Minimization of Electronic Noise
The magnitude of the equivalent noise charge (ENC) of charge sensitive
preamplifiers is affected by a number of different factors. Although these
factors produce a certain unavoidable level of noise, additional and
preventable noise may also be introduced into the detection system by
some aspects of the circuit design. The purpose of this section is to
help the engineer design detector circuits
having the minimum electronic noise
possible, given the constraints of the
detector.
One of the more avoidable noise sources
which may be present is inductive 'pick-up'
from nearby circuitry. This can generally
be eliminated by adequately shielding the
detection circuitry and by avoiding 'ground
loops' in the layout of the circuitry. The
power supply may also contain 'ripple' that
will not be completely rejected by the
amplification circuitry. For this reason, it is advisable to RC filter both the
positive and negative power supply lines at a point close to the preamplifiers.
In addition to ensuring a quiet power supply, it is also important to RC filter
the detector bias supply at a point near the detector and preamplifier. The
types of noise described in this paragraph can be identified by their periodic
behavior. With careful circuit design, these noise sources can be eliminated
as significant factors affecting the performance of the circuitry.
One final note on providing clean power supply voltages
: Surprisingly,
some power supply regulation ICs (such as the LM317 and LM337) produce
outputs that are very noisy. This noise can couple to the preamplifier
output, producing unsatisfactory results. If these regulator chips are used to
provide supply voltages for the CR-110, it is recommended that an RC filter
combination of 4.7 /1000 F be used to filter both the positive and negative
power supplies. Alternatively, a quieter regulator circuit (such as that used
in the CR-150-X evaluation boards) could be used. See
http://www.cremat.com/CR-150.pdf for more information on this regulator
circuit.
In typical detection systems using charge sensitive preamplifiers, the ENC
(equivalent noise charge) is due to a combination of 5 factors:
1). The
series thermal noise
of the input JFET in the preamplifier (which is
proportional to the total capacitance to ground at the input node),
2). The
parallel thermal noise
of the feedback resistor and any 'biasing'
resistor attached to the detector,
3). The
shot noise
of the detector leakage current,
4). The
series 1/f noise
, which is produced by the electrical contacts of the
detector and preamplifier input JFET,
5). The
parallel f noise
caused by the proximity of lossy dielectric material
near the preamplifier input node.
These noise sources can often be individually quantified in an operating
detection system by measuring the dependence of the ENC on the "shaping
time" of the pulse amplifier which usually follows the preamplification stage.
This method is described in more detail in the article:
Bertuccio G; Pullia A; "A Method for the Determination of the Noise
Parameters in Preamplifying Systems for Semiconductor Radiation
Detectors", Rev. Sci. Instrum., 64, p.3294, (1993).
Other articles which describe typical noise sources and signal processing
techniques when using charge sensitive preamplifiers are:
Radeka V; "Low-Noise Techniques in Detectors", Ann. Rev. Nucl. Part.
Sci., 38, p.217, (1988).
Goulding FS; Landis DA; "Signal Processing for Semiconductor
Detectors", IEEE Trans. Nuc. Sci., NS-29, p.1125, (1982).
In the interest of avoiding unnecessary noise, there are a few factors
requiring attention. If AC coupling is used, an important decision to make is
the value of the "bias resistor" (resistor placed between the detector and
the filtered detector bias supply). Because this resistor is effectively "in
parallel" with the preamplifier input, it is a source of
parallel thermal noise
.
The magnitude of this noise is proportional to the reciprocal of the square
root of the resistor value. To choose a good value for this resistor, one
should have approximate knowledge of the detector leakage current. It
should be noted that the thermal noise of the bias resistor has the same
power spectrum as the
shot noise
produced by the detector leakage current.
To keep the bias resistor from being a significant source of noise, one
should choose a bias resistance that keeps the thermal noise of the bias
resistor significantly less than the detector shot noise. The point at which
the thermal noise of the resistor equals this shot noise is when the bias
resistor voltage drop is =2kT/q, or approximately 50 mV. If the voltage drop
is significantly greater than this, then you can be certain that the thermal
noise of the resistor is not limiting the performance of the circuit. To be
safe, you can use a bias resistor that will drop approximately half a volt.
Because the CR-110 uses a 100 M feedback resistor (which produces its
own thermal noise) there is no need to increase the value of the bias
resistor higher than approximately 200 M . Another consideration in the
choice of bias resistor is that a very large voltage drop across it (in excess
of several volts) may significantly subtract from the voltage drop across the
detector.
Another source of electronic noise is the thermal noise of resistances
effectively "in series" between the detector and the preamplifier input. The
thermal noise voltage that the effective series resistance produces is
converted to a "noise charge" (remember that the preamplifier output is
proportional to the charge flowing into the input) which is proportional to the
capacitance to ground at the preamplifier input. Of course it is
recommended that the circuitry minimize the series resistance between the
preamplifier input and the detector, and usually this resistance can be
reduced to a figure of less than a few ohms. Unfortunately, effective
resistance in the input stage of the preamplifier add to this figure, making it
the dominant source of the "series thermal noise". As mentioned, this
noise component is proportional to the capacitance at the preamplifier input,
and for this reason it is important to seek to minimize the input capacitance
as much as possible. Using even short sections of coaxial cable to connect
a detector to the preamplifier, for example, can significantly degrade
the noise performance. Assuming a shaping time of 1 s, this noise
component adds 3 electrons RMS of noise charge for each pF of
capacitance added to the input.
Another noise concern in the design of your detection system is the
introduction of
parallel f noise
, which is introduced by the proximity of lossy
dielectric materials at the preamplifier input. To minimize this source
of noise, which in some situations can be quite significant in magnitude,
detector circuit designs should keep the input traces on the circuit board as
short as possible. This is because the circuit board itself is often the lossy
dielectric material introducing this form of noise. Epoxy and glass, which
are usually considered to be good dielectrics (and circuit board materials) in
most circuit applications, are actually too lossy to be used in the usual
manner when designing detector circuits. Better construction materials are
Teflon and to a lesser extent alumina. These materials, however, are more
unusual and expensive than standard FR-4. To avoid the expense of Teflon
boards, consider lifting the input lines off the circuit board in some fashion,
perhaps by suspending the input lines above the board using Teflon
standoffs. If electronic noise is not a primary consideration, however, it
may suffice to use short traces on an epoxy-based circuit board. The use of
coaxial cable to couple the detector to the preamplifier may introduce noise,
not only by adding capacitance (as mentioned previously), but also because
of the lossiness of the cable's dielectric layer. If coaxial cable absolutely
must be used between the detector and preamplifier, its length should be
as short as possible.
Estimating the Electronic Noise in a Detection System
It is often useful to know what the expected electronic noise will be in a
detection system while the system is still in the design phase.
The following equation can be used to estimate the noise level in a detection
system based on the CR-110 charge sensitive preamplifier. Estimates have
been made for factors (d) and (e) mentioned previously, assuming short
traces on an FR-4 circuit board (such as those found on Cremat's CR-150-
AC-C evaluation board). This equation may be useful in allowing the user to
calculate the optimal shaping time ( in s) minimizing the electronic noise
(ENC in electrons rms) for a given detector capacitance (Cin in pF) and
detector leakage current (Idin pA).
Frequently Asked Questions
What are charge sensitive preamplifiers?
Charge sensitive preamplifiers were developed to detect the total amount of
charge flowing from a detector as the result of a 'pulse' event, such as the
detection of individual particles or gamma-rays. The preamplifiers integrate
the pulse of current flowing from the detector over time (by virtue of a small
Figure 5. Ground loops, caused
by multiple paths to ground, can
make a detection system
sensitive to external RF.
det.
