application

application
note
All you need to know about
SINAD and its measurement
using 2023 signal generators

The 2023A, 2023B and 2025 can be

supplied with an optional SINAD measuring
capability. This article explains what SINAD
measurements are, when they are used and
how the SINAD option on 2023A, 2023B
and 2025 performs this important task.

[Link]
 SINAD and its measurements using the 2023

What is SINAD? C-MESSAGE filter used in North America

SINAD is a parameter which provides a quantitative Psophometric filter specified in ITU-T Recommendation
measurement of the quality of an audio signal from a O.41, more commonly known from its original description as a
communication device. For the purpose of this article the CCITT filter (also often referred to as a P53 filter)
device is a radio receiver. The definition of SINAD is very simple A third type of filter is also sometimes used which is
- its the ratio of the total signal power level (wanted Signal + unweighted (i.e. flat) over a broader bandwidth.
Noise + Distortion or SND) to unwanted signal power (Noise + The telephony filter responses are tabulated in Figure 2. The
Distortion or ND). It follows that the higher the figure the better differences in frequency response result in different SINAD
the quality of the audio signal. The ratio is expressed as a values for the same signal. The C-MES signal uses a reference
logarithmic value (in dB) from the formulae 10Log (SND/ND). frequency of 1 kHz while the CCITT filter uses a reference of
Remember that this a power ratio, not a voltage ratio, so a 800 Hz, which results in the filter having "gain" at 1 kHz. In
factor of 10 in voltage is 20 dB. performing the SINAD measurement however, the filters are
For most applications SINAD is measured by setting up effectively "normalized" to their 1 kHz value since the
conditions so that the audio output contains a nominal 1 kHz measurements are performed with a tone at this frequency and
tone. For a radio receiver this could be generated by applying SINAD is inherently a ratiometric measurement.
an FM signal with a specified deviation at 1 kHz rate to the Frequency Response Tolerance Frequency Response Tolerance
antenna. The audio output will have the 1 kHz tone present (Hz) (dB) (dB) (Hz) (dB) (dB)

plus noise and distortion products. 16.66 -85.0

60 -55.7 2 50 -63.0 2
To measure the SINAD ratio the audio output from the 100 -42.5 2 100 -41.0 2
receiver is measured (wanted signal + noise + distortion) and 200 -25.1 2 200 -21.0 2
300 -16.3 2 300 -10.6 1
then passed through a notch filter which removes the 1 kHz 400 -11.2 1 400 -6.3 1
tone as shown in Figure 1. The resulting signal is measured 500 -7.7 1 500 -3.6 1
600 -5.0 1 600 -2.0 11
(noise + distortion) and compared with the first measurement. 700 -2.8 1 700 -0.9 1
The ratio is the SINAD value. 800 -1.3 1 800 0 0 (reference)
900 -0.3 1 900 +0.6 1
1000 0.0 0 (reference) 1000 +1.0 1
1200 -0.4 1 1200 0.0 1
1300 -0.7 1 1400 -0.9 1
1500 -1.2 1 1600 -1.7 1
1800 -1.3 1 1800 -2.4 1
2000 -1.1 1 2000 -3.0 1
2500 -1.1 1 2500 -4.2 1
2800 -2.0 1
3000 -3.0 1 3000 -5.6 1
3300 -5.1 2
3500 -7.1 2 3500 -8.5 2
4000 -14.6 3 4000 -15.0 3
4500 -22.3 3 4500 -25.0 3
5000 -28.7 3 5000 -36.0 3
6000 -43.0

Figure 2 - C-MES Response CCITT Response

Figure 1.- SINAD measurement system

The measurement can be made either from an audio output
or via an acoustic coupler from the loudspeaker.
For most radio systems the important SINAD numbers are
usually between 12 dB and 20 dB (the threshold for
reasonable intelligibility of voice) or the "ultimate" SINAD under
ideal reception conditions, which is typically in excess of 40 dB.
The bandwidth of the audio signal is limited by a filter, which
has the effect of limiting the amount of noise present and
reflecting the ability of users to extract speech information from
an audio signal. The audio filter restricts the bandwidth of the
signal that is measured, concentrating the measurement on
the important part of the voice around 1 kHz. For radio Figure 3 - Comparison of C-MES and CCITT Frequency Response
receivers the filters specified are usually based on those used When used the filters clearly restrict the effects of harmonics
in the fixed analogue telephony systems but, because of on the SINAD measurement since only the second and third
international differences in telephony systems two filters are harmonics pass through the filters and contribute significantly
commonly quoted. to the measurement.

2 SINAD [Link]
 The ITU-T Recommendation O.41 specifies that the the RF signal level that is applied to its antenna input to obtain
measurements in CCITT filters should be performed with an a specified value of SINAD. Values of SINAD are typically 12 dB
RMS (root mean square) detector (i.e. they are power or 20 dB and a 1 kHz modulating tone is used with a specified
measurements) and require that any alternative detectors used modulation depth (AM) or deviation (FM or F M). For FM
provide the same answer when measuring Gaussian noise analogue systems ETSI typically specifies the use of a deviation
signals, sine waves and any periodic signals having a peak to of 12.5% of the channel spacing.
RMS ratio of less than 8 dB. Adjacent channel rejection
The notch filter used to remove the 1 kHz tone must be deep This parameter is an indication of the ability of the receiver
enough to ensure the tone is not passed onto the noise and to adequately demodulate the wanted signal when there is a
distortion detector, but narrow enough to ensure that distortion strong interfering signal in the adjacent RF channel. In this
and noise products are passed on. measurement the reference sensitivity is typically measured,
the RF level is then raised by 3dB and an interferer (a second
source) is added in the adjacent channel which has a
modulation tone at 400 Hz present. The interferer signal level
is raised until the SINAD value is degraded to the original value
obtained at the reference sensitivity. The ratio of the interfering
level to the wanted signal is the adjacent channel rejection. The
degradation in SINAD is typically caused by the 400 Hz tone
becoming apparent in the audio signal which is passed by both
the weighting filter and the notch filter.
Receiver intermodulation
This measures the ability of the receiver to reject two high
level interference signals whose frequencies are chosen such
that the third order intermodulation product falls at the receiver
Figure 4 - Limits for Notch Filter ITU-T 0.132
input frequency. Again it is found by finding the reference
sensitivity, raising the wanted signal by 3 dB and then adding
ITU-T Recommendation O.132 describes a filter used with
two more interfering signals of equal amplitude (one of which
1.02 kHz tones for distortion measurements on telephone
is typically modulated at 400 Hz rate) until the SINAD degrades
systems which requires the notch to be a minimum of 50 dB
to the original value. Intermodulation causes a co-channel
deep from 1 kHz to 1.025 kHz. For 1 kHz SINAD
interference to occur which again generates a 400 Hz
measurements the filter has to be adjusted for a 1 kHz centre
interference tone. The receiver intermodulation is then the ratio
frequency.
of the interfering signal levels to the wanted signal level.
The European Telecommunications Standards Institute
Receiver blocking
(ETSI) defines a notch filter in ETR 027 with the following
This is the ability of the receiver to reject a strong interfering
characteristics for use in SINAD measurements:
signal far from the wanted frequency. It is measured again by
2.2.3 Band-stop filter (for the SINAD meter)
The characteristics of the band-stop filter used in the audio distortion factor meter finding the reference sensitivity, raising the signal level by 3 dB
and SINAD meter shall be such that at the output the 1000 Hz tone will be
attenuated by at least 40 dB and at 2000 Hz the attenuation will not exceed 0.6 and then adding a CW interfering signal whose level is raised
dB. The filter characteristic shall be flat within 0.6 dB over the ranges 20 Hz to
500 Hz and 2000 Hz to 4000 Hz. In the absence of modulation the filter shall not
until the SINAD again degrades to the original value. An
cause more than 1 dB attenuation of the total noise power of the audio frequency extensive search using different frequencies has to be
output of the receiver under test.
performed to check for image and non linear behaviour in the
The notch in ETR 027 is not as deep as that defined in the receiver.
ITU Recommendation O.132. If the ultimate SINAD (i.e. under It can be seen that SINAD is used as a basic measurement
good conditions with a high input signal level) is to be technique for many of the important (and fundamental)
measured a notch deeper than 40 dB is required. receiver parameters. Its application is mainly relevant to
WHEN ARE SINAD MEASUREMENTS USED? analogue systems, but digital systems can also be measured in
Since SINAD measurements provide a quantitative this way if the voice codec allows the generation of suitable test
measurement of audio quality from a receiver they can be used tones. The SINAD measurements are particularly useful for dual
to assess the impact of different conditions on the radio mode radios in North America supporting both digital systems
performance. The principal measurements which require the and analogue AMPS systems.
use SINAD measurements are: Measuring SINAD with 2023 series Option 12
Reference sensitivity To simplify SINAD measurements the 2023A, 2023B and
The sensitivity of the radio to an RF input is measured to be 2025 Signal Generators from IFR are offered with an optional
built in SINAD meter. The RF output of the signal generator can

3
 SINAD and its measurements using the 2023

be used stimulate the receiver while the SINAD meter is used

to measure the audio output.
The SINAD meter is implemented by an analogue to digital
converter which converts the audio signal to a digital format
and a digital signal processor (DSP) which compute the SINAD
value. The DSP first filters the audio signal using a flat
(unweighted) filter, a C-MES filter or a CCITT filter. The resulting
RMS signal level (signal + noise + distortion) is computed as
a reference value. The signal is then further filtered by the DSP
to produce a notch filter which suppresses the 1 kHz tone. The
RMS signal level of the resulting signal (noise + distortion) is
then computed and compared with the first reading and the Figure 8 - Frequency response of the notch filter on 2023 series Option 12
SINAD value computed and displayed on the LCD panel. The notch filter is also digitally implemented, allowing a
notch to be generated which is deep, but has steep sides to
maximize the mount of signal passed to the (noise +
distortion) measurement stage while allowing some variation in
the nominal 1 kHz tone frequency. The tolerance of the notch
filter stop band to small frequency errors (approximately ± 25
Hz for -50 dB) is useful when dealing with signals generated
from codecs where there is no guarantee that the input
Figure 5 - SINAD display on 2023 series
frequency is precisely the same as the output frequency. At
The digital implementation ensures that the measurements frequency offsets of greater than 100 Hz the flat, low loss, filter
are made quickly and very accurately. The C-MES and CCITT response ensures that the measurement accuracy is excellent.
filters computed by the DSP are much more accurate than the Since the signals being measured are noisy by nature there
traditional analogue filters, and since they are defined by a is some time variation in the value displayed. The use of digital
digital algorithm they are highly repeatable and stable with time techniques allows averaging to be performed simply over a
and temperature. number of measurement cycles. The signal generator defaults
to an average over 5 readings but the user can set the
averaging to between 1 (no averaging) and 127. Furthermore
the use of digital techniques ensures that the measuring
circuits are accurate true RMS responding with no errors
introduced because of the detector techniques.
Use of automatic mode for SINAD measurements
When making measurements such as RF sensitivity, the user
has to adjust the RF signal level until the required SINAD is
obtained. The RF level set is then the required sensitivity. The
2023 series Option 12 SINAD Measurement has a unique
automatic mode whereby the required SINAD is set (to, say, 20
dB) by the user and the signal generator then automatically
Figure 6 - Filter response compared to CCITT
standard error band lowers its RF output level from a preset value until the set
SINAD is obtained. This can considerably simplify both manual
testing and automatic testing.
To use the automatic mode the operator first sets the SINAD
target and the initial RF level. An RF step size is set (using the
increment facility) which is the amount by which the level is
reduced for each step down in the algorithm. The user then
starts the automatic mode and the generator automatically
lowers its level until the required SINAD target is reached. For
most applications a 1 dB step size is the most useful since a
finer step size would require more averaging to be applied to
overcome the unavoidable reading variance caused by the
noise in the signal being measured. This in turn would increase
Figure 7 - Filter response compared to CMESSAGE standard error band the time taken by the algorithm to arrive at the required SINAD

[Link] SINAD 4
 SINAD and its measurements using the 2023

target. Once the target SINAD has been reached the rotary intrinsic RF level accuracy of the signal generator at low
control can be used to make finer adjustments to the level (if levels. This should include both response and attenuator errors
required) to get closer to the SINAD target. (if separately quoted) and any environmental effects
GPIB measurement (temperature)
Like all the other functions on the 2023 series, the SINAD
Example: At 1 GHz the RF level accuracy of 2023 series is ± 0.8 dB and has an
option is fully supported over the GPIB. The GPIB control of the environmental impact of ± 0.02 dB/° C outside the temperature range of 17° C to
27° C . For normal factory or laboratory use the environmental factors do not add
SINAD is simple to operate and has been designed to minimize to the uncertainty. The uncertainty is directly reflected onto the uncertainty of the
the chances of an invalid response. RF sensitivity.

To find the current SINAD value a query is sent: If the impact of RF level uncertainty is needed to be known
sinad: value?
(i.e. what is the SINAD at a specific RF level) then according to
The signal generator will return a single value in dB. The
ETSI ETR 028 Section 5.3.2 an empirical relationship between
returned value is the average value over the number of
SINAD and RF level in a receiver operating below the knee point
measurements specified (also controllable over GPIB).
Compound queries such as: (i.e. when the receiver is subject to the FM capture effect)
sinad: value?; value?; value? is:0.375 dB (RF Input Level) = 1 dB (SINAD).Applying to the
will return the appropriate number of measurement above example this results in a SINAD uncertainty of
responses in one message, each measurement being (0.8/0.375) or ± 2.13 dB.
from independent measurements. If three values are Typically the uncertainty is much lower than this for 2023
requested it increases the delay until the values are [Link] effects. This is a particularly relevant factor
ready to sent. To reduce the controller waiting time the because often the input impedance to a receiver is not well
SRQ "mav" (message available) can be used to alert the defined (and often not even intended to be 50 W ) and a
controller when a measurement has been completed. standing wave is generated between the signal generator and
The weighting filters can also be selected over the GPIB. the receiver causing level errors in addition to those specified
Whenever the weighting filter is changed it automatically in the data sheet.
restarts the measurement averaging to avoid erroneous results.
Example: At 1 GHz 2023 series VSWR is specified as 1.3 (max). If the generator
The system programmer needs to ensure that the timing of is used to test a radio with an input VSWR of 1.5 the resulting standing wave
any system settling (including receiver AGC) does not disturb causes an uncertainty in the delivered RF power according to the following
formula:
the SINAD measurement - if there is a danger of this happening
two measurements can be requested and the first discarded. Power uncertainty =10 log(1/(1 ± ΓsΓl)2
SOURCES OF ERROR IN SINAD MEASUREMENTS
ETSI maintain a useful document, ETR 028, which describes where Γs and Γl are the modulus of the reflection coefficients
methods of estimating measurement uncertainty on various of the source and load respectively and are related to the
parameters, including SINAD. Copies can be obtained through VSWR by the formula:
the ETSI documentation service, national standards Γ = (r-1)/(r+1)
organizations or from the ETSI CD ROM based document Putting the example into the above formula Γs=0.13 and Γl
subscription service. =0.2
In principle SINAD measurements appear to be straight- And inserting into the power formula the level uncertainty is
forward, but there are considerable opportunities for apparent approximately 0.22 dB.
differences in readings between different manufacturers' The accuracy can be improved by masking the radio input
implementations, including the SINAD voltmeters, weighting VSWR using an attenuator pad on its input.
filters and notch filters. The principle sources of errors are cable losses between the signal generator and the receiver
identified below: input
RF Level errors Example: calculate as above for absolute level errors
Uncertainty in the RF level applied to the receiver will result
in errors either in the derived measurement (e.g. RF sensitivity) All IFR signal generators feature good RF level accuracy and
or the SINAD value. As the RF level is decreased the receiver low output VSWR to minimise RF level errors.
noise in the audio output increases and the SINAD ratio
reduces. For FM receivers this error can be compounded by the
"capture effect" observed on receivers where a change in 1 dB
of RF level causes a greater change in the SINAD value. RF
level errors can be caused by a variety of effects including:

5 SINAD [Link]
 SINAD and its measurements using the 2023

RF leakage characteristics that minimizes errors in the SINAD

From the signal generator. If the signal generator leaks RF measurement. The notch filter is sufficiently wide to allow some
signals they can add (or subtract) with the directly applied frequency errors in the 1 kHz tone without affecting the SINAD
signals and cause RF level errors. This can be minimized by measurement.
ensuring that the signal generator does not leak RF signals at
Example. The ideal response is a very narrow filter at 1 kHz with virtually no
its carrier frequency. All IFR signal generators feature low RF bandwidth and a flat response outside this region. ETSI ETR 027 expects SINAD
signal leakage to minimise this type of error. notches to have less than 1 dB impact on the noise + distortion measurement.
The filter on 2023 has a noise bandwidth of approximately 150 Hz. Assuming the
To check for RF leakage effects add a high value 20 dB pad measurement bandwidth is around 3 kHz and the noise is reasonably "white" the
filter removes around 5% of the signal or 0.21 dB. Compared to meters only just
between the signal generator and the radio while it is operating meeting the ETSI objective the 2023 will read 0.79 dB higher.
at a level similar to the RF sensitivity. The SINAD ratio should
worsen considerably (to less than 3 dB). Do not test using the The above examples all give worse case analyses for the
level control or the carrier on off function since it changes the errors involved. In practice some of the errors may be lower in
internal settings of the signal generator. real systems. In particular the RF output level accuracy and
VSWR of signal generators sometimes has the potential for
Modulation errors
"double counting" errors since the VSWR issues have to be
If the modulation on the carrier is not accurate it will affect
taken into account since the manufacturers of the signal
the audio level in the receiver and therefore affect the SINAD
generator have to include an allowance for VSWR in their error
ratio. The excellent modulation accuracy of 2023 series
budgets. For those with the ambition to do a more rigorous
ensures low modulation errors are introduced.
analysis ETSI ETR 028 provides a statistical approach to error
Example: On 2023 the FM modulation accuracy is ± 4%. To a first approximation
this will have a similar effect on the SINAD analysis
The nature of some the errors in SINAD measurements
SINAD detector errors makes it difficult to assign a definitive statement of the
The use of a detector with other than a true RMS responding accuracy. However the excellent RF accuracy and VSWR
characteristic will introduce an additional error which is hard to of the 2023 series signal generators and the precise
quantify since the mixture of signals which constitute the (noise digital implementation of the SINAD meter facilities on
+ distortion) signal is a complex signal which varies between Option 12 provide the users with simple and fast
test situations and radio types. The SINAD option on 2023 measurement results which have an accuracy much
series uses true RMS detectors which introduce negligible error. better than competitive solutions based on more
Example for 2023: Neglible traditional analogue technologies.
Weighting filter errors
Errors in the frequency response of the weighting filter (C-
MES or CCITT) will cause measurement errors which are
dependent on the spectral content of the audio signal. The
2023 series SINAD option minimizes these errors by using a
digital weighting filter which gives a very close and repeatable
response compared with the theoretical filter.
Example for 2023: Less than 0.1 dB
Notch filter errors
The wider the bandwidth of the notch filter used to remove
the 1 kHz tone the more noise and spurious signals it fails to
pass on to the (noise + distortion) meter and "improves" the
observed SINAD. In contrast if the notch is not deep enough it
will pass on additional 1 kHz tones to the (noise + distortion)
meter and make the observed SINAD appear worse than it is.
The use of a digital filter on 2023 series SINAD option ensures
a deep but steep sided notch filter which minimizes
measurement error. The digital filter is able to produce a deep
filter with a well defined narrow bandwidth and flat pass band

IFR, 10200 West York Street, Wichita, Kansas IFR, Longacres House, Norton Green Road, Stevenage, Herts
67215-8999, USA. E-mail: info@[Link] SG1 2BA, United Kingdom. E-mail: info@[Link]
Tel: +1 316 522 4981 Toll Free USA: 1 800 835 2352 Fax: +1 316 522 1360 Tel: +44 (0) 1438 742200 Freephone UK: 0800 282 388 Fax: +44 (0) 1438 727601
As we are always seeking to improve our products, the information in this document gives only a general indication of the product capacity, performance and suitability, none of which shall form part of any
contract. We reserve the right to make design changes without notice. All trademarks are acknowledged. Parent Company IFR Systems, Inc. © IFR. 2000.

[Link]
Part no. 46891/885
6 SINAD Revision A