Beruflich Dokumente
Kultur Dokumente
Lock-in amplifier
G Gurunadha Reddy
7th semester, SPS
NISER
Dt: 09/10/2017
Abstract
Precision experiments involve special techniques to obtain the small signal from noise,
magnitudes greater than the signal itself. Such is the phase sensitive detection technique.
A lock-in amplifier is a device which utilizes this technique to perform detections which
may seem like finding a needle in a haystack. This experiment is aimed to learn the theory
of PSD and to design simple lock-in amplifier to test the same. The working and utilization
of the lock-in amplifier was learnt and implemented. The final part of the experiment
involves measuring an unknown resistance. This has been successfully completed and the
observations of each stage were recorded.
Contents
1. Introduction
2. Lock-in amplifier
a. Phase sensitive detector
b. Band-width and noise reduction
4. Debugging
5. Calibration of Lock-in
7. Sources of error
8. Conclusion
9. References
Introduction
In signal detection it is often the case that a signal is overwhelmed by noise. Usual
amplification of the signal also unnecessarily amplifies the noise. One of the main
challenges is to isolate and obtain the signal with minimal loss. The signal is usually
within a narrow frequency band whereas, the noise can span over a wide range of
frequencies. At low frequencies, the flicker noise, which is inversely proportional to the
operating frequency, is ever-present in electronic circuits. The noise due to
electromagnetic pickup from motors and choke circuits has its power spectrum at the
frequency of motor or the mains. The Johnson-Nyquist noise, which is of thermodynamic
origin, spans over all frequencies. It is approximately equivalent to white noise and varies
with temperature T as
< ( )2 > = 4
Where, the LHS term represents the mean square fluctuation of voltage and is the
frequency bandwidth. These noises can be respectively minimized by working at higher
frequencies, electromagnetic shielding and reducing the temperature or bandwidth of
amplifier. At room temperature, weak signal detection is achieved by minimizing the
bandwidth of the amplifier to the frequency range of the signal. This is usually achieved
by a lock-in amplifier.
The speciality and working of lock-in amplifiers is explained in the following chapters of
the report and experiments which have been designed to test its effectiveness are also
discussed.
Lock-in amplifier
A lock-in amplifier is a narrow band pass filter attached to a phase-sensitive detector
followed by a low pass filter. It can extract signal from an extremely noisy background
with the help of a carrier or reference waveform. To achieve optimum clarity in the signal
detection, the frequency of the reference wave must be close enough to that of the signal.
State-of-art lock-ins utilize digital signal processors for phase sensitive detection whereas
the traditional models rely on synchronous demodulation technique. The working of a
lock-in amplifier can be algorithmically represented in four sections as shown in Fig. 1.
Signal in DC out
Signal Low
channel Mixer pass filter
Reference
channel
Reference in
Fig. 1. Working of a Lock-in Amplifier
The signal channel comprises of a preamplifier to amplify the weak signal input and a
bandpass filter to select only the band which accommodates the signal. Background noise
outside the selected band is eliminated in this channel. The reference channel comprises
of a phase-shifter for shifting the phase of reference input. A weak signal is often
modulated with a high frequency carrier waveform to eliminate the 1/ noise. The
demodulation or phase sensitive detection process occurs at the mixer channel by
combining the outputs of both reference and signal channels. The output from the mixer
passes through a narrow low-pass filter to give an almost DC output. The greatest
contribution in noise reduction is from the mixer and the low-pass filter sections. These
sections are explored in the later chapters.
The principle of phase sensitive detection, which forms the core of a lock-in amplifier, is
discussed in detail in the following section.
Highly accurate lock-ins employ digital signal processors for PSD whereas, analog ones
such as the one built in this experiment rely on synchronous demodulation. The PSD used
in this experiment can be visualized as an electronic reversing switch whose position is
determined by the polarity of the reference waveform. The switch is used to power an
inverting or a non-inverting amplifier, one at a time such that the output is synchronously
demodulated.
1
Lets consider a signal denoted by sin() and reference by 2 ( + ), where is the
phase shift of the reference wave. The reference wave can be of any shape but should have
same frequency as the signal. Now the PSD flips the sign of signal wave if the reference
is negative (or positive) at a given time and leaves it unchanged if the reference is positive
(or negative). This can be graphically visualized as follows.
1.0
0.5
2 4 6 8 10 12
0.5
1.0
0.5
2 4 6 8 10 12
0.5
1.0
1.0
0.5
2 4 6 8 10 12
0.5
1.0
1.0
0.5
2 4 6 8 10 12
0.5
1.0
Clearly, the PSD output voltage is dependent on the amplitude of only the signal,
frequency of both reference and signal waveforms and the phase difference between them.
This output from PSD passes through the low-pass filter, where any AC component is
bypassed to earth to give a DC output. If the signal waveform is sin() and the gain of
lock-in amplifier is , the final DC output of the lock-in amplifier can be calculated as
follows.
sin() for 0 +
= {
sin() for + 2
By calculating the area under the output graph for range [0,2] we obtain .
2
= { sin() sin()}
2
1
= {() ( ) ( ) + (2 )}
2
2
= cos()
Notice that the DC output of lock-in amplifier is only dependent. The amplifier locks in
phase of the signal with the phase shifted reference, thus giving the amplifier its name.
2
When the reference and signal are in phase, the DC output is maximum ( =
)
2
Similarly, the minimum ( =
) occurs at a phase shift of .
where, = sin( + )
1
= 2 {cos(( ) ) + cos(( + ) + )}
The output frequency is thus, a mixture of sum and difference of signal and reference
frequencies. After passing through the narrow low-pass filter, the output becomes nearly
DC with extremely low frequencies contributing to the output. So, the (( + ) + )
term is always filtered out and the (( ) ) term contributes to the DC output
only if ( ) is very small. The contribution of this noise to drops rapidly as
1
differs from . At = , the output is pure DC which is equal to 2 cos().
If , the integration time is large compared to the period of the reference signal, then only
noise frequencies differing from by / (where is a small number) will contribute to
VDC. Thus, the effective bandwidth of the lock-in amplifier is /. This implies that the
effective bandwidth W is a few Hz. So, one can understand how thermal noise and other
noises are suppressed by the lock-in amplifier.
Construction and Working
This section deals with the construction of individual channels of the lock-in amplifier.
Their working is demonstrated, and the data is recorded and interpreted. The pin
diagrams of ICs used are given in the Appendix section.
The Pre-amplifier
Often the signal is very small to be fed into the PSD. So, the pre-amplifier is used to
amplify the signal before further processing. It makes up the signal channel of this lock-
in amplifier.
There can also be a bandpass filter connected to the output of pre-amplifier, but this had
been omitted in this experiment as the signal was directly taken from the function
generator.
Phase Shifter
The reference signals phase should be adjusted to obtain maximum DC output from the
lock-in amplifier. So, a phase shifter is placed in the reference channel.
If is the angular frequency of the reference waveform then, by using Kirchoffs Laws
and phasor algebra, we obtain
Freqency(KHz Freqency(KHz
( ) ) Delay (s) ( ) ) Delay (s)
0 -495 0 -100
0.04 -485 0.03 -105
0.49 -420 0.05 -91
0.92 -360 0.1 -82
1.3 -315 0.16 -75
1.74 1 -265 0.26 5 -63
2.28 -225 0.37 -53
3.04 -185 0.45 -47
4.16 -145 0.66 -37
5.94 -100 0.81 -29
9.96 -70 1.11 -22.5
1.61 -16
3.1 -8
9.97 -5
-200
Delay [/) in s
-300
-400
-500
-600
Resistance R5 in (K)
-40
-60
-80
-100
-120
Resistance R5 in (K)
Low pass filter
The output of the PSD has AC component from the signal and also due to noise. So, a
narrow low pass filter is used to bypass these components to the ground and obtain a
nearly DC output. The narrower the filter, greater the efficiency in giving DC output.
R7
6 1
= ( )
7 1 + 6
Hence,
6 1
= ( )
7 1 + 6
1
As the frequency varies from 0 Hz, the gain is almost constant till = . Above this
6
1
frequency, the gain falls rapidly. Hence, = is the cutoff frequency of the low pass
6
filter. In this experiment, the cutoff frequency of the low pass filter is found to be
= 974.659 / or = 155.122
This can be verified in the observations and plots below as the observed value of cutoff
obtained from the graph is 156.104 Hz
Table. 2. Frequency vs. Gain for Low-pass filter
0
-3dB
-5
-10
Gain in dB
-15
-20
-25
-30
156.104 Hz
-35
In a chip there are two-unit gain amplifiers A & B. The comparator is used to switch the
required amplifier through selection pins SEL A and B. Phase shifted reference is fed
through one of these pins while the other is grounded. The internal compensation
capacitor can be controlled by using COMP pin. In this experiment, this pin is connected
to the output pin. The pins are , , are used to give feedback and one of them is
connected to one of the amplifiers non-inverting inputs for feedback. The other non-
inverting input is grounded. Either of the remaining feedback pins are connected
simultaneously to the inverting inputs of both the amplifiers. The remaining feedback
pin is grounded. The differential and common mode offset can be adjusted by connecting
impedances across suitable pins.
The complete circuit diagram of the Lock-in amplifier built in this experiment is given
below.
Fig. 10. Connections across the pins of AD630 PSD.
Fig. 12. Output of AD630 before offset adjustment. The sine wave is of the signal.
Fig. 13. After adjusting the differential offset the peak heights became equal
(notice that the scales of signal and output waves are changed for better comparison)
Fig. 14. Final output from the low-pass filter (yellow).
It is nearly DC but has a low frequency AC component due to the higher low-pass cutoff
frequency
V1 V2
The pins V1 and V2 are fed into pre-amplifier and phase shifter respectively. The former
serves as Signal whereas the latter as reference. The voltage difference across the small
2
resistance R3 is given by = + +
1 2 3
By substituting the values, we get
We now calculate the signal using the above formula and plot it across the output
observed to get the amplification of the lock-in.
1.005KHz
6 2.000KHz
3.000KHz
Output voltage of lock-in (V)
4.000KHz
5.051KHz
5
4
Equation y = a + b*x
Weight No Weighting
3 Residual Sum of
Squares
0.01304 0.00533 0.00703 0.00581 0.0029
2 D (1)
Intercept
Slope
-0.14178
90170.85545
0.03572
711.53616
Intercept -0.19466 0.02189
D (2)
Slope 85862.17712 445.25848
Intercept -0.22337 0.02258
D (3)
1 Slope
Intercept
79503.76171
-0.22346
419.72337
0.02122
D (4)
Slope 71750.46564 415.08178
Intercept -0.188 0.01677
D (5)
Slope 63773.37666 338.48662
0
0.00001 0.00002 0.00003 0.00004 0.00005 0.00006 0.00007 0.00008 0.00009
Signal Voltage (V)
Note, that the amplification factor or gain of the lock-in is dependent on the frequency of
the signal. The respective gains with error for each given frequency are shown below.
.
Frequency Standard
(KHz) Value Error
1.005 Intercept -0.14178 0.03572
Slope 90170.86 711.5362
2.000 Intercept -0.19466 0.02189
Slope 85862.18 445.2585
3.000 Intercept -0.22337 0.02258
Slope 79503.76 419.7234
4.000 Intercept -0.22346 0.02122
Slope 71750.47 415.0818
5.051 Intercept -0.188 0.01677
Slope 63773.38 338.4866
We follow the same procedure as in the previous section to obtain the unknown resistance.
But, now we calculate the signal voltage by dividing the lock-in output with the gain
obtained for respective frequencies. Then the unknown resistance is obtained by back
calculating the signal voltage from the lock-in output.
Frequency Unknown
(kHz) Vin p-p (V) Vout (V) Vsig (V) resistance
1.005 1.04 1.27 1.41E-05 0.183861454
1.32 1.71 1.9E-05 0.195048443
1.76 2.31 2.56E-05 0.19761487
2.04 2.71 3.01E-05 0.200013556
2.72 3.57 3.96E-05 0.19761487
3.2 4.26 4.72E-05 0.200437939
3.92 5.25 5.82E-05 0.201647826
4.24 5.61 6.22E-05 0.199212834
4.88 6.43 7.13E-05 0.1983862
2 1 1.14 1.33E-05 0.180255775
1.24 1.48 1.72E-05 0.188722797
1.84 2.23 2.6E-05 0.191633475
2.24 2.84 3.31E-05 0.200472431
2.84 3.53 4.11E-05 0.196535361
3.48 4.4 5.12E-05 0.199920702
3.92 4.93 5.74E-05 0.198858971
4.4 4.54 5.29E-05 0.163150163
4.88 6.12 7.13E-05 0.198296905
5.24 6.54 7.62E-05 0.197347122
3 1 1.03 1.3E-05 0.175887826
1.28 1.35 1.7E-05 0.180103584
1.96 2.19 2.75E-05 0.190803616
2.32 2.6 3.27E-05 0.191374434
2.72 3.09 3.89E-05 0.193993926
3.16 3.48 4.38E-05 0.188057526
3.64 4.18 5.26E-05 0.196098131
4.08 4.69 5.9E-05 0.196295904
4.72 5.48 6.89E-05 0.19826092
5 5.81 7.31E-05 0.19842879
4 1 0.9 1.41E-05 0.191597277
1.32 1.26 1.98E-05 0.203209233
1.88 1.82 2.85E-05 0.206091634
2.24 2.27 3.56E-05 0.215737013
2.72 2.77 4.34E-05 0.216799206
3.24 3.36 5.27E-05 0.220770525
3.88 4.05 6.35E-05 0.222213337
4.36 4.56 7.15E-05 0.22265127
5.04 5.28 8.28E-05 0.223023285
5 1 0.78 1.22E-05 0.166050973
1.24 1.03 1.62E-05 0.176832612
1.64 1.39 2.18E-05 0.18043375
2.04 1.82 2.85E-05 0.189927584
2.68 2.4 3.76E-05 0.190644057
3.12 2.89 4.53E-05 0.197192354
3.56 3.29 5.16E-05 0.196740025
4.04 3.74 5.86E-05 0.197077507
4.56 4.26 6.68E-05 0.198880214
5.04 4.74 7.43E-05 0.200214086
3 1 2 2 2 2 2 2
= ( ) +( ) +( ) +( ) +( )
3 1 2
Frequency Unknown
(Hz) resistance
R3 ()
1.005 0.197093
2 0.191519
3 0.19093
4 0.213566
5 0.189399
Sources of error
1. The wires and oscilloscope probes act as antenna and pickup signal from the mains
or even worse, from the parts of the lock-in amplifier. This can be avoided by using
shielding like an earthed metal box to contain the lock-in.
2. Multiple core wire should be used for connections in the circuit as single core would
cause more noise to be picked up from the surroundings.
3. The pre-amplifier gain must be set low as at gain above 1000, it begins deviate
significantly.
4. The offset adjustment should always be taken care of to obtain reliable outputs.
5. Perform calibrations for various frequencies as the gain varies with frequency. And
perform measurements by considering respective gain for each frequency.
6. After testing transfer the circuit to a pcb as there is a chance of loose connections
in the breadboard.
Conclusion
Principle of phase sensitive detection has been learned and understood. Lock-in
amplifier has been successfully constructed and calibrated. An unknown resistance
was also found using the lock-in. A small Ac component was found in the output which
couldnt be eliminated due to lack of time. This shall be done in the next classes and a
better version will be created. More experiments will also be designed to test the
reliability of lock-in.
References
1. AD630 data sheet.
2. A Simple Low-Cost Lock-In Amplifier for the Laboratory, Sengupta
3. Design and Implementation of the Phase-locked Amplifier, Song and Liu
4. Princeton applied research paper
APPENDIX