A picture gallery of the "Level-1
Trigger" electronics, as it was in the ALEPH experiment. (links hereunder point to a picture of the item;
use the browser's "back" to return here)
Our institute (then called IHEP) was responsible for the design, construction
and operation of the Level-1 trigger in the ALEPH Experiment at LEP. The
system consisted of appr. 130 FastBus-size printed-circuit boards, about
half of which followed the FastBus form-factor for reasons of signal input/output
("analog boards") while the other half were "real" FastBus boards with
data-access.
The Level-1 Trigger system, as it was installed in the electronics
"barracks" right next to the detector, was primarily based on information
from the ALEPH calorimetry -- electromagnetic and hadronic. In addition,
the Inner Tracking Chamber (ITC) delivered Trigger information fast enough
to be used in Level-1. Hit-combinations forming track-candidates were found
in its own processor and delivered as a pattern in r-phi-z coordinates
to the Level-1 system.
The energy-sampling in the calorimeters was used in four different ways
for the purpose of triggering :
projective trigger towers (72 in total distributed in azimuthal angle and
pseudo-rapidity) were formed from the cathode signals of the electro-magnetic
calorimeter on Mixer Boards (ECTmix). Their
assembly
in the experiment required a FastBus-size crate with some wiring
on the back-plane for energy-summation and other purposes.
anode signals from wire-planes in the electro-magnetic calorimeters yielded
signals of high sensitivity (reflected in low energy thresholds due to
a very favorable signal-to-noise ratio). The granularity was naturally
limited to the physical size of calorimeter modules (barrel, end-caps).
Furthermore, sets of odd/even planes were used in coincidence hardening
the signal even more against noise-contributions (ECWmix).
to capture purely hadronic energy-deposits, projective trigger towers matching
up with the electro-magnetic ones were formed in Mixers (HCTmix)
designed for this calorimeter-type.
signals from "strips" on the High-Voltage side were merged for detector-"planes".
Each "plane" contributed to an analog sum. Thus, "plane"-signals could
be used to detect particle-penetration in the hadronic calorimeter (HCWmix)
- an important signature for the trigger on muons. Also here, the granularity
was limited by the "modularity" of the detector (barrel, end-caps).
All analog trigger signals from the Mixers require "discrimination" before
they can be used in any "decision logic". This was achieved in a "trigger
universal" discriminator module (DIS). The
module applies FOUR different thresholds to each detector item above. Thresholds
can be chosen to match with constraints applied in a trigger-condition
later on. It is apparent, that all modules had to exist in larger numbers.
Before these were produced, prototypes of all were built in house (Example:
DIS-prototype's Component- and
Solder-Side).
Control of the "analog electronics" described above was achieved through
an interface called the ControlBusInterface (CBI).
It is a uni-directional device to load data from the experiment's FastBus
system (e.g. voltage-offsets for analog input signals, thresholds for banks
of discrminiators) into DACs (DigitalAnalogConverters) located on the "analog
printed-circuit" boards.
Following the stage of "discrimination" digital trigger patterns (Discriminator
patterns, track-candidate patterns from the ITC - all 72 bit wide) were
available for the actual decision making process.
For this purpose the information was first transferred to TriggerSegmentRegisters
(TSR), where some detector-specific re-ordering
is performed on plug-on daughterboards as well as a reduction of the width
of the pattern to 60 bit by application of an "OR" to the 2 *12 most "central"
trigger segments. Most important, however, was the fact that the patterns
were made available to data-readout through FastBus. Availability of these
data was essential for monitoring the performance of the system on-line
and for verification of the decision-making process in quantitative off-line
analyses. The TSR distributed the trigger patterns on a backplane-fragment
to neighboring modules (TSS, TSD, ECLfo). [1]
As data read-out only occured upon acception of an "event" by the Level-1
trigger and the following higher level decisions, information on all the
non-accepted particle collisions would have been lost. To monitor signal-occurences
for these collisions at least on a statistical level, the TriggerSegmentScaler
(TSS) was built. The scaler module covers
64 channels with 16 bit counting depth. It recorded occurences of each
trigger bit in the respective pattern until read-out fetched the data and
cleared the scaler-content. [2]
An auxilliary module helped for visual inspection of the running system.
The bit patterns were displayed on LED-arrays in the TriggerSegmentDisplay
(TSD), e.g. "stuck" bits from faulty input
channels could be easily identified. [3]
The bit-patterns were distributed (in multiple copies, if needed) to e.g.
coincidence-logic as differential signals by a Fanout-module (ECLfo). [4]
Actual trigger decisions were made using "commercial" electronics
housed in CAMAC-crates. Correlations
constituting "physics triggers" were implemented here.
The one-bit result from each "physics trigger" was transmitted to a TriggerPatternRegister
(TPR), which also derived the final Level-1
decision. Furthermore, the TPR was also the place, where the Level-2 trigger
decision of ALEPH was taken in. The mechanism was a relatively simple one,
because it was only dealing with track-detectors (calorimetry information
was not updated in Level-2):
The track-pattern found in the ITC within 3-5 microseconds (Level-1
time-scale) was replaced by a pattern derived from the TPC by a dedicated
processor after drift and processing time (70-80 microseconds = Level-2
time-scale).
The ALEPH "Trigger Supervisor", built by CEA Saclay, picked up the pattern
of "physics triggers", derived the final Yes/No decision for Level-1 /
Level-2 and dealt with protocol signals to steer Data-Acquisition.
Seen from the DAQ point-of-view, the trigger is yet another subdetector
of the experiment and has as such to follow the normal DAQ protocol. This
task was managed in the TriggerProtocolUint (TPU), of which two versions
existed. The one of the early days (89-94)
was replaced by a newer "twin"-module (94-00)
after upgrading of DAQ electronics.
Data produced in the trigger system itself were also read-out for purposes
mentioned above. This was the task of software running in a standard ALEPH
Event-builder (AEB with its associated
memory) housed in racks [ 1,2,3
] near the trigger electronics.
