ADDA

1
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Vladimír Haasz
• Czech Technical University
• in Prague
• Faculty of Electrical Engineering
• Department of Measurement

2
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• WHAT INFLUENCES THE PRECISION
• OF MEASUREMENT USING A/D  MODULES?
• The Effective Number Of Bits (ENOB) or SIgnal to
  Noise And Distortion (SINAD) are typical integral
  parameters characterising the dynamic quality of
  an A/D module. However, they are influenced by
  several main factors

• Quality of ADC and S/H circuit
• Quality of pre-amplifier
• EMC aspects

3
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• static values of DNL and INL,
• jitter of a S/H circuit that causes the frequency
  dependence of ENOB (or SINAD)
• settling time of a pre-amplifier and frequency
  dependence of a pre-amplifier non-linearity.
• EMC aspect
• A/D modules disturbance immunity,
• internal noise of A/D module and cross-talk
  between channels.

4
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Effect A decrease in the effective number of
  bits
• Disturbing sources in the system case - power
  supply,
• - digital signals on mother board,
• - near-by plug-in boards
  (especially network board)

5
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Sources of disturbance outside the PC case -
  pulse controlled motors - monitors

6
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• The disturbance level in the area of PC plug-in
  boards or in slots of notebooks

Disturbing magnetic field Special board (PC
Card) with measuring coils Induced disturbing
voltage fDHD Results PC fDHD lt 20 kHzA/m
for fD lt 100 kHz fDHD lt 100 kHzA/m for fD gt
100 kHz Notebook fDHD lt 50 kHzA/m for 30 kHz lt
fD lt 800 kHz
Disturbance propagating over supply
lines Waveform, amplitude, RMS value, and
spectrum of disturbing components of supply
voltages were measured Results Amplitude
hundreds of millivolts, RMS value several
tens of millivolts, Fundamental freq. tens to
hundreds kHz.
7
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Measuring of disturbance immunity to disturbing
  magnetic field

The tested AD board was inserted into the area of
Helmholtz coils. The input of the tested AD
board was short-circuited using a 50 W resistor,
an output noise was measured.
8
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Measuring of disturbance immunity to disturbing
  magnetic field

a) fDHD 20 kHzA/m, G 100 b) fDHD
100 kHzA/m, G 10
9
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Measuring of disturbance immunity to disturbing
  magnetic field
• The influence in all three axes

a) fDHD 20 kHzA/m, G 100 b) fDHD
100 kHzA/m, G 10
A1 component perpendicular to the plug-in
board A-2 component parallel with the plug-in
board and perpendicular to the
motherboard A-3 component parallel both with
the plug-in board and with the mother-board
10
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
Effect of the disturbing magnetic
fields Disturbing sources in the system case A
decrease in the effective number of bits of
12-bit AD plug-in board placed in a PC case

• Sources of disturbance outside the PC case
• A decrease in the effective number of bits of
  12-bit AD plug-in board (Gain G 10)

CONCLUSION Disturbance causes a significant
decrease in the effective number of bits for the
gain G ³ 10, but there is no standard concerning
the disturbance immunity of A/D modules
11
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
The influence of a disturbance propagating over
supply lines

• The sinewave voltage with a frequency range from
  several tens to several hundreds of kHz was
  injected using a toroidal transformer into supply
  lines, which were protracted through the toroid.
• The AC component of the supply voltages was
  measured using a digitising scope.
• The AC voltage was injected both into separate
  supply lines 5 V, 12 V, 12 V and into all
  supply lines suddenly (but in this case, an AC
  component cannot be set independently in each
  supply line using one generator only).

CONCLUSION The disturbing effect of AC
components of supply voltage in tested PCs and
notebooks is several times smaller than in the
case of a disturbing magnetic field for all
tested AD plug-in boards and PC Cards.
12
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
INTERNAL DISTURBING SIGNALS (NOISE) IN A/D
MODULE Three components of internal noise - the
basic internal noise, - the additional internal
noise arising due to auxiliary digital signals
(e.g. external clock etc.) - disturbing signal
arising due to cross-talk between channels by
multichannel measurement. Basic level of internal
noise of an A/D module - the system noise
(NOISEIN) - short-circuited input by 50 W
resistor, no signals at the other analogue and
digital inputs - it is not generaly proportional
to the set gain - its value radically limits the
real resolution. The noise arisen due to external
digital signals (NOISEEDS), - it can be
determined from an output noise by connecting the
external digital signal (NOISEINEDS) and the
short-circuited analogue input according the
formula
(1)

• - if more external digital signals i
  (i  1, 2, ) are applicable in A/D module
  (external clock,
• signals of digital inputs/outputs, etc.), then

(2)
13
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
INTERNAL DISTURBING SIGNALS (NOISE) IN A/D
MODULE Cross-talk - It arises using multichannel
measurement (the additional disturbing signal -
NOISECT). - The value of the cross-talk CT in
dB should be specified in a data sheet - The
maximum value of NOISECT,j (in LSB) in the
channel j with set input range Uj and n-bit ADC
can be determined using formula
(3)

• where CT is the value of the cross-talk in dB,
• Ui is the set range of the other used channels
  i ¹ j,
• ch is the number of the used channels.

14
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
MEASUREMENT UNCERTAINTY ESTIMATION The standard
uncertainty of a measurement made by a DAQ
device Combined standard uncertainty
(Guide to Expression of
Uncertainty in Measurement) Uncertainty type A
(uA) Undesirable components of the digital output
signal, which sources are unknown or have a
random character Uncertainty type B (uB) The
sources of uncertainties are known and they have
not random character (the gain error, the offset
and the integral non-linearity). Usually there
are several components of type B, then
Expanded uncertainty U k uC interval,
where probable the measurement result
lies (coverage factor k 2 is recommended -
probability about 95 for normal
distribution) Measurement of a DC or a very low
frequency signal Sampling an AC signal
15
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation

• Measurement uncertainty estimation for DC or very
  low frequency input signal
• uA it is given by the internal noise (NOISEIN)
  and noise arising due to external disturbing
  signal (NOISEEMC), which sources are unknown or
  have a random character
• (LSB) (4)
• uB is given by two components
• uB1 - the gain and offset errors, the integral
  non-linearity (INL) of used ADM
• (LSB) (5)
• where dG is the gain error in ,
• dOff is the offset error in of full scale
  range,
• INL is the integral non-linearity in LSB,
• X is the measured value in LSB,
• n is the number of bits of the used A/D module.
• uB2 - the cross-talk (NOISECT) the noise arisen
  due to external digital signals (NOISEEDS)
• (LSB) (6)

16
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
Remarks A. Some producers publish in data sheets
Absolute Error - the first component
corresponds with the gain error, - the second
one incorporates obviously the offset error and
INL. The gain error in , or in LSB relative
to full scale range The offset error in mV
(mV), or of full scale range, or in LSB
B. The limits of the gain and offset errors,
and INL are assigned, and the uniform
distribution is presupposed therefore the
component uB1 is equal to the sum of these limits
divided by root of 3 C. The component uA - all
influences described by the total noise RMS
noise or System noise - expressed in LSB, or
in mV (mV). D. The measurement conditions are
not known (e.g. EMC in system case above all).
The best way for an estimation of the
component uA is to short-circuit the analogue
input of the system by the real operating
condition, to acquire the digital output, and to
calculate the RMS value of the output noise.
17
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
Measurement uncertainty estimation for dynamic
measurement The effective number of bits (ENOB),
or the Signal to noise and distortion (SINAD) and
analogue bandwidth of input amplifier (fBW) must
be known ! u is given by three components
(type A and type B is not separable in the second
one) uB1 - is given by the gain and offset
errors of the used ADM uB2 - is given by the
effective number of bits (ENOB), or the Signal to
noise and distortion (SINAD) uB3 - is given
by the cross-talk (NOISECT) and the noise arisen
due to external digital signals (NOISEEDS
) ________________________________________________
__________________________ The gain error for the
input signal with frequency f can be estimated
using ) (7) ) the frequency
bandwidth fBW is defined using classical way
(decrease of amplitude of 3 dB) and the 1st
order system is presupposed. Then the frequency
dependence of gain can be described for f
smaller then about 0.4fBW by
18
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
uB1 - is given by the gain and offset errors of
the used ADM (LSB) (8) uB2 - is
given by the Signal to noise and distortion
(SINAD), or by the effective number of bits
(ENOB) SINAD  6.02 ENOB  1.76 (LSB)
(9) uB3 - is given by the cross-talk
(NOISECT) and the disturbance arisen due to
external digital signals (NOISEEDS )
(LSB) (10) where dG,f is the gain error
corresponds to the input signal frequency f (in
), dOff is the offset error in of full
scale range, X is the measured value in
LSB, n is the number of bits of the used A/D
module SINAD is signal to noise and
distortion in dB NOISECT and NOISEEDS - see
(2) and (3)
19
ADDAEWADC 2002 - TutorialModular DAQ systems -
EMC aspect and measurement uncertainty estimation
CONCLUSION Pseudo-static signal The static
parameters are usually known (they are published
in data sheets of well-known producers) It
enables to estimate the measurement uncertainty
by sampling without problem. Dynamic signal
Not all of well-known producers publish all
necessary parameters, which are needed for the
measurement uncertainty estimation by sampling
the input signal with frequency near (in order)
to the declared analogue bandwidth or to the
Nyquist frequency referred to the maximum
sampling rate. (It concerns of the input signal
with frequency about 10 - 30 times smaller than
the maximum sampling frequency according our
experience.) It fundamentally complicates the
estimation of the measurement uncertainty for
virtual instruments.