Professional Documents
Culture Documents
Abstract
We describe a low-noise preamplifier to be used with generic gas flow proportional detectors in a
Mössbauer spectroscopy setup. This preamplifier is responsible for providing the detector high
voltage bias, for converting the positive or negative charge pulse of the detector to a voltage signal,
for signal amplification and pulse shaping and finally, for the matching between the high impedance
detector output and the low impedance of the coaxial cables.
To test the preamplifier in a real Mössbauer spectroscopy setup a printed circuit board prototype was
made. A detailed circuit analysis and the most relevant simulation results, together with some
experimental ones, will be presented.
Keywords: low-noise preamplifier, compact preamplifier, gas flow proportional detectors, Mössbauer
spectroscopy.
D4
HV C5 10n
R15 R16 (up to 3KV) R3 L1 10u C12 +12 V
R17 6.8k 10u
50M,0.5W 50M,0.5W 33M,0.5W AGND PVB
-12 V
C2 10n R7
C10 C11 3.3k C6 10u
OP2
2.2n,3kV 2.2n,3kV C13
+
10u
D3 R6 -
C4 100n
1.2k
C7 10u C9 10n
OP1
R10 1k OUTPUT
DET Q1 +
R2 22
-
Q2 R14 51
R1 200 R9
C1 R13
D2 4.7k
18.8n,3kV R4 33k
2.2k C8 10n
D1 D5
R5 200 R8
C3 56
15p L2 10u
R11 1k R18
51
added. The others notable features are the The major source of noise in resistors is
preamplifier voltage bias, PBV, (±12 V) and the thermal noise. This source of noise could be
2
small size of the printed circuit board. modelled as voltage source, V R(f), in series with
2
The preamplifier itself consists of two the resistor or as a current source, I R(f), in
cascaded stages. The input stage uses two parallel. Depending on the noise analysis, one
high-frequency discrete bipolar transistors of the following expressions could be used [6]:
(2N3904 and 2N3906) and serves as a current
VR2 (f ) = 4 ⋅ k ⋅ T ⋅ R (1)
to voltage converter. This stage is also
responsible for the low output noise and for part 4 ⋅ k ⋅T
I R2 (f ) =
(2)
of the preamplifier gain (9.6 dB). The diodes D1 R
and D2 at the preamplifier input prevent the where k is the Boltzmann constant and T is the
input stage transistors from breakdown in case temperature in Kelvin.
of high voltage arising at the input. The
capacitor C2 increases the gain at high V2Q1
frequencies, and its value (15 pF) was selected
experimentally.
I2Q1
* V2R6
R6
R7
* V2Q2
2
where V in_i and Ai are respectively the input- The integrated preamplifier output noise is:
referred noise spectral density and the gain of 2 2
the stage i. So in a multi-stage amplifier the Von = (Von _ in _ stg (f ) + Von _ out _ stg (f )) ⋅ ∆f (9)
noise spectral density at the output is mainly
due to the first amplifier stage. This was the where ∆f is the preamplifier noise bandwidth.
principal reason for using bipolar transistors in The noise bandwidth is calculated for the brick-
the input stage. wall response approach. The frequency limits
2 2
If V in_Q1(f) and V in_Q2(f) are, respectively, are obtained by:
the input-referred spectral density noise due to π ⋅ f −3dB
fl = (10)
transistors Q1 and Q2, the output noise spectral 2
density of the preamplifier input stage is: The calculated preamplifier output noise
V 2 2 values are presented in table 1. The parameter
2 in _ R 2 (f ) + Vin _ R 4 (f )
Von _ in _ stg (f ) = values needed to calculate those noise values
+ Vin2 _ Q1(f ) are taken from the transistors and the
(6)
operational amplifier datasheets. As expected,
2
hieQ1 the output noise in mainly due to the
( )
+ Vin2 _ R 6 (f ) + Vin2 _ Q 2 (f ) ⋅
hfe ⋅ R 6
preamplifier input stage contribution . This could
Q1 be reduced for higher gain of the input stage.
To analyse the noise of the output stage of The noise of the output stage of the
the preamplifier the circuit shown on Fig. 3 was preamplifier could be reduced in applications
used. where inverted polarity is not need. In these
I2Req cases, a non inverter voltage configuration could
be used, and the first operational amplifier is not
I2OP2- Req = R11+R12 necessary.
OP1
- R10
+
-
Simulated and Experimental Results
+
OP2
V2OP1 I2R10
* V2OP2
*
The proposed circuit was simulated using
the Multsim simulation program. Several
simulations indicated the optimal values for the
Figure 3 – Circuit equivalent for output stage noise analysis. capacitor C1 and for the potentiometer R1. The
value of C1 affects the low frequency limit and
The simplified noise spectral density at the
the output signal fall time. The value of R1 also
output of each operational amplifier in Fig. 3 can
changes the preamplifier gain. A compromise
be represented by ([5] and [6]):
between the minimum gain, the output signal fall
2
Von _ OP
(f ) = Vin2 _ OP (f ) + I in
2
_ OP −
(f ) ⋅ R −2 time and the bandwidth are needed to define
(7) these values.
2
+ I in _ OP +
(f ) ⋅ R +2 The simulation results obtained for the
extreme values of potentiometers R1 and R12
the second and third term of this equation are
are resumed in table 1.
null if there are not resistors at the inverter and
non-inverter operational amplifier inputs. Using Table I – Preamplifier simulated performance.
superposition, and assuming all noise sources Fall Time Von
Gain
are uncorrelated, the output noise spectral [dB]
Bandwidth
[µs] [µV]
density of the preamplifier output stage is given
by: R1=200 Ω 38.0 kHz to
9.6 30.0 38.5
R12=0 Ω 24.0 MHz
V 2 (f ) +
in _ Op1 R1=200 Ω
30.3 30.0
38.0 kHz to
16.7
2 R12=10 kΩ 4.5 MHz
V 2 Re q
(f ) ⋅ + R1=0 Ω
in _ Op 2 R10 31.1 4.0
456.4 kHz
37.1
R12=0 Ω to 22.5 MHz
2
Von ( f ) = I 2 ( f )
in _ R10 R1=0 Ω 381.6 kHz
_ out _ stg 51.0 4.0 17.8
+ I2 2 (8) R12=10 kΩ to 5.3 MHz
in _ Re q
( f ) ⋅ Re q
2 Some of the simulated results are presented in
+ I in _ OP −. (f ) Figs. 4 and 5. Fig. 4 shows the working
bandwidth for a preamplifier gain of 25.1 dB.
The high frequency limit and, the output noise
2 are highly dependant of the output stage gain.
hieQ1
⋅ The transient characteristics for an input
hfe ⋅ R 6
Q1 square signal are shown in Fig. 5. In this figure
Ibersensor 2010, 9-11 November 2010, Lisbon, Portugal IB-119
the preamplifier gain factor is 25.1 dB also, and probe points (with matched terminations at both
the output signal rise and fall time are, ends). Due to that, the measured output signal
respectively, 20 ns and 30 µs. In all the amplitude and noise level are reduced by a
simulations, the rise times are in the pretended factor of two.
range. The output signal fall time values are Fig. 7 shows a more detailed photograph of
slower and sit in the range 4 to 30 µs. one of the experimental results for a preamplifier
gain of 23.6 dB. The output signal fall time is
9.7 µs, which is quite near of the expected one
(10 µs).
The noise visible in the input signal shown
in Fig. 7, and when the preamplifier supply
voltage was switched off, shows that the
oscilloscope output noise is of the same order of
magnitude of the output preamplifier noise.
These two sources of noise are not correlated,
and so in the measured output signal the
oscilloscope noise is added to the preamplifier
noise. This could be minimized with suitable
Figure 4 – Simulated amplitude-frequency response.
oscilloscope probes.
Conclusions
References