCESSING SYSTEMS BY TIME DELAY SPECTROMETRY

by James W. Brown
SOUND ENGINEERING ASSOCIATES
936 W. Montana
Chicago, Ill. 60614

ABSTRACT
Multiband broadcast signal processing systems have defied measurement of
their dynamic transfer characteristics under the heavy processing ratios
commonly in use. The high noise immunity of Time Delay Spectrometry and
the averaging capability of a microprocessor based TDS analyzer are
combined the investigate the dynamic complex transfer characteristics of
several popular processing systems. Measurements will be demonstrated with
near inaudibility of the test signal.

BROADCAST SIGNAL PROCESSING

Broadcast signal processing systems are a special combination of audio and
radio frequency signal processing. It is common for the audio signal
feeding the transmitter to be split into as many as six frequency bands,
compressed, peak limited, and even expanded within those bands,
recombined, and sent to the transmitter. An additional stage of peak
limiting is often incorporated within the modulation system. FM systems
(including television audio) use pre-emphasis. For these systems, this
final peak limiter is applied to pre-emphasized audio (or with
pre-emphasis in the side chain) to prevent overmodulation. With amplitude
modulated transmitters, asymmetrical peak limiting is often employed.

The complexity of these systems has always raised questions about their
static and dynamic characteristics. Since these signal processing systems
are, by their very nature, changing their transfer characteristics with
the signal applied, use of audio test signals alone as excitation will not
yield their response to a dynamic signal (i.e., program material). If,
however, a test signal could be mixed with program material such that the
test signal is small in comparison to the program material, the processing
system will be responding to the program material rather than to the test
signal.

Specialized broadcast signal processin_ systems thus fall into two

distinct groups. The first consists of devices inserted directly into the
audio chain between studio output and transmitter input. The signal
appears as audio at both the input and output of this type of processor.
Figures 1 a and 1 b are examples of this group of processor. The second
group of processors are built into the broadcast station's stereo
generator. Because the two functions are combined into a single unit,
signal processing may occur both before and after generation of the stereo
signal. Both single band and multiband processing are used. Signal
processing after the stereo generation stage has certain advantages, and
this approach has come into widespread use. The input signal to this type
of processor is discrete left/right audio, but the output consists of the
matrixed and modulated main channel L+R and stereo L-R sub-carrier used
for FM broadcasting. The audio output also is pre-emphasized. Figure 2
illustrates this type of processor.

SPECTRAL ENERGY DISTRIBUTION FOR BROADCAST AUDIO

Broadcast audio consists primarily of two basic kinds of sources. The
first consists of the music and commercials produced outside the radio
facility. This material is normally compressed and peak limited with the
pre-emphasis and dynamic range limitations of the phonograph record or
analog tape recorder in mind before reaching the broadcast facility.
Although its high frequency headroom is considerably greater, material
released on compact disc is often processed in about the same way. The
spectral content and dynamic characteristics of all of these materials are
generally well controlled for the nature of the program material.

The second kind consists of live announcers and field audio from sources
such as news and sports. In some broadcast facilities (primarily the
larger ones) this material is also processed, but not usually as well as
the commercially recorded music and commercials. In many broadcast
facilities it is not processed at all prior to reaching the main
processor; its spectrum is weighted toward the midrange and has more
dynamic range.

The spectrum of the first type of program material at the input to the
broadcast processor approximates pink noise for the frequency range
between 50 Hz and 1.5 KHz, and pink noise with an additional 3 dB/octave
rolloff for the range between 1.5 KHz and 15 KHz. Since pink noise can be
thought of as white noise with a 3 dB/octave rolloff, the 1 KHz-15 KHz
reszon may also be seen as rolled off 6 dB/octave as compared to white
noise. Figure 3 is the spectrum of this material as it would be indicated
on a constant percentage bandwidth real time analyzer.

At the output of the broadcast processor, the spectrum is essentially the

same as at the input except that the previously unprocessed audio looks
more like the commercially recorded material and the peak to average ratio
is reduced by several dB.

TIME DELAY SPECTROMETRY AND NOISE IMMUNITY

Time Delay Spectrometry has been thoroughly documented by its inventor,
Richard C. Heyser, and others. [1, 2, 3, 4, 5] It is a powerful method
for the analysis of two-port networks, and one of its major advantages is
its noise immunity. Another is that it is valid for non-linear systems.
The noise immunity is a function of the signal bandwidth, measurement
bandwidth, and the length of time of the measurement.

The immunity of TDS energy-frequency measurements to white noise is given

by
S/N = 10 LoglO[(Program Bandwidth) / (Measurement Bandwidth)] (1)
For critical bandwidth, (maximum frequency resolution for a given sweep
rate) the measurement bandwidth is equal to the square root of the sweep
rate and (1) may be expressed as
S/N = 10 Logl0[(Measurement Time) X (Program Bandwidth)] (2)

-2-
If the interfering signal were pink noise, band-limited at 50 Hz and
15,000 Hz, the above expression would represent the signal to noise ratio
at mid band (866 Hz) and the immunity at any frequency would be improved
by 3 dB/octave of increasing frequency. For the spectral distribution of
program audio, the mid-band point is approximately 500 Hz. The signal to
noise degrades by 3 dB/octave of decreasing frequency down to 50 Hz,
improves by 3 dB/octave of increasing frequency to about 1 KHz, and
improves by 6 dB/octave above 1 KHz.

THE TECHRON TDS SWEEP

The Techron implementation of TDS [4, 6] employs an especially useful
variation of the linear sweep. To eliminate the perturbations in response
which would otherwise be obtained by sweeping through or near zero
frequency, two sweeps (one a cosine sweep, the other a sine) through zero
frequency may be used. The negative frequency portion of the sweeps
extends to only one quarter of the frequency of the positive going sweep.
A sweep to provide analysis between zero and 1,000 Hz would thus sweep the
range from -250 Hz to 1,000 Hz twice in succession. An added benefit of
the technique for this application is that the measurement time in the low
frequency .portion of the spectrum is doubled, improving the signal to
noise ratio by 3 dB for those low frequencies.

AVERAGING AND NOISE IMMUNITY

This noise immunity may be further improved by averaging. If the signal
delay through the medium is constant, complex (i.e. magnitude and phase)
averaging may be used, and the improvement is given by
S/N -- 10 Logl0(Nr of sweeps) (3)
The averaging may also be thought of as simply increasing the measurement
time, and that new time simply substituted into the corresponding
expression (2). This lengthening of the measurement time required for
averaging and the desired frequency resolution for the measurement set
some practical restrictions on the sweep times and filter bandwidths which
will be used in the analyzer. The author has found bandwidths of about
one-third critical bandwidth to be good compromises for practical
measurements. Figure 4 shows the signal to noise ratio (noise immunity) to
be expected from a TDS measurement using a complex average of 64 sweeps to
15,000 Hz at a sweep rate of 2,000 Hz per second.

If the travel time through the system changes with successive sweeps,
complex averaging will lose non-correlated data and scalar averaging must
be used. In that instance, the improvement in noise immunity due to
averaging will be approximately one half of that provided by complex
averaging.

THE TEST METHOD

Time delay spectrometry thus allows recovery of a test signal in the
presence of a considerable amount of noise. Currently available
implementations of TDS systems are based on sine waves swept at a linear
rate through the portion of the audio spectrum to be investigated. The TDS
sweep signal is mixed with program audio at a level well below program
peaks, so that the test signal is a negligible part of the input to the
processzng system.

-3-
Since the signal processing system is changing its transfer characteristic
(gain) on an instantaneous basis, the amplitude response to this sweep
will vary with time. The measured response will thus be a function both of
the program applied at the instant the sweep is at each frequency and the
overall response of the system. If many measurements were made of the
system, each would be different.

If.many sweeps could be averaged, the instantaneous variations in system

gain due to applied program material would be averaged out and the average
response of the system obtained. The Techron implementation of TDS,
developed by Gerald Stanley and using software created by Heyser, allows
this type of measurement. The Techron system was also well documented at
the time of its introduction.

The software allows averaging of either scalar or complex data using four,
sixteen, or sixty four sweeps. If complex averaging is used, the averaged
data may be stored on disk and printed to hardcopy. Scalar averaged data
may not currently be written to disk and may not be used at the nearly
inaudible levels of which complex averaging is capable. Both types of data
may be differenced to other data for measurement of gain reduction or
comparison of different devices.

With practical values of sweep rates and bandwidths, noise immunity on the
order of 36 dB is available at midfrequencies, improved by 18 dB with the
complex averaging of 64 sweeps. If the program audio has a form factor of
10 dB, the overall noise rejection would allow mid-frequency sweeps at
nearly 60 dB below 100% modulation. By 100 Hz, however, the noise immunity
for the same level of test signal has degraded by 10 dB.

With highly compressed pop music formats, good data may be obtained with
the sweep 60-70 dB below program peaks. At these levels the sweep will be
quite well masked by program material except during periods of silence
("dead air"). For all practical purposes, dead air does not exist in such
formats, and measurements can be made on the air with the audience unaware
of the sweep. With less compressed formats, the high frequency portion of
the sweep (above about 2 KHz) must be further reduced in level to make it
inaudible. Measurement in less compressed formats and those with more dead
air require that the test signal be at a lower level.

PRE-EMPHASIZED MEASUREMENT
Some means is therefore needed to improve the noise immunity at low
frequencies. Several approaches could be taken. A pre-emphasized test
signal could be used, with the sweep having a 3 dB/octave rolloff applied
to it prior to its being mixed with program material, and a corresponding
3 dB/octave de-emphasis applied to the processed signal prior to analysis.
The Techron system contazns such a rolloff filter which may be switched
into the test oscillator signal path, although current software does not
support it.

For the spectral energy distribution of broadcast audio, an alternative

pre-emphasis network might be a 75-150 microsecond 6 dB/octave low pass
applied between the generator and the device under test, with a
corresponding de-emphasis network at the input to the analyzer. An
additional 3 dB/octave low pass network for low frequencies would be

-4-
useful but more difficult to implement. It should operate between the
range of 100.Hz and 1 KHz, shifting to a 6 dB/octave high pass at 50 Hz.
Figure 5a is the desired response of the generator's network. Such a
pre-emphasized measurin_ system has not yet been implemented by the
author. Figure 5bis the signal to noise ratio to be expected from these
pre-emphasized measurement systems. The sweep signal to the simple 75
microsecond network was raised 10 dB at the input to the filter. The
signal from the combined network was assumed to be unchanged from the flat
measurement condition.

Alternatively, it might be desirable to split the spectrum into two or

more segments for analysis. Thus, data for the range from 1,000 to 15,000
Hz is obtained from one set of averaged sweeps and combined with a second
set for the range of 30 - 1,000 Hz. The experimental work done by this
author uses this method of measurement. Figure 6 shows the signal to noise
ratio of two possible setups of to analyze from 50 Hz to 1,000 Hz and
2,000 Hz respectively, using 64 sweeps of 19 seconds each.

OPERATING LEVELS
In order not to be audible in the transmitted audio, the TDS sweep at high
frequencies must be at least 70 dB below 100% modulation level. If gain
reduction of 10 dB occurs with processing, it must be 80 dB down at the
processor input. With the 150 microsecond pre-emphasis suggested above,
the sweep could be maintained at 60 dB below 100% modulation at the input
and still be expected to provide good data down to a low frequency limit
of about 150 Hz. Use of the additional 3 dB/octave low frequency
pre-emphasis could improve that low frequency limit to about 50 Hz.

Audibility is a particularly important factor, since in major cities, most

stations do not sign off for maintenance but rather maintain standby
transmitters and even antennas. Many others will do so for no more than
one night a year and there are usually many more important projects
needing the four hours available.

CHOOSING TDS PARAMETERS AND INTERPRETING THE DISPLAY

Any choice of TDS measurement parameters defines certain limitations on
the measurement. These limitations must be clearly under, stood to avoid
making incorrect interpretatiofis of the results of any measurement. As in
any swept frequency measurement, the choice of a sweep rate and
measurement bandwidth will define the frequency and time resolution of the
data. This well known relationship is gzven by
(Sweep Rate)
(Frequency
Resolution)
= (Bandwidth) (4)
and
(Bandwidth) 1
(Time Resolution)= (Sweep Rate) = (Frequency Resolution) (5)

There are several effects of these relationships which must be kept in

mind in the course of setting up for and interpreting the measurements.
First, the response of most systems will be zero at zero frequency. First,
the low frequency limit of a given measurement is one half of the
frequency resolution for that measurement, and any details of that
response which are more closely spaced that the frequency resolution will

-5-
be smoothed. Second, the time resolution may be thought of as a time
window, during which the data will be taken. Any system response which
occurs outside that window, either before or after, will be rejected by
the analysis filter, and any response within that window will appear to
have occurred at the same time.

A limitation of the Techron system, used in this work, is that the data is
stored in only 400 discrete frequency registers. Since TDS uses a linear
sweep, this means that when converted to a log display these points are
sometimes not close enough together to yield the desired resolution, even
though the above relationships might indicate otherwise. One must thus be
careful not to misinterpret a smoothness in, or lack of, a low frequency
response for a given measurement.

EXPERIMENTAL WORK
The proposed measurement method was implemented using several popular
broadcast signal processing systems as the device under test. The test
setup of Figure 7 was used for the Orban Optimod 8100, a device known to
use considerable amounts of all pass filtering in its processing. Since
the 8100 is a stereo generator which provides composite, pre-emphasized
audio at its output, a 75 microsecond de-emphasis network was needed to
provide "flat" audio for analysis. Additional measurements were performed
on other processing systems which do not perform pre-emphasis. The same
setup was used for those devices, but the 75 microsecond filter was
omitted.

The method of dividing the spectrum into two parts for analysis was used.
As is seen from the documentation included with the data, sweep rates were
such that a complete set of data (i.e.,two sets of averaged sweeps) were
obtained in about 45 minutes. The sweep signal was mixed with program at a
level that was nearly inaudible, except during periods of silence. The
program material used for the test was a compact disc reissue of a popular
contemporary recording of acoustic music having a wide variety of sound
dynamics. The systems under test were adjusted so that they were doing
relatively large amounts of signal processing, representative of their use
in competitive pop music formats. A tape recording of the mixed
combination of program material a_d test signal will be included in the
presentation. Data from thes_ sW,eeps is shown in Figures 8A and 8B.

Another set of data was taken, !uSing single sweeps for each of the two
frequency bands at the same level of test signal but without program
material. Since this level was well below the threshold of the signal
processing, these sweeps represent the steady state transfer function of
the device under test. The first sweeps, with program material, were
subtracted from those obtained without program. The resulting data are the
average dynamic gain reduction of the device, and are shown in Figures 9A
and 9B.

The phase responses were then displayed and processed to obtain the group
delay, with and without processing. The resulting phase data are shown in
Figures 10A and 10B, and in differenced form in Figuresf llA and liB. The
group delay data is shown in Figures 12A and 12B. The data for processed
and unprocessed audio (i.e.,with and without program a_dio present) show
near perfect correlation, differing only in the presence of noise at the

-6-
low frequency extremes for each set of sweeps for the averaged data. Note
that the high frequency sweep begins to get noisy below about 2 Khz, but
that it merges very well with the corresponding data taken below 2 Khz.
An analysis of the low frequency data for the Optimod seemed to indicate a
surprising loss of low frequency response below about 100 Hz. Further
analysis of the group delay data showed, however, that the unit's all pass
filter characteristic was producing so much delay that the frequencies
below 100 Hz were present at normal amplitude, but were outside the time
window of the TDS analysis filter. A new set of sweeps were made at the
time offset indicated by the group delay measurements for low frequency
information, and confirming data having good noise immunity obtained. The
results of these new low frequency sweeps is shown as Figures 13A and 13B,
which were a measurement of a single sweep only.

Another set of measurements was made with the same test setup but with the
input level to the Optimod 8100 adjusted to provide considerably more gain
reduction. Figure 14A shows the averaged amplitude response with and
without program. Figure 14B is the gain reduction. Since this measurement
extended to 20 KHz, the stereo generator's 15 KHz low pass filter is
clearly visible in the response. It did not show up in the previous series
of measurements because that sweep extended to only 15 KHz. The program
material used for both this measurement and for the previous series was
the same.

Figure 15A shows the low frequency response of a typical compressor set
for an indicated 6 dB of gain reduction. Figure 15B is its phase response,
both with and without signal. Note that there is no significant difference
between the two curves.

Figure 16 shows the gain reduction of a popular processor which combines

the functions of compression, limiting, and AGC in a single unit. It was
adjusted to provide an indicated 10 dB of gain reduction for the
measurement. The same program material was also used for this measurement.

AUDIBILITY OF THE TDS SWEEP SIGNAL

A listening test was set up by the author wherein program audio (a
contemporary music FM broadcast station) was adjusted to provide 95 dB
program peaks at the listening position. The TDS sweep signal was then
added to the listening room and increased in level until audible. That
level was then measured in the absence of program. At 30 dB SPL, the sweep
was clearly audible at high frequencies (above about 5 KHz) in periods of
moderate modulation density. During periods of low modulation density
(disc jockey speech) it was clearly audible at frequencies as low as 800
Hz. At no time was it audible at the lower frequencies. This result
supports the feasibility of a pre-emphasized TDS sweep.

The program material used as input for the device under test was analyzed
for energy content using the averaging function of the Techron TEF
analyzer. It is believed to be representative of typical popular music in
its spectral content. [7]

Spectral distributions were also obtained by monitoring the program

material with a one-third octave real time analyzer. Three types of
measurements were made. First, the RMS value of program peaks (actually,

-7-
fast responding RMS values) for a three minute segment of highly dynamic
music were accumulated in memory. Second, an accumulation of long-term
integrated levels was performed. Third, the long-term averaged levels were
observed for the same period and the general shape of the curve noted. A
plot was made of the data, and the second set of data subtracted from the
first. This differenced plot shows a relatively constant peak to average
ratio of 10 dB below I KHz, decreasing to 6 dB between 1 KHz and 10 KHz.
It is likely that this additional high frequency amplitude compression is
due to the pre-emphasized peak limiting used to provide for recording
pre-emphasis. All three sets of data from the real time analyzer show the
same trends as that obtained from the FFT.

SUMMARY AND CONCLUSIONS

Time delay spectrometry is a very useful tool for the evaluation of signal
processing systems under dynamic operating conditions. Good data may be
obtained quickly if the audibility of the sweep can be tolerated (as when
the test is run in the laboratory) and more slowly if the sweep must not
be audible (as in a broadcast station which is on the air). Pre-emphasized
measuring techniques provide the most efficient method of obtaining data.
The ability of the system to reject interference to the measurement from
program material is up to 70 dB, depending upon the length of time allowed
to obtain data.

ACKNOWLEDGEMENTS
The author woOlite-to thank Richard Heyser for his encouragement and
for the software used to accomplish these results; and Harry Priester of
the American Broadcasting Company, and John Freberg of WXRT for the loan
of equipment for measurement.

REFERENCES
1) Heyser, Richard C., "Acoustical Measurements by Time Delay
Spectrometry", Journal of the Audio Engineering Society, October, 1967

2) Heyser, Richard C., "Determination of Loudspeaker Arrival Times, Parts

I, II, and III", Journal of the Audio Engineering Society, October,
November, and December, 1971

3) Heyser, Richard C., "Concepts in the Frequency and Time Domain Response
of Loudspeakers", Proceedings of the IREE (Australia), March, 1976

4) Stanley, Gerald R., "A Microprocessor Based TEF Analyzer", Paper

presented at 72nd Convention of Audio Engineering Society, October, 1982

5) Biering, H. and Pedersen, O. Z. , "System Analysis and Time Delay

Spectrometry, Parts I and II", Technical Review No. 1, No. 2, 1983, Bruel
& Kjaer Instruments, Marlboro, Mass.

6) Bartlett, Bruce, "Testing Microphones With Time Delay Spectrometry",

Paper presented at 76th Convention of Audio Engineering Society, October,
1984

7) Quincy Jones, "The Best", A. & M. Records Compact Disc CD-3200,

Recorded 1974, A & M Records, Inc., Hollywood, CA.

-8-
 L COMPRESSOR/ L
LIMITER
Input Output

FromStudio ToTransmitter

I LIMITER Output
nRu_
t COMPRESSOR/ R

FIGURE iA

COMPRESSOR/

_=!_o_so_, _sso_J_ _o_ _o

Studio I ] LIMITER I NETWORK Transmitter

ICoMpRESSOR/
/
i LDIITER

FIGURE lB
 I
T, I PROCESSOR/ To Transmitter

R Baseband
From Studio I STEREO GENERATOR Main Channel (L+R)
38KHz Sub Carrier (L-R)

FIGURE 2
 i__i:_-_
_!i:,_
!iiii::!:!
.'.:_!'
.:_,..
........
.........
_..":
,::_ :,-_
_:_.:,,,_.::
_.,_:_'_
.......... _,:
±_ .... !_'-_: _!!!_!-t!i_!-!Tr::'l:_7'iT: 8
g
_'-**_._- _I i T_ ',.'-z_.
_r'<:-_:;_:_i__ _'.--_;-_
_:._i _-_i'.-_ .."L_i-_i:!'!:iF_!'! !i_ .! I. '
-r_-_ -'X'_!_L_::!:_:
''X:L''X:__¥.::-_ :_X?._X._LL:iT:!:
;.:T;!::! ;--F_:_'::-.: :!.. .
_.*_: _XSZ-_X_ _'. ,*_%',,_i_T
_', :_F- 2:: :_._ _H.', ::' :_;7::_:_ ';*- :;-:*.':' -::: ::.*_ :::7: _** ;_¥: :.:. ::: i '
iF_"_ S-cS"__'_-_S :_::i:;.:_ :.:_:':::;: .-::x .'z: ._;_ ::._ :!:: ::::x ::.: :':: :.:.: _:_: :::: _;:: ' '_ - 0'5
_ X:i.ii_i;_._ i;_-_/..;-_
;_i:;.i_il
__;;i _ iii_ i--: :.::
:_i:.ii_'_ .'X:_:
i-i:ii!i_i -:;_-z_::::_
i:_L-"ii:_ ::_:x
_;ii ,::i_[-
:z_c-- _:, 7';-'__-_z-*t_x_ _-: :_: _:::, z::z ::_:. :;:: :' x_z_*
;:::-:, ;_ 'z_::: ..... _4.
:::' '::: ::::::::::::'_ ::.': ':ii::::: :::. :;::::::' :c:;:::.:::::: "-.I ! r.'.,:-,
_:::..::'_..;].iu::x ::::::x:-:.:::::-::::::._:':::'z.:::-__x:
::x::-::::-:x;-:_..:::1:. ,.8
-- ;x::'.'i:'_E:::::-':, :: 7:.: ::.... :: :x: :::: ::X :!:: !:i:: ::-:: '::: :::: ;::: :::7 ::" ::"i" ¢',
, .,,;! , --\:::' .-1- _ ,.-,
:.F_::_=_ F=z=:_ _= ........ _...s= .............. 7 ...... _....... _'- -_: ...... r....... t ........ ,
':;: _L::_i-S_ X-S-i-:i:_!:; ::_:i:! i:i_.hi.-;: ::' :<_'_:_-_: :_:_i _.i__? ::i:::.i : _ z
-/::: &%T%L[:%":
:;_-_- %ZR.L'
[* ; :7'21 -'£:tZ':L';C':? - :t:-- ':*; %X:.Z:. ; ' : ;TI'i;. ::Z:,---. -- , i 4-1 14_
':!!'!;-7-::!fF::i
;; :FL';::;:::::[
...."! :.i..:
........
.\' j ::.:I. . I w
:. _::_x:::: ..:_.:':'I-' :-: : -:i .... -:: :-[:. : !: "_.: '' t: - - !- , _-
.L:..::---:-:
.... 1 "' : -'.:.. r '. -:'1' ":'":-""t'\:':
i::.:: _: i :
:"_.........
'":::i' ::i ': - "-...... :_:i' :'' :::' '
:i":,.:ii
::- : ',,!: :::. :
':. !-'-! 7.--:[:IX:_..:
?' -i'- ' :::'!X& :;___::_L
:L_..i_ " "_'::','
:::"4---:-'!i':'
'"!:'::i-'' :::-!:
'" ..... -'i'I ._ : ;-L;
I : I' '1'' '.':_ _-_ .........
-o
 '*- - 77
i!_i!.nz: ill _ "_, _' _1" 'u .... ......... __, ',_ '
.......... _,,?_................ :.... ., ,,_.,. __._ _X__!? -_ _ii_
· · ii ,
.................
...._::_:"_
......
:::_
_,.-:_-,_-_:,_?:'_:-':__
_L'_'?_'_:,
_:_,_'
_i__ii_
'_':i_
!_'?'_'-"_*'*-_,_
_ _® _*-:,::;_
_'_:,,_:-_
_,_"_',
....
_; [:_C;:;z:;;
:71:c,_: .::
_i _ l_._ -!;_ :. :_,._ _:_!' ..it; _ };_ _ :_: _i;_! I_:, _ ....... -= _'i_ _'_4_-_C_-_',_TT .... _-_- {_-_t_ _- _ '_?
.__ _74:4: -f:_f:.:::-_Cfi: _ *_:--:;= ,i.=: ::::: _':: _ _ '-_l_,...
-'_,_ _ :::
7,;_T
"'-_ i _ ::::. ,' "i ......W",,""i"_'_:_ ''_-:_
, :: _
:'; ', ":"--ii _,,1!;i'!:-? _,; :ii_!i r?-,_tJlq!_HffrltfflfH; TTr,!?
:_-:_ '--_i_'_ Ix_7T''-_ t _i-:s:d_X _i_ ,_71 ............. i_,_ :"'_-__ :..,'_:
;¥7,_:-i:--- .--: -. : _-'-_ ;tr_: _:::
Z,,,.,,r_TP_q"_kq_,, +t+l+_ff+ _+N-'
 o
_i_!!_
..........
!_!!!___ !_-!!_!!_i_
ID , !!!!ix _ ' i ....
-i_-'L-']'i-'i£!_ L;LE;.i_i;C'r';:
i'X _7,_
;.__X_-£
1.7i';' 7'-;L'.;'
h:
:_------_ :_' =:;: -<' _f_ _!E :._ _x;.; :cT' c;:Lc_"_c_:x_z':::-;.c:;_ :=:...=:: h:-_':z_ ::
:--'_ _-- FY'i TM :=_-F:F' ::!':: :" =:-:' ::::= ::= :':_ ==_':"_- :=:-'_'x=-_: '¢
:.::..¥:.:-.
,:i.-:i,i....::_.(!i--_.::::'::7:.:_xi':7:
::':::::._:;[!:1:%i
':::' .f,_
:z:c' ___-:z:_ _'c-_:z"_r c---z :--_c::_z -_c2 '::c: :*x: ._: =cz_:*xz _-c;_._
7'-_
_. z:== __-_:.-:-'ccc.;:: :: _, _, ,_
__-$_ci
:-;c:-;:c::...c:_::-:-:'-::-c-:_;:: :- ._/.:- 1_-!
i[_F'_
_2
_!!_
1'_5:/5!ii!_ i_}ii_-_ o -_
'-_ o
 oo
f ' _.... _. , .. m -F ' ' _'_
8
" ' :' · , _- : , i' i ..... I...... r_ _'
' ': : ! I , , ' :. : ,.._ ..'_:'::': _
. , _ .Z
I I ! I ' _ \, _ ' ,
, , ! i · ' I t _ .i : ' !
: ! I" ' _L i ! '! _ "_
I .,.. _ ! ! ! I-.V\ -I .! .
 o
[-_
f
' 0"_
_._11 _
 D
C_
cD_
QJ
0
_v
U
o
c_
o
F_
0 Z
_ocO
m r.D
[.-.4
O_
&J
 Mag. vs Hz (EFC) of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85
At S. E. A. LAB

i:::! ii......
iiii
· i'!'i'])i' ' ' : · ':'' : :':'i' . .! SINGLE SWEEP ·

!!:!! ! i :!ii!!! _ · .'_

6 dB :::::: : : :/:::':_'%, : .... ::::

:::::: : _ : :::::: 64 S:_Y_PAVERAGE

'!'!'i...... i!! ...... !"ii!i'"

..... ! ' ; ; : :::::: : ; '- : ',;;;i
J t_t'3 Pr ecluen _.v

100 Hz 2 KHz
Vertical: bdB/d_v with base of display at -52.8dB
_dE: is lorated at .775 Volt

Horizontal: Auto 4.00Hz to 2008.24Hz

Log freq axis (2.7decades)

Resolution: 25 milliseconds ,x° 39.97Hz

Time of test: 571 microseconds, 4.2518E-01 Feet

Sweep Rate & Bandwidth: 199.85Hz/Sec & 5.00Hz

Input configuration: Channel 1 Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE 8A

Amplitude response of the processor at low frequencies, with and withou_

signal. The time of the measurement is the 371 microseconds found to be
the signal delay for the processor at high and medium frequencies.
 Mag. vs Hz (EFC) of OPTIMOD 81_0 #2
By JIM BROWN
On 10/3/85
At S. E. A. LAB

iiiiil i i i!iiiii i ! _ !iiill

iii!
i ii
i........
i ilii
' i'i'i'iii'"i"i'iiiiiii
.... i"i'!'i'iiiii"
''_
_" '-_-_ _-i__:',
._.._--il..

"i'i'i
.....?iii......._i"_i
_ii-
Log
Prequency
1_ 20K2_
Vertical: OdB/div with base of display at -52.SdB
OdB is located at .775 Volt

rtorlzontal: Auto 40.00Hz to 19998. 10Hz

Log freq a:,'is (2.7decades)

Resol ut ion: 10 milliseconds & 98.87Hz

Time of test: 571 microseconds, 4.2318E-01 Feet

Sweep Rate & Bandwidth: 1977.45Hz/Sec & 2_.00Hz

Input configuration: Channel 1 Balanced

with OdB of inpL(t gain & 12dB of IF gain.

FIGURE 8B

Magnitude response of the processor with signal, as measured by the 64

sweep complex average.
 Mag. vs Hz <EFC) of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85
At S. E. A. LAB

·i!!iil
...iii iii!......iiiiili
6dB .........

_-_-_' i..!.i ii'iii_....i..i.i.i,i ii.,,

!i!!i! ! i i iiii:i : : ::::::_

Log Frequency
2I_3_
Vertical: 6dB/div 1OO_
Differenced data

Horizontal: Auto 4.00Hz to 2000.24Hz

Log _req aris (2.7decades)

Resolution: 25 milliseconds & 59.97Hz

Time of test: 571 microseconds_ 4.2518E-01 Feet

Sweep Rate & Bandwidth: 199.85Hz/Sec & 5.00Hz

Input configuration: Channel i Balanced

with OdB of input gain &, 12dB of IF gain.

FIGURE 9A

The gain reduction of the processor at low frequencies. The single swee p
without signal is differenced to the 64 sweep compfex average.
 Mag. ve Hz (EFC) of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85
At S. E. A. LAB

----_.
_..: . .... _ .

6 dB

..... !, . ....... : ........ :

Log Frequency
i_ 20
*_i_
Vertical: 6dB/div
Differenced data

Horizontal: Auto 40.00Hz to 199_8.10Hz

Log freq axis (2.7decades)

Resolution: 1.1538E+01 Feet & 98.87Hz

Time of test: 571 microseconds, 4.2318E-01 Fee_

Sweep Rate & Bandwidth: 1977.45Hz/Sec & 20.00Hz

Input configuration: Channel 1 Balanced

with OdB of input gain & 12dB of IF gain,

FIGURE 9B

Gain Reduction for the processor. The single sweep without signal is
referenced to the 64-sweep complex average.
 Mag. vs Hz (RFC) of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85
At S. E. A. LAB

'. ::::i : : : : ::::i : , : : ::::i

iii:iii i:i
i :iii:iiii:
i
6 dB

Loq
Frequency
100Hz 2 kq4z
Vertical: _dB/div with base of display at -52.8dB
OdB is located at .775 Volt

Horizontal: Auto 4.00Hz to 2000.24Hz

Log freq axis (2.Tdeeades)

Resolution: 25 milliseconds & 39.97Hz

Time of test: .571 mi_rosecends_ 4.2_18E-01 Feet

Sweep Rate & Bandwidth: 199.85Hz/Sec & 5.00Hz

Input configuration: Channel I Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE 10A

Amplitude and phase response of the processor at low frequencies. The

apparent oscillation of the phase data at low frequencies is actually very
high phase shift_ plotted at multiples of the 360 degrees which is
displayed. This zs a 64 sweep compte x average, wzth program.
 Mag. vs Hz (EFC) of OPTIMOD 8100 #2
By OIM BROWN
On 10/5/85
At 8. E. A. LAB

..:.:.::i... SE , .'.,'.'.'.''":

iiii! ' 'ii'iiii

45°6
d_ b._:....i__

Loq I:requencv
1 KHz 20KHz
Vertical: 6dB/div with base o+ display at -52.8dB
OdB is located at .775 Voit

Horizontal: Auto 40._)0Hz to 19998. l_Hz

Log freq axis (2.7decades)

Resolution: 1.1538E+01 Feet & 98.87Hz

Time of test: 571 micromeconds_ 4.2.S18E-Ot Feet

Sweep Rate & Bandwidth: 1977.45Hz/Sec &. 20.0_]Hz

Input configuration: Channel 1 Balanced

with OdB o4 input gain & 12dB of IF gain.

FIGURE 10B

Magnitude and phase response of the processor with signal as measured by

the 64 sweep complex average. The apparent oscillation of the data at high
and low frequenczes is really the phase being.plotted at multiples of the
360 degrees which are displayed on axis.
 Phase vs Hz (EFC> of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85
At S. E. A. LAB

ii!i......ici![......iiiiiii
':.....
....'::-.-_
""!i'' "!"'i- ii!....... i"ili:i'"
.iiiii!.,i 'i')i['iq!
....i i !'ii!'-ii'

iiiiiiiii¢iiiiiiiiiii
'i'i'!'_i'"_'"_'ri_i .... i"?i'i'!ii?i'"

_._._._.._'_4_i;_
...._.._.i..z_
...
Log Frequency
100 Hz 2
Vertical: .45 degrees/div.
0 degrees is at the dashed horizontal line.

Differenced data

Horizontal: Autto 4.00Hz to 2000.24Hz

Log frl_q axis (2.Tdecades)

Resol u_ i on: 25 milliseconds & .79.97Hz

Time of test: 371 microseconds_ 4.2118E-0I Feet

_weep Rate & Bandwidth: 199.85Hz/Sec & 5.08Hz

Input configuration: Channel 1 Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE llA

The di'fference between the phase response of the processor with and
without signal. The complex average is differenced to the single sweep
without signal. The measurement is subject to considerable noise below 100
Hz, partially because of the signal delay at those frequencies and partly
due to .the reduced, noise i_unitv
 Phase vs Hz (EFC) of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85
At S. E. A. LAB

?il......iii.......
i iii
· ........
4..

Log Frequency
1_ 20KIts
Vertical: 45 Oegrees/dlv.
_,iOegrees is at the dashed horizontal line.

Differenced data

Horizontal: Auto 40.00Hz to 19998. 10Hz

Log freq axis (2.7decades)

Resol ut ion: 10 milliseconds & 98.87Hz

Time of test: .571 microseconds, 4.2518E-01 Feet

Sweep Rate & Bandwidth: 1977.45Hz/Sec & 20.00Hz

Input configuration: Channel i Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE lib

The difference between the phase response of the single sweep without
signal and the 64 sweep complex average.
 Group Delay of OPTIMOD 8100 #2
By JIM BROWN
On 10/.3/85
At S. E. A. I_AB

,%-:
.: ..... :::i :. iil

: : : :','.::: : : : : :::;!
............. : ........ : t

Log
Frequency
100Hz 2 KHz
Vertical: Delay in microseconds, 1562 uSec/dlv
zero delay is at khe dashed Inorizontal line.

Horizontal: Auto 4.00Hz to 2008.24Hz

Log freq axis (2.7decades)

Resol ut ion: 25 milliseconds _< _.9.97Hz

Time of tI._st: .-'-',71

microseconds_ 4.2318E-01 Feet

Sweep Rate g_ Bandwidth: 199.85Hz/Sec & 5.00Hz

Input configuration: Channel 1 Balanced

with OdB of input gain _ 12dB of IF gain.

FIGURE 12A

Group delay for the processor from the low frequency sweep. The apparent
oscillation of the data at low frequencies is really the computer plotting
that data off axis above the display. Eight divisions of delay should thus
be added to the displayed data in the range between 30 Hz and 60 Hz. Note
that the frequency resolution of this data is 40 Hz, so that data below
about 30 Hz has little significance.
 Gr'oup Delay of OF'TIMOD 8100 #2
By JIM BROWN
On 10/.5/85
At S. E..A. LAB

..... : _ _ '.'_.'._i ........ :

1.25ms ....__6.
68._.ee:p:____
7 :J_2i ......... ] J _ : [::_i
·!.i.i.ii!.,, !.-::.
........
_i i!i_'i
.... _.,_.-._,i-':_i
! ........ !
·-

Log I-'requenc y
1FJ{z 20KHz
Vertical: Delay in microseconds, 1250 uSec/div
zero delay is at the dashed Inorizontal line.

Horizontal: AuEo 40.00Hz to 19998. 10Hz

Log freq axis (2.Tdecades)

Resol ut ion: i0 milliseconds & 98.87Hz

Time of test: 371 microseconds, 4.2318E-01 Fee__

Sweep Rate & Bandwidth: 1977.45Hz/Sec _ 20.00Hz

Input c-onfiguration: Clnannel 1 Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE lib

Group delay for the processor from the high frequency sweep. The time
reference for this data is the 371 microseconds found to be the signal
delay through the device at high and medium audio frequencies.
 Mag. vs Hz (EFC) of OPTIMOD 8188 #2
By JIM BROWN
On 1815/85
At S. E. A. LAB

· :.:.:.:;i. . . '. .:, ' : i':21 .... '., .:. '.:.:2 '_J . .

: 1:11{ : : : : ::'.:1 : : : : :1:::

Frequency Log
100Hz 2 KHz
Vertical: 6dB/div with base of display at -52.8dB
OdB is located at .775 Volt

Horizontal: Auto 4.08Hz to 2000.24Hz

Log freq axis (2.7decades)

Resolution: 25 milliseconds & _,9.97Hz

Time of test: 8371 microseconds, 9.5492E+80 Fee:

Sweep Rate & Bandwidth: 199.85Hz/Sec & 5.88Hz

Input configuration: Channel 1 Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE 13A

Amplitude response of the processor with the time of measurement changed

to include the delayed low frequency re?ponse. Note that the
mid-frequencies are starting to move out of the tune window. This data is
from a single sweep with no signal.
 Group Delay of OPTIMOD 8100 #2
By JIM BROWN
On 10/5/85 PHASE
At S. E. A. LAB

12.5 ms

GROUP DELAY

LogFrequency,
100 Hz 2 KHz
Vertical: Delay in microseconds, % 12499 uSec/div
zero delay is at the dashed horizontal line.

Horizontal: Auto 4.00Hz to 2000.24Hz

Log freq axis (2.7decades)

Resolution: 25 mill'iseconds 59.97Hz

Time of test: 8571 microsecondsg _.5492E+00 Feet

Sweep Rate & Bandwidth: 199.85Hz/Sec & 5.00Hz

InpLIt _onfiguration: Channel I Balanced

with OdB o_ input gain & 12dB of IF gain.

FIGURE 13B

The phase response, and its negative first derivative, the group delay for
the measurement of Figure 13A. Note that this measurement shows the low
frequencies are delayed considerably more than the 8 msec time chosen for
the measurement, so that some low frequency data is still falling outside
the analysis time window. The amplitude response at those low frequencies
is thus higher than indicated, but with a significant time offset from the
major part of the program audio.
 Amplitude ETC of OPTIMOD 8100 #2
By JIM BROWN
On 1015/85
At S. E. A. LAB

6 d_ .....................................

¥'._...i...i...!...!...
.. i...i...

Time
10.5
ms 52.5
ms
Vertical: 6dB/div with base of display at -55.8dB
OdB is located at .775 Volt

Horizontal: 0 microseconds or 0 Feet to

526.58 microseconds or 80.0.'581 Feet
scale: 1.6415E+01 Feet/inch or 6.4626E+00 Feet/cm.
14.'589microseconds/inch or 5665 microseconds/cm.

Line Spacing: 151.905 microseconds or .150471 Feet

Line Width: 179.591 microseconds or .204641 Feet

Sweep rate: 99.92Hz/Sec

Sweep range: 50.5.5Hz to 997.98Nz

Window file name: A:HAMMING. WST

Input configuration: Channel I Balanced

with 24dB of input gain & 15dB of IF gain.

FIGURE 13C

Energy Time Curves for the processor taken without program material, but at the
same level as during program tests. Note that the sweep which covers only the
region above 200 Hz is more compact, indicating the lower frequency energy is
arriving later. Since the TDS sweep is linear, the Energy-Time Curve emphasizes
the higher frequency octaves. This measurement confirms the findings made using
the data of curves 13A, 13B, 12A, and 12B.
 OPTIMOD _2
By OIM BROWN
On 10/8/85
At MADISON BROADCAST CLINIC

i i_!!i i ! i i!iiii i i _ i!iii!

iiilii .........
i ! iiiiii! i i _!!iii!

.... i ..................
""'",'i' ' ' '' %'7 ] ]?]:_ .... 1' '2 '.'iq]ii' ' ' '

_!}il i i i?)iill _ i _ii_':i!

iiliii i i iiiiiii i iiiiiii

Coq
I:requency
1 KHz 20 kq-lz
Vertical: 6dB/div with base of display at -58.8dB
OdB is located at .775 Volt

Horzzontal: Auto 40.00Hz to 19998. 18Hz

Log freq axzs (2. Tdecades)

Resolution: 10 milliseconds ._t & 98.87Hz

Time of test: 571 mzcroseconds, 4.2318E-01 Feet

Sweep Rate & Bandwidth: 1977.45Hz/Sec & 20.88Hz

Input configuration: Channel 1 Balanced

with OdS of input gain & 12dB o_ IF gain.

Remarks:

FIGURE 14A

Amplitude response with and without signal, wit_{ a much higher drive
signal to the processor. The processor is doing much higher compression.
The sweep with signal is a 64 sweep complex average.
 OPTIMOD #2
By JIM BROWN
On 1018/85
At MADISON BROADCAST CLINIC

6 dB

'i'i'!
......iii ......!'-iii-"
Lag Frequency
1 i_Iz 20 Ktlz
Vertical: 6dB/div
Differenced data

Horizontal: Auto 40.00Hz to 19998.10Hz

Log freq axis (2.7decades)

Reso 1ut ion: 10 milliseconds & 98.87Hz

Time of test: 371 microseconds, 4.2318E-01 Feet

Sweep Rate & Bandwidth: 1977.45Hz/Sec & 20.00Hz

Input configuration: Channel 1 Balanced

with OdB of input gain & 12dB of IF gain.

Remarks:

FIGURE 14B

The difference between the two sets of data of Figure 14A, which
represents the gain reduction of the processor.
 Mag. vs Hz (EFC) of AUDIO & DESIGN VOCAL STRESSOR
By JIM BROWN
On 10/19/85
At S. E. A. LAB

4 5808E+'_,_

-12 61dB
8._.!Gz_eq
Ot99E-O! AMPLITUDERE POHSE
v011

...... ¢i ;!ii-: iiii ii i:!ii/!i

i-.._-.,-?.i
..............................................
S,_.R_,:
199,85Hz/S
i i::::i::ii/ i i iiiiiii : : :::::::
14,14H,- _i" _EDiJCiTih,_[ii ! ii i::::iL

ti, ,:,:q.)i:! , · -'.-, i .i. i,i,:.:'!. · , ' · .:,::: i:.:[

Rcms,
e Lo._ _- : :/ ':_ .... ::::: : : : : :'::
99 2.._8
_'' '_ _ _ 'm_''_'''_ _ i _ S _ _ _ _ _ _"= _ _ _ _ _:_'

Log Frequency

Vertical: 6dB/div with base of display at -46.8dB

OdB is located at .775 volt

Horizontal: Auto 2.00Hz to 997.98Hz

Log freq axis (2.7decades)

Resolution: 70 milliseconds _ 14.14Hz

Time of test: 15 microseconds, 4.5000E+05 Meter

Sweep Rate & Bandwidth: lc?_.85Hz/Sec & 14.14Hz

Input configuration: Channel 1 Balanced

with OdB of input gain & 12dB of IF gain.

FIGURE 15A

The low frequency response of a typical compressor set for an indicated 6

dB of compression. This is a 64 sweep complex average. The sweep level was
set at a level which was much more audibl_ than for the Optimod, but was
well below the average p:ogram level.
 Phase vs Hz (EFC) of AUDIO & DESIGN VOCAL STRESSOR
By JIM BROWN
On 10/19/85
At 8. E. A. LAB

(D_taR,,,i_)<O_erW_i
t : : ' * '_ i :' :::::ii
' : '.'": :' :: i::::il
.... :'
ll§_Sec ......
';/!_- _ . : : . . ; :._ : . . : : ;..
· 8E _ . ,:. .:.:,::!_N'_,i . : .:.:.:.:.::i. , . '. .:. : ::::.:
Heter k !/: : :/:::i _ : : : : ::::i : : : : ;:::
'_._-_'-_.'::::! _ : : :::::i : : : :::::1
' :' ': ': ': ': !:'_ ' ' ' :_,_: ' P'H'R$'E"I_ESFORS"EZ,
'_1TH' fiND":
: : : ::.:i : '_i WITHOUTS:IGNAL SMOOTH'":.
i ! i !!!!i i ._ CURVEIS HO SIGN,L: : !ii!

Sweep Rate: : : : ::::: : : : : ::::: : : : : :!!!,

[9_,85Hz/S : : : : :;:; : ; : : : ;::; : ; : ; : ,-.
Banduidth ':' ': ':':': ::_ ' ' ': ' : ':' :' :':'::_' ' ' :' ':' : : :::':
14.14HZ :: i i ii::il : : : : ::::: : : : : ::::

Channel t '!' '!'!'!'!!i:.'''? 'i'i'!'!'i'iii' ' ''.'' 'i'! !!!!'!

_:_o,_-_,2.TD_:: iiiiii_z ! i iil._.;.
! i i!::5:_OH:

fi.... ......i ' Log Frequency

Vertical: 45 degrees/div.
0 degrees is at the dashed horizontal line.

Horizontal: Auto 2.00Hz to 997.98Hz

Log freq axis (2.7decades)

Resolution: 70 milliseconds _ 14.14Hz

Time of test: 15 microseconds, 4.5000E+05 Meter

Sweep Rate & Bandwidth: 1c_9.85Hz/Sec & 14.14Hz

Input configuration: Channel I Balanced

with OdB of input gain ._ 12dB of IF gain.

FIGURE 15B

The phase response of the compressor of Figure 15A. The two curves show
the measured phase response, with and without signal.
 Mag. vs Hz (EFC) of APHEX COMPELLOR
By JIM BROWN
On 10/19/85
At S. E. A. LAB

(Data Review) : ::::i ........ : : : : : :::::

3.6888E+03 : : : : ::: :_

........: ............. . ....................

dl
-'i-'i-':-'-!..........
' -' ":--"-"-'-"_ _?
..........
i'--:~ '+-.-_.+_'F..........
Sweep Rate: : !i!i:_ ........ :
5B09.55Hz/S ...... ' .................

BondwLdth ii:_ :

i !i! t
Channel t .!.!.i.ii::.. i.Gh, _iD_JC;r_ION..i..i. !.i.;iirl -' ]

'!'!'i!!" !' ': _ !!!'! .... !''!' :.'!'i!!'i' I

Ran_,eLc_ ..- : - · l
199.'1
_. I;)
Log Fr'ec_[.Jenc y

Verti cai: 6dB/dj v

Dif fer-_nced data

Horizontal: Auto 40.00Hz to 19998. 10Hz

Log freq axis (2.7decades>

Resolution: 2 milliseconds & 5oo. q6Hz

Time of test: 12 microseconds, 5.AOOZE+05 Metsr

Sweep Rate & Bandwidth: 5009.55Hz/Sec & 10.00Hz

Input configuration: Channel 1 Inverting

with OdB of input gain ,_. 12dB of IF gain.

FIGURE 16

Measured gain reduction at medium and high frequencies for an integrated

compressor, limiter, and AGC. The sweep signal was audible, but well below
the average program level.