Professional Documents
Culture Documents
Lecture 2: Electrical Measurements: Brown
Lecture 2: Electrical Measurements: Brown
2
Lecture 2: Electrical Measurements 2-6
power: A
2
R/2 = R
t
1
+1/
t
1
V
2
(t) dt = RV (t)
2
However, for other types of signal (e.g. square or triangle waves) or for noise these relationships ar
dierent i.e. for white noise the peak-to-peak value (99/
Ways to measure AC signals:
1. If one can sample an AC signal fast enough then it is possible to sample the waveform and therefore
obtain all of these measurements by calculation, or even just give the waveform and therefore avoid
the diculty. One could, for example, just look at the waveform on a CRO (up to perhaps 100 GHz
these days).
2. Traditionally one used an true rms voltmeter to deliver the rms value for a wave. The voltmeter
obtains this by essentially determining the average power in the signal and taking the square root
of it i.e. the input waveform is squared (rectied) and then averaged, and then the square root
is taken of the nal value. The AC voltage signal produces the same heating in a resistor as DC
signal with a voltage equal to the root mean-square value of the AC signal.
3. Spectrum Analyzer: measures the spectrum of the signal from which is it possible to recover the
full time domain behaviour (i.e. give the waveform), or use the spectrum analyzer to deliver the
amplitude (in whatever form) of the most important sinusoidal components.
4. Lock-in Amplier: a tool that allows you to directly measure noise at various frequencies -discussed
later in course
2.4 Measuring Resistance
To measure resistance accurately one really needs to compare them with precision resistance standards.
The classic accurate method of doing this is the Wheatstone Bridge (actually invented by Hunter Christie
see Figure 8). The resistors labelled R
1
, R
2
, and R
s
are all standard resistors used to measure the
Null
Detector
R1
Rs
Rx
R2
V
Figure 8: The Classic Wheatstone Bridge
unknown resistance R
x
. R
s
is a variable resistor and is adjusted until the null detector declares that
the bridge is balanced, i.e. that there is no current owing along the central branch. We can solve to
determine what this means in terms of the values of the resistors and we nd that the balanced condition
corresponds to R
x
= R
s
R
1
/R
2
. The ratio R
1
/R
2
is referred to as a multiplier, and the ratio of these
two resistors could well be switch selectable for values between say 10
3
and 10
3
so that a broad range
Lecture 2: Electrical Measurements 2-7
of resistances can be measured. In general the sensitivity is optimal when all impedances are about the
same. One can use this bridge for both AC and DC measurements but you need to take into account
any reactances in the circuit. We often use such bridges for the sensor in sensitive thermometry in our
lab.
2.4.1 Errors in Resistance Measurement
The usual method for measuring the resistance is to pass current through it and measure the voltage
drop. Unfortunately, this very simple approach can cause some issues. If the resistor is based on many
turns of resistive wire then there may be some associated inductance, and there is always a little bit of
self-capacitance as well. The resulting impedance is dominated by resistance and may be represented as
the network shown here on Figure 9. One can calculate the impedance of this arrangement as:
Figure 9: Equivalent circuit of real resistor
Z =
R + j
CR
2
CL
2
2
+ L
C
2
L
2
4
+ C
2
R
2
2
2CL
2
+ 1
(1)
If R has been constructed so that L and C are small (obviously the desirable situation), then we can
assume that
2
LC 1 and
4
L
2
C
2
being the square of a small number will be real small. The previous
expression collapses to:
Z =
R +
L(1
2
LC) CR
2
1 +
2
C(CR
2
2L)
(2)
From this we nd that the real part is:
R
e
=
R
C (CR
2
2L)
2
+ 1
(3)
and the phase angle is approximately (assuming
2
LC 1):
tan =
L CR
2
R
(4)
So we can choose to have a frequency independent resistance by setting 2L/C = R
2
when we design the
resistor, or we can choose to have a resistor that has no phase shift (or reactance) by choosing L/C = R
2
.
Of course we cant do both at the same time. The usual choice is to do the later which means that most
resistors have a small frequency dependence.
2.4.2 Small Resistance
A resistor needs to be connected to the circuit with some wires and they of course possess resistance
as well. This lead resistance adds to the measurement of the resistance, and this is especially a big
problem if the resistance to be measured is small. It also becomes a big problem if we are trying to
make measurements in the presence of changing lead resistances (perhaps because they are in a large
and changing temperature environment), or if the leads are necessarily long. We can overcome this issue
by using a very sneaky technique called the 4-wire approach. In this case we connect four leads to the
resistor: two leads apply current through the resistor while the other two measure the voltage drop across
the resistor. If virtually no current ows in the sense leads then the resistance of the sense leads does
not enter into the measurement.
Lecture 2: Electrical Measurements 2-8
The resistance can be measured in a bridge conguration (called a Kelvin bridge - a 4 wire variant of the
Wheatstone bridge) or one can just make use of modern DVMs with input impedances of 10 M-10 G
in the sense leads. The unit will pass 1mA through the current leads (depends on the resistor under
study but is about this level for low resistances) while passing a few nA through the sense leads. This
leads to a reduction in the eect of lead resistance of about a factor of 10
6
.
2.5 Errors of Electrical Measurements
Remember from kindergarten that it is always possible to express any network of resistors and batteries
(however complicated) as a Thevenin equivalent or as a Norton equivalent circuit (this isnt a statement
that only one of these will work - both are equally valid for any situation, however often one is more
appropriate).
R
V R I
Figure 10: Thevenin and Norton equivalents
One nds the equivalent circuit elements from a real network by a couple of measurements. One measures
the open circuit voltage of the network to determine the voltage in the Thevenin equivalent circuit. If
one then also measures the short-circuit current then one obtains a series resistance. For the Norton
equivalent one measures the short-circuit current and that determines the value of the current source. If
one also measures the open circuit voltage then one can determine the value of the shunt resistor.
R V
Ideal
Voltmeter
R
I
Ideal
Ammeter
Figure 11: Here we show equivalent circuits for real meters
Lets model the meter when one is making voltage and current measurements. In Figure 11 we show
circuit models for real meters. They are shown as ideal meters in parallel or series with resistors. The
ideal voltmeter has an innite impedance while the ideal ammeter has a zero resistance. If we combine
the equivalent circuits with the Thevenin and Norton equivalents we obtain the following:
The intention of course is to measure the output voltage, V of the Thevenin circuit or the current, I of
the Norton equivalent. Any error in this measurement comes about from the non-ideality of the meter
and the non-ideality of the source under study.
Lets look rst at the voltmeter measurement in Figure 12. The ideal voltmeter is just making a simple
resistor divider measurement and so we obtain:
V
m
V
=
V + V
V
=
R
1
R
1
+ R
(5)
V
V
=
R
R + R
1
(6)
Lecture 2: Electrical Measurements 2-9
R1 V
Ideal
Voltmeter
R2
I
Ideal
Ammeter
R
V
R I
Figure 12: Here we show real meters connected to real sources
So the fractional error is given in the nal expression. If we now turn to the current measurement one
can show that it is equal to :
I
I
=
R
2
R + R
2
(7)
2.5.1 Power Transfer
It is often the case that one wants to transfer the maximum power from a source into a measurement
device (e.g. a meter) - this is especially so when one is trying to minimize noise.
Power absorbed by a measuring system (using the Thevenin circuit as the example):
P =
V
2
m
R
=
R
1
V
2
(R + R
1
)
2
(8)
where V
m
is the voltage over the internal resistor in the meter. This is a maximum as a function of R
1
when:
P
R
1
= 0 R
1
= RandV
m
=
1
2
V (9)
This is potentially a major diculty as it heavily loads the source output voltage i.e. V V/2 but if
R and R
1
are known (and stable) the error can be calculated and corrected. We will see that this power
matched position is often also the point at which the signal-to-noise ratio is maximized.
2.5.2 Thermal EMFs: from HP34401 manual
If one is trying to make low level voltage measurements it is frequently the case that thermoelectric
voltages will cause errors. These temperature dependent voltages are generated whenever one connects
two dissimilar metals together. The table below, Table 1, shows some typical levels:
Copper to Approx microV/degree
Copper < 0.3
Silver 0.5
Brass 3
Aluminium 5
Tin-Lead Solder 5
Copper Oxide 1000
Cadmium-Tin Solder 0.2
Table 1: Thermal EMFs
Lecture 2: Electrical Measurements 2-10
2.5.3 Power Line pick-up: from HP34401 manual
It is almost always the case that however careful one is when building electronics that there will always
be some electrical pick-up from the mains. In analog systems one can attempt to average these issues
away but making lengthy measurements but if you are required to make quick measurements there is
no real solution. If on the other hand one uses digital measurement techniques then it is possible to
suppress sensitivity by making measurements with an averaging time exactly equal to a multiple of the
power line cycle time.
2.5.4 Magnetic Field pick-up: from HP34401 manual
If you are attempting measurements in the presence of large varying magnetic elds then you need to
avoid induced currents in the circuit. It is sensible to use twisted pair wires to bring the signal to the
meter. Loose or vibrating leads can also induce error voltages so you should make sure the leads are
well secured. Whenever possible use physical separation or shielding materials to reduce the magnitude
of the magnetic elds.
2.5.5 Power dissipation eects: from HP34401 manual
Since real currents ow through resistors that are being measured it can heat the resistor and therefore
change its impedance. This often happens for resistors in cryogenic environments in which the specic
heat of materials has become very small, or in vacuum systems in which devices can have very high
degree of thermal isolation.
2.6 Amplication
Almost all of the techniques discussed in this section require signal levels of the order of a volt, and
currents of the order of 10mA in order to obtain good results. If the signal available is not of this level
then one needs to amplify the signal. We looked at some issues associated with noise in amplication
etc in the last lecture but will suspend discussion of the details of this until next time.