RoentDek CFD8c User manual

RoentDek
HandelsGmbH
Supersonic Gas Jets
Detection Techniques
Data Acquisition Systems
Multifragment Imaging Systems
The RoentDek Constant Fraction Discriminators
CFD8c, CFD7x, CFD4c, CFD1c and CFD1x
(11.0.1701.1)

Page 2 of 25 The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1)
Mail Addresses:
Headquarter
RoentDek Handels GmbH
Im Vogelshaag 8
D-65779 Kelkheim-Ruppertshain
Germany
Frankfurt branch
RoentDek Handels GmbH
c/o Institut für Kernphysik
Max-von-Laue-Str. 1
D-60438 Frankfurt am Main
Germany
Web-Site:
www.roentdek.com
WEEE:
DE48573152
Product names used in this publication are for identification purposes only and may be trademarks of their respective companies.
All rights reserved. Technical changes may be made without prior notice. The figures are not binding.
We make no representations or warranties with respect to the accuracy or completeness of the contents of this publication

The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1) Page 3 of 25
Table of Contents
3B. THE CFD TIMING DISCRIMINATOR MODULES.........................................................................................6
3B.1 GENERAL DESCRIPTION .........................................................................................................................................6
3B.2 PRINCIPLES OF THE CFD CIRCUIT..........................................................................................................................8
3B.3 ADJUSTING THE CFD PARAMETER.........................................................................................................................9
3b.3.1 Threshold level............................................................................................................................................ 10
3b.3.2 CFD delay................................................................................................................................................... 11
3b.3.3 CFD fraction...............................................................................................................................................12
3b.3.4 Walk level....................................................................................................................................................13
3b.3.5 Output signal width.....................................................................................................................................14
3B.4 THE VETO OPTION ............................................................................................................................................... 14
3b.4.1 Accessing the jumpers in CFD1c / CFD1x.................................................................................................16
3b.4.2 Accessing the jumpers in CFD8c and CFD7x ............................................................................................16
3b.4.3 Accessing the jumpers in CFD4c................................................................................................................17
3B.5 PERFORMANCE LIMITATIONS ...............................................................................................................................18
3b.5.1 Temperature drift........................................................................................................................................18
3b.5.2 Pre-trigger problem.................................................................................................................................... 18
3b.5.3 Cross talk problem......................................................................................................................................19
3B.6 THE CFDX PULSE HEIGHT DETERMINATION OPTION ............................................................................................19
3B.7 THE BCFD VERSION OF READ-OUT OF BIPOLAR INPUT SIGNALS...........................................................................22
LIST OF FIGURES..........................................................................................................................................................24
LIST OF EQUATIONS....................................................................................................................................................25

Page 4 of 25 The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1)

Constant Fraction DiscriminatorsCFD8b, CFD4b and CFD1b (11.0.1701.1) Page 5 of 25

Page 6 of 25 The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1)
3b. The CFD timing discriminator Modules
The RoentDek CFD (Constant Fraction Discriminator) units can be used for improving the timing response of electronic
signals, e.g. as obtained from secondary electron multipliers like MCPs, Photomultipliers, Channeltrons or similar avalanche
counters.
Figure 3.1b: The CFD8 (here: CFD8b version), CFD7x similar
Especially for advanced RoentDek detectors with Hexanode (Hex75/Hex100) the combination of the RoentDek
FAMP amplifiers and these CFD units has considerable advantage over the RoentDek ATR19 modules in terms of pulse-
pair resolution and timing precision. Using the RoentDek CFDs is recommended for LC-delay-line anode (DLD40X or
RS-PMT25/40) and may also improve the performance of the RoentDek DLD or DET MCP detectors.
It has to be noted that any operational improvement over the easy-to-use ATR19 units requires a detailed understanding of
the CFD function and a proper setting of the CFD parameters. The RoentDek CFD units allow a precise adjustment of all
these parameters to different input signal properties (e.g. rise time) and their control by two types of monitor outputs for
achieving optimal timing results. Following the description here is mandatory for successful use.
This manual describes the latest product series CFD8c/4c/1c and CFD1x/7x.which are part of the RoentDek
FEE2/5/7x/8frontend electronics assemblies. Units with earlier version numbers (b or a) are very similar and this manual
may serve as a reference, too. However, the older product versions have fewer features (e.g. no veto option) and can differ in
power requirements, positions of control elements and output options. You may request an older manual version from
RoentDek for those devices.
All CFD units (except for the “x” channels) can be equipped with boards to operate bipolar signals as from the RoentDek
bFAMP modules. If you have received CFD modules containing such boards please review first Chapters 3b.1 to 3b.3 and
then refer to Chapter 3b.7 to learn about the difference between the standard CFD and the RoentDek bCFD settings.
3b.1General description
If a source produces signals with different pulse heights, a simple comparator circuit or leading-edge discriminator unit will
produce a digital output signal with a time jitter on the order of the input signal’s rise time, thus limiting the achievable timing
precision.
The CFD circuit was designed to deliver a “digital” output signal (e.g. NIM) with timing properties almost independent from
the pulse height for all signals above a selectable threshold level. If the CFD is properly adjusted to the input signal’s
properties, the jitter of this digital “timing output” signal can be reduced by at least a factor of 100 compared to leading-edge
discrimination, as long as the pulses are fairly above noise and the normalized input signal width (the FWHM) does not vary as
function of pulse height. The RoentDek CFD units have a bandwidth of about 300 MHz, require negative input signals
on the In socket and provide 3 such timing outputs as NIM signals from sockets labelled NIM-Out on the front panel. Some
units are available with additional ECL outputs on request or TTL output (CFD1c only). The input signal pulse height should
not exceed -2 V and there is no good temporal resolution to expect for signals smaller than -100 mV. If signals or levels >
2 V or smaller -2 V are connected to the CFD input the circuit can be damaged. For operation with signals beyond the
safe range RoentDek can supply passive attenuators or inverter plugs (pAtt and pInv).
RoentDek offers three different case versions hosting 8(7), 4 or 1 CFD channels, some with pulse height determination
option (CFDx) on one of the channels.
The CFD8c (see Figure 3.1b) is a standalone unit with 8 independent channels. In the version CFD7x one channel is replaced
by a circuit allowing for a pulse height determination on one channel. These units are the recommended versions for operating
RoentDek Hexanode Delay-line detectors. The CFD8c/7x are standalone modules for 19” racks (one height unit) and

The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1) Page 7 of 25
come with an external mains adapter for 100-250 V AC (50-60 Hz). The power consumption is
max. 3.7 A at 12 V DC (<45 W). The case size is (approx.): 485 mm x 45 mm x 375 mm (width x
height x depth incl. power connector), weighing: 2.5 kg (without 12 V power adapter)
The single-channel units CFD1c and CFD1x (version with pulse height determination circuit) come
in a stand-alone 3HU case (W61 mm/L129 mm/H232 mm, weight 0.8 kg) with an external
12 V DC mains adapter for use with 100-250 V AC sockets. The power consumption is max. 0.5 A
at 12 V DC (6 W) or max. 0.7 A at 12 V DC (<10 W) for the CFD1x.
The CFD4c has the same functions as the CFD8c for 4 channel operation in a 1/12 NIM cassette
(W34 mm/L280 mm/H220 mm, weight 1.2 kg) and requires a NIM-bin for operation: The current
drawn is max. 0.2 A from the 6 V line and -1.8 A from - 6 V*.
Figure 3.2b: Photos of the CFD4 and CFD1 series cases (here: ‘b’ versions) and the SPS3 power supply
*The CFD4c can optionally be ordered with a 9-pin sub D input for external +6 V/-6 V/-5.2 V powering via the SPS3 mains
adapter. The SPS3 is also required for the older CFD1b model.

Page 8 of 25 The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1)
3b.2Principles of the CFD circuit
It is important to note that an optimal operation of the RoentDek CFD units requires optimal settings of the parameters
•Threshold level (Th)
•Walk level (Z)
•CFD delay
•CFD fraction
with respect to the input signal properties. This will be explained in the following.
Figure 3.3b: Front panel inputs, test points and control potentiometers , not shown is the veto input
and the
CFD fraction potentiometer being on the front panel for c and x versions (see Figure 3.14)
If a signal as shown in Figure 3.4b (left)*is turned into a bipolar signal (right) by an appropriate electronic chain it can be
shown that the zero-crossing of the bipolar signal does not jitter in time as function of the input pulse height.
The CFD circuit consists of two parts:
The “ analogue ” circuit produces this bipolar signal from a unipolar input signal: A resistor-divider splits the signal into two
with different relative pulse heights. The ratio of the pulse heights (
CFD fraction ) is determined by the choice of fix resistors
or a potentiometer (as in case of the RoentDek CFDs). While one of the signals is inverted later on, the other experiences a
certain delay (
CFD delay ) before both signals are superimposed to form the bipolar signal. On the RoentDek CFDs, this
delay is formed by an external coax cable (LEMO 00 series) connected between the front-panel sockets labelled Delay. The
bipolar signal can be observed on the analogue walk monitor output socket labelled Ma(damped by a factor of 10). The CFD
fraction can be adjusted via a potentiometer on the front panel side panel (CFD4b: side panel, CFD1b and CFD8b: on the lid).
Figure 3.4b: Schematics of a CFD analogue chain
Figure 3.6b shows the digital circuit chain of two comparators and a logical AND-gate: One comparator switches to “high”
when the input signal exceeds a certain (adjustable) reference value (the
Threshold level) and produces a norm signal (this is
how a simple “leading-edge” discriminator unit operates). When the input signal level is below the Threshold level, its output
returns to “low”. Ideally, the Threshold level is set above electronic noise but low enough to register even the smallest valid input
signals.
*Note that for RoentDek CFDs the input signal polarity is negative

The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1) Page 9 of 25
A second so-called “walk comparator” in the circuit operates in the
same way with the bipolar signal. Its reference, the
Walk level , shall
be set very near and slightly above the baseline level of the input
signal so that it “triggers in the noise”. Then the bipolar signal
produces a distinct series of transitions as the noise fluctuates around
this level (see Figure 3.6b). In presence of a significant input signal a
certain transition from “low” to “high” represents the zero-crossing
of the bipolar signal (the “timing transition”) which shall serve as the
accurate timing output of the CFD circuit. This signal is produced by
a logic AND-gate of the two comparator outputs.
The RoentDek CFDs allow monitoring the output of the walk
comparator via the digital monitor output socket labelled Md.*
The Threshold level and the Walk level can be set via two potentiometers labelled Th and Z (Z standing for Zero Crossing Level)
with corresponding test points to measure the actual levels with a voltmeter. A ground reference point labelled GND is
provided on the front panel of most units. The width of the CFD timing output signals is adjustable via a potentiometer
labelled Won the front panel. The width can be increased by a factor of 10 by placing a jumper on the dedicated position
through a hole in the side panel (CFD8/7x: top lid) for each channel individually.
Note that the test points can only give coarse information about the setting (within about ±5 mV). For optimal results and
adjustment, the signals have to be verified on an oscilloscope. The CFD Fraction can be adjusted by another potentiometer
labelled Fr on the front panel (CFD4b, CFD1b: side panel, CFD8b: top lid (see Figure 3.6b). Figure 3.6b also shows the
position of the jumper terminal for increasing the timing signal output width
The RoentDek CFD is useful for signals with rise/fall time > 1 ns. If the input signal source has less than 3 ns FWHM the
CFD might not operate optimally even with small CFD delay and optimized settings of the fraction (see below). The input
signal heights should be kept between 50 mV and -2 V (on 50 Ω impedance) by selecting a proper gain on the given amplifier.
The input should not be exposed to signals higher than -3 V or DC levels higher than ±1 V.
The oscilloscope traces in the next chapter are shown as examples for optimal parameter settings and the effect of changing a
certain parameter.
3b.3Adjusting the CFD parameter
Figure 3.7b shows all relevant signals with a near-optimal adjustment of the parameters Threshold level, Walk level, CFD delay and
CFD fraction for a given input signal of about 5 ns rise time (upper trace). The traces trigger on the negative slope of the lowest
trace, which is the CFD timing output. The analogue and digital walk monitor outputs are displayed on the second and third
trace. The CFD fraction was set to 0.35 (default) and cable produces 4 ns external delay which results in an effective CFD delay
of about 4.5 ns, i.e. sightly increased by internal delays on the circuit board).
*Note the Mdoutput has a DC offset of about -0.1 V
Figure 3.6b: Schematics of the
CFD digital circuit chain
Figure 3.5b: top/side panel with fraction
potentiometer only versions a/b) and optional
jumper terminal for increasing the CFD output
width by a factor of 10 (optional)

Page 10 of 25 The Constant Fraction Discriminators CFD8c, CFD7x, CFD4c, CFD1c and CFD1x (11.0.1701.1)
Figure 3.7b: Oscilloscope traces of in- and output signals of the CFD, traces 1-4 from top: input signal, analogue
monitor output signal, digital monitor output signal, CFD timing output signal (which triggers all traces). The red
arrows indicate the
timing transition which appears slightly shifted in the different traces due to internal delays.
This and the following images of signal traces have been obtained with a 300 MHz analogue oscilloscope (a large
number of signals with varying pulse heights are superimposed). Similar views can be obtained from digital
oscilloscopes with adequate “persistence” setting
The Threshold level (see below) was set so that even the smaller signals from the pulse height distribution are registered. This can
be verified both on the first trace (input signal) and on the analogue monitor output (trace 2). The input signal cannot always
be displayed together with the CFD outputs unless the signal source (e.g. amplifier) has two independent outputs sockets.
Therefore the trace of the signal input was omitted on the next figures. The analogue walk monitor may serve as well for
verifying the threshold level but it is to note that the analogue walk monitor signal height is damped by a factor of 10.
For achieving the ideal settings it is first of all important to note that the CFD operates best (in all respects) and yields the
optimal temporal resolution if the input signals are as high as possible but only up to -2 V. The threshold should not be set
smaller than 50 mV for the multi-channel versions because then there is a risk of inter-channel cross talk (see Chapter 3b.5.3).
Please note that it is preferable to obtain such high signals rather by a low amplifier gain but high “raw” signal from the given
detector (e.g. high gain on the physical electron multiplier MCP, PMT, Channeltron …) if one has this choice. This can for
example be achieved by increasing the bias voltage on the physical electron multiplier.
For a given electronic noise level, signal-to-noise ratio can only be improved by increasing the gain of the primary charge input
from the detector source to the electronics. Only then signals have optimal signal-to-noise ratio which will finally determine
achievable resolution.
3b.3.1 Threshold level
The Threshold level shall be set just above the input signal line’s noise level. This can be verified by observing the CFD output
simultaneously with the analogue walk output on an oscilloscope as shown in Figure 3.8b.
If the threshold is set too high, small (but valid) signals are not registered (see Figure 3.8b, left picture), if it is set too low (right
picture), the CFD output will also trigger on spurious noise components and produce background counts.