ArticlePDF Available

A 1GHz Low-cost, Ultra Low-noise Preamplifier

Authors:

Abstract

A high quality, compact 1GHz preamplifier suitable for operation in conjunction with micro channel plates (MCP) and silicon Photomultipliers (SiPM), that is comprised of two integrated circuits is described in this paper. The amplifier requires no adjustment and has a flat response from low frequencies and adequate bandwidth for high speed measurement systems.
A 1GHz low-cost, ultra low-noise preamplifier
J. CHATZAKIS1,2, S. HASSAN2, E. CLARK2, and M. TATARAKIS1,2
1Department of Electronic Engineering
Hellenic Mediterranean University
Romanou 3, 73133 Chania, Crete
2Institute of Plasma Physics and Lasers
Hellenic Mediterranean University
74100 Rethymno, Crete
GREECE
Abstract: - A high quality, compact 1GHz preamplifier suitable for operation in conjunction with micro channel
plates (MCP) and silicon Photomultipliers (SiPM), that is comprised of two integrated circuits is described in
this paper. The amplifier requires no adjustment and has a flat response from low frequencies and adequate
bandwidth for high speed measurement systems.
Key-Words: - ultra low-noise preamplifier, MCP, SiPM
Received:
July 22, 2020. Revised: August 31, 2020. Accepted: September 1, 2020.
Published: September 1, 2020.
1 Introduction
There are numerous detection systems such as
neutron time-of-flight [1], photon counting [2] and
charged particles detection [3] where the signal
level is sufficiently low that amplification with low
noise and high time resolution is required to
measure the radiation signal. When the signal level
is low, there is a requirement for a low noise and
low distortion amplifier to make the measurement
feasible. Proposed designs of amplification modules
[3, 4] can be found in literature that can attain very
low noise spectrum densities but they may use DC
coupling and lack the ability to reach an adequate
high frequency cut-off close to 1GHz. DC coupling
also amplifies DC errors and low frequency noise.
Typically, low frequencies do not contain useful
information for these types of applications and the
suppression of these frequencies is preferred. AC
coupling, however, can reduce the total output noise
and eliminate the requirement for adjustment of the
amplifier. Another advantage of AC coupling is that
it eliminates the need to use a symmetrical supply to
the amplifier and this simplifies the amplifier
design.
In this paper a low cost, ultra low-noise preamplifier
is described, with a flexible design that can be
modified and adapted to any specific need. The final
design is compact and adjustment free. A typical
equivalent input noise voltage density of about
1.6nV/Hz1/2 and a high cut-off frequency (-3dB)
better than 1GHz is achieved.
2 Design Considerations
There are three requirements that are difficult to
satisfy in a preamplifier that is used in
instrumentation. These are low noise, high gain and
high bandwidth. For the attainment of low noise, the
noise of the first amplification stage is a significant
consideration. The contribution to the equivalent
input noise of the following stages is divided by the
gain of their previous stages so that generally the
following stages do not contribute significantly to
the equivalent input noise [5]. To achieve high gain,
it is possible to connect several stages in series,
however, this makes it difficult to achieve a high
bandwidth as each stage must have a much higher
bandwidth than the series combination.
For the first amplification stage, there are solutions
that have an input voltage noise density less than
1nV/Hz1/2 for example the Texas Instruments
LMH6554 [6]. However, such a device needs a lot
of external components and for this reason the final
design objective may not be achieved. The gain of
such a device and its bandwidth are also marginal.
Currently, the input stage of choice has become the
Analog Devices ADL5566 [7]. Each stage of this
device has a typical input voltage noise density
about 1.3nV/Hz1/2 at a voltage gain greater than 3
for single end operation and a bandwidth of
4.5GHz under these conditions. Most of the passive
components are inside the device and this enables
easy construction. The quality of the external
passive components can easily affect the
characteristics of the device close to the 1GHz
design specification. The minimization of the
external components in addition to the dual
construction in a single package (there are two
identical stages in a single package) allows for an
efficient printed circuit board (PCB) design.
Fig. 1. The proposed preamplifier circuit.
Fig. 1 shows the preamplifier circuit design. To
make an AC coupling from some 10’s of kHz up to
GHz frequencies special RF capacitors are required.
Regular low frequency capacitors have a low
resonance frequency that affects the response of the
amplifier to frequencies close to 1 GHz. To
overcome this problem, a parallel combination of a
100nF X7R dielectric, a 1nF C0G and a 47pF C0G,
all 0402 footprint (Murata
GRM155R71C104KA88D, GRM1555C1H102JA01
and GRM1555C1H470JA1D) is formed as a 101nF
equivalent capacitor. This combination has low
impedance at high frequencies. In this case, the low
cut-off frequency of the amplifier is close to 50KHz.
If a lower cut-off frequency is desirable another
capacitance can be connected in parallel with this
combination.
The Input stage is single-ended and asymmetrical.
Although this type of design is not supported from
the ADL5566 datasheet and application notes, it is
used as it is expected to deliver higher voltage gain
and lower noise. There is also an input impedance
change with this topology but it is likely to be
insignificant. The no-load voltage gain of the first
stage is expected to be x6.25 (15.9dB). Between the
first and second stage there is a direct (without the
use of other components) DC coupling. The DC
errors are insignificant and are unlikely to saturate
the second stage output. This in turn minimizes the
number of components used. In addition, the
termination resistance of the first stage is only 160
Ω when compared to the 200 Ω that is specified. It
is known that compensation and the high cut-off
frequency are affected by this asymmetrical
operation and the non-specified load. In this case it
is not a problem as this stage has a much higher
bandwidth than that required and thus the operation
within the useful bandwidth is not affected. The
second stage is normally terminated on a 200 Ω
resistor. At the second stage’s output a total voltage
gain of x35 is expected.
According to the ADL5566 datasheet RG=80Ω and
RF=500Ω. The input internal resistance (Ri) of the
first stage is given [8,9]:
 
12
G
iF
GF
R
RR
RR

 
80 140.6
500
12 80 500
i
R
 
 
(1)
The lower cut-off frequency of the input formed RC
can be calculated:
31~11.5
2
dB
f kHz
RC
(2)
More than one high-pass RC are formed and
determine the amplifier lower cut-off frequency.
The actual amplifier lower cut-off frequency is
always higher than the one from this calculation. To
match the amplifier input internal resistance to 50Ω
an external 77Ω resistor (R1) is placed in parallel to
the input resistance Ri (140.6Ω||77Ω=49.75Ω). The
first stage gain without load can also be found in
respect to the differential output:
1500 6.25
80
F
G
R
GR
 
(3)
The 5Ω output internal resistor of the ADL5566
forms a voltage divider with the 80Ω (RG) input
resistance of the second stage. The divider ratio can
be found equal to 0.941. The no-load voltage gain to
the ADL5566 second stage output can be calculated
to 6.252·0.941=36.76. The second stage output
formed voltage divider rounds the total ADL5566
voltage gain to 35.
For the final stage the fixed gain amplifier THS4303
(Texas Instruments) was chosen. This amplifier
forms the final frequency response and provides a
single-ended gain of 10. As only the one half of the
previous stage signal is used (one end only) the
amplification on this amplifier output is expected to
be 175. Because this amplifier is specified to
operate at 100Ω output load, a 50Ω resistor is placed
in series with its output. This resistor forms a
voltage divider with the 50 Ω output load and the
final voltage gain will be 87.5. A resistor network
can be used at the THS4303 output, as is show in
Fig. 2, if a round value of amplification about 50 is
desirable. Resistors R5 and R6 can be rounded in a
higher value without a significant performance
change.
Fig. 2. Resistive network rounding voltage gain
at x50 on 50 Ohm load.
3 Construction tips and operation
Some experimental prototypes were constructed in
order to verify the operation and to ascertain the
construction demands. The Printed Circuit Board
(PCB) was designed to use two layers for easy
construction. The signal path tracks that connect the
output of the input stage with the input of the second
stage (the connection between the two parts of the
ADL5566) are paralleled with cables to reduce the
tracks’ inductance. One of the most difficult
construction issues is to solder the 0402 size
capacitors for the signal paths. 0402 size capacitors
perform better than the 0603 at high frequencies and
their use simplifies the circuit. All the resistors must
be thin film. Small size is also preferred for the
resistors. To supply power to the circuit, a mini
USB connector was used. The supply voltage can be
applied from a 5V mobile phone charger with a mini
USB output plug. An AMS1117 low drop out
(LDO) regulator was used to convert the input
voltage to 3.3V that is required for a low noise
circuit operation. A prototype circuit photograph is
shown in Fig.3, where the dimensions of the final
construction can be seen. The bottom layer “L”
shape track is the heat sink of THS4303 that is
connected to its inverting input (pin 14). It is likely
that the thermal pad of this integrated Circuit (IC)
can be connected to the negative plane, but this is
not stated anywhere in the datasheet or the
application notes.
Fig. 3. Prototype circuit photograph
Some measurements of the prototype circuit
frequency response are shown in Fig. 7 and a
frequency response curve has been draw based on
the measurements that were taken. The frequency
response (-3dB) exceeds 1GHz when the
preamplifier is terminated on a 50 Ohm load.
4 Simulation Results
In order to prove the proper operation of the
proposed amplifier, and have a connection between
theoretical calculations and experimental results the
circuit should be modeled and simulated. This
simulation is somewhat challenging. The proposed
amplifier uses two integrated circuits from two
different manufacturers. Thus the simulation is
difficult in a single simulation environment.
ADL5566 has no Spice [10] model, but THS4303
has PSpice and Spice for TINA-TI models. TINA-
TI is a Spice-based simulator and for this reason is
capable of simulating complex circuits. In order to
use a Spice-based simulator the ADL5566 must be
modeled. The datasheet of the ADL5566 contains an
equivalent circuit and a frequency response that
shows an increment of amplification before the
upper cut-off frequency. Usually this type of
amplifiers have low loop-gain and their frequency
response is affected from two poles that are
relatively close to each-other. The simulation model
for the ADL5566 that was used is shown in Figure
4.
Fig. 4. The ADL5566 approximated Spice
model.
The frequency response and the behavior of the
ADL5566 equivalent circuit is close to those
described in the ADL5566 datasheet. The
simulation, using the TINA-TI, frequency response
of the ADL5566 is shown in Figure 5.
Using the previously described Spice model for the
ADL5566 the circuit can be modeled and simulated
without problems using the TINA-TI. Each one of
the capacitors that form the 101nF capacitor is
modeled as a series RLC circuit. This is not accurate
as the resistive part is not constant vs frequency,
because it is depends on the dielectric losses. The R
and L values were approximately extracted from
their datasheet frequency characteristics. These
values are shown on Table 1.
GRM155R71C
104KA88D
GRM1555C1
H470JA1D
C
100nF
47pF
R
20mΩ
200mΩ
L
0.4nH
0.024nH
Table 1. Capacitors RLC model values
Fig. 5. The ADL5566 Spice model frequency response.
Using the models and parameters and the default
TINA-TI model for the THS4303 several
simulations of the proposed amplifier circuit were
performed. The frequency response resulted from
the simulation is shown in Figure 6:
Frequency [Hz]
10.0MEG 100.0MEG 1.0G 10.0G
Gain [dB]
-10.0
0.0
10.0
20.0
Fig. 6. The proposed amplifier simulated frequency response.
The frequency response curve that flattens exactly
before the upper cut-off frequency was determined
after several trials to originate from the parallel
combination of capacitors that form the 101nF
capacitor. In simulations this capacitor combination
extends the upper cut-off frequency about 100MHz
compared to the single 100nF capacitor. The
simulated upper cut-off frequency is about 2.4GHz.
Although THS4303 bandwidth is only 1GHz, the
increment of the amplification close to 1GHz caused
by the ADL5566 extends the upper cut-off
frequency in simulations.
Fig. 7. Prototype circuit frequency response.
Frequency [Hz]
10.0k 100.0k 1.0MEG 10.0MEG 100.0MEG 1.0G 10.0G
Gain [dB]
0.0
10.0
20.0
30.0
40.0
Some measurements of the prototype circuit
frequency response are shown in Fig. 7 and a
frequency response curve has been plotted based on
the measurements performed. The upper cut-off
frequency (-3dB) slightly exceeds 1GHz when the
preamplifier is terminated on a 50 Ohm load. In
comparison to the simulated frequency response we
observe a small amplification decrement as the
Fig. 8. The proposed amplifier simulated frequency response using a 50Ω||5pF load.
frequency increases. In practice this is normal and it
is difficult effect to be modeled in a simulation in
these types of high frequency circuits. This is
affected by several factors such as capacitances on
the PCB, inductances of the tracks, inaccurate
models of the passive components and poles that are
missing in integrated circuits default models. These
factors are significant at high frequencies. In
practice the circuit upper cut-off frequency exceeds
the 1GHz and the circuit satisfies the purpose that it
was designed for.
Several simulation trials were made to determine the
cause of the simulated and measured frequency
response curves. It was found that the series 50Ω
resistor with the THS4303 output, limits the
amplifier frequency response forming an RL low
pass filter with the load capacitance. A small track
over a ground plane on a common double layer FR4
PCB forms a capacitance of about 0.6pF/cm. As
shown in Fig. 8, a 5pF of load capacitance limits the
simulated frequency response close to the frequency
response curve has been plotted based on the
measurements performed. This capacitance is close
to the measurement used setup.
5 Conclusion
The design of a high quality, compact 1GHz
preamplifier suitable for operation in conjunction
with micro channel plates (MCP) and silicon
Photomultipliers (SiPM) is described in this paper.
The 1GHz preamplifier comprised of two integrated
circuits. The minimization of the external
components ensures the ultra-low-noise operation
and the low-cost design.
References:
[1] Mauri G., Mariotti M., Casinini F., Sacchetti
F., C.Petrillo, “Development of pulse shape
analysis for noise reduction in Si-based neutron
detectors”, Nuclear Instruments and Methods
in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated
Equipment, Vol.910, 2018, pp. 184-193.
Frequency [Hz]
10.0k 100.0k 1.0MEG 10.0MEG 100.0MEG 1.0G 10.0G
Gain [dB]
-10.0
0.0
10.0
20.0
30.0
40.0
[2] Tosi A., Scarcella C., Boso G., Acerbi F.
“Gate-Free InGaAs/InP Single-Photon Detector
Working at Up to 100 Mcount/s”, IEEE
Photonics Journal ,Vol. 5, Issue 4, 2013, DOI:
10.1109/JPHOT.2013.2278526.
[3] Domienikan C., Costa P., Genezini F. and Zahn
G. “Low-cost Amplifier for Alpha Detection
with Photodiode”, International Nuclear
Atlantic Conference - INAC 2017, October 22-
27, 2017.
[4] Flaxer, E., "A low-cost, ultra-fast and low-
noise preamplifier for micro channel plates",
Measurement Science and Technology, Vol.17,
No.8, 2006, pp. N37-N40
[5] “Noise Specs Confusing?”, National
Semiconductor Application Note 104, May
1974
[6] lmh6554.pdf, “LMH6554 2.8-GHz Ultra Linear
Fully Differential Amplifier”, datasheet,
www.ti.com
[7] ADL5566.pdf, “4.5 GHz Ultrahigh Dynamic
Range, Dual Differential Amplifier”, datasheet,
www.analog.com
[8] MT-076.pdf, “Differential Driver Analysis”,
Tutorial, www.analog.com.
[9] AN-0990, “Terminating a Differential
Amplifier in Single-Ended Input Applications”,
Application Note, www.analog.com.
[10] , https://en.wikipedia.org/wiki/SPICE.
Creative Commons Attribution
License 4.0 (Attribution 4.0
International , CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
... Integrated inductors are essential components in modern RF circuits and have been studied in several recent works, [1], [2], [3], [4], [5]. In the early development of variable inductors, switchable inductors were the first design choice as a means of varying inductance values. ...
Article
This paper presents the design and simulation of two-variable MEMS inductors for high-frequency applications. In two cases, the inductors' cores are enclosed between two tunable metal plates. The medium permeability changes in response to iron plate movement, ensuring a high-tuning range above 250%. The first inductor is designed for the 10 GHz range, and the inductance value is comprised of Lmin = 3.37 nH and increases to Lmax = 11.8 nH. Similar inductance values are obtained for the second inductor, but it is designed to work at 20 GHz. Furthermore, the maximum value of the quality factor of the second inductor is slightly higher than the first one and reaches Qmax = 17.5.
Chapter
This chapter summarizes the experimental setup (such as targets and detectors), measuring, and data reduction procedures towards obtaining the angular distribution of accelerated ions elastically or inelastically scattered from targets. Different normalization methods for obtaining the absolute elastic scattering cross section and its ratio to Rutherford are discussed.
Article
The application of Si sensors coupled to Gd converters as thermal neutron counters is assessed in a series of test experiments on the neutron beam, among which scattering from standard samples. The prototype 1d detector is a Si microstrip sensor directly coupled to natural Gd converter and equipped with standard front-end electronics. The raw scattering data, collected by interfacing the detector with the data acquisition system available at the neutron source, show advantages and limits of this technology when applied to neutron detection. To improve the performances of the Si-based detector by means of an optimized discrimination of the neutron signals from noise and background radiation, a pulse shape analysis method is proposed. The effectiveness of this method is then explored by experimental tests on the neutron beam of two more prototype detectors, namely a PIN diode coupled to ¹⁵⁷Gd2O3 converters, and a Silicon photo-multiplier (SiPM) coupled to neutron scintillators. This study is aimed to real time applications and single event storage of the neutron information in time of flight instrumentation.
Article
Recently, there has been considerable effort to develop photon-counting detectors for the near-infrared wavelength range, but the main limitation is to have a practical detector with both high count rates and low noise. Here, we show a novel technique to operate InGaAs/InP single-photon avalanche diodes (SPADs) in a free-running equivalent mode at high count rate up to 100 Mcount/s. The photodetector is enabled with a 915-MHz sinusoidal gate signal that is kept unlocked with respect to the light stimulus, resulting in a free-running equivalent operation of the SPAD, with an afterpulsing probability below 0.3%, a photon detection efficiency value of 3% at 1550 nm, a temporal resolution of 150 ps, and a dark count rate below 2000 count/s. Such gate-free approach can be used to measure, at high count rate, signals in continuous wave or with slow time decays, where standard gated detectors would not be suitable.
Article
A compact, low-cost, ultra fast, low distortion and low noise preamplifier suitable for data-acquisition applications utilization micro channel plate (MCP) detectors has been built with a new monolithic integrated circuit amplifier in two stages. The gain, bandwidth and noise are optimized to the MCP performance in order to achieve the best rise time, response and stability with minimal jitter. The use of commercial components makes this instrument very low in cost and easy to build.
Low-cost Amplifier for Alpha Detection with Photodiode
  • C Domienikan
  • P Costa
  • F Genezini
  • G Zahn
Domienikan C., Costa P., Genezini F. and Zahn G. "Low-cost Amplifier for Alpha Detection with Photodiode", International Nuclear Atlantic Conference -INAC 2017, October 22-27, 2017.
Noise Specs Confusing?
"Noise Specs Confusing?", National Semiconductor Application Note 104, May 1974