4.3 Validation and Testing
121
4.3.5 System Integration Tests
To integrate the SAMPA with the rest of the front-end and readout electronics, the
TPC, MCH, STAR, and MPD have all designed prototypes of their final front-end
cards, as shown in Fig. 4.12. They have also performed tests of their system to verify
that the new front-end cards with SAMPA devices operate as they require them too
for the final installation. This includes verifying signal and noise integrity, the proper
operation of the slow control, the data readout, triggering and any other parts of the
functionality that they require in their system.
The TPC front-end card contains five SAMPA devices running in direct readout
mode and connected to two GBTx devices which are further connected to a CRU
through optical links. For the TPC, their tests have confirmed that the device operates
reliably in combination with the GBTx and GBT-SCA. This includes confirming
that the I
2 C communication with the GBT-SCA works at full speed with several
SAMPAs on the same bus, the multi-drop clock and trigger distribution are operating
satisfactorily as well as the GBTx’s ability to sample the transmitted signals reliably.
Several front-end cards have also been tested when connected to a small TPC chamber
in a beam test and has been shown to work reliably in this setting as well. Some
optimization has been applied in v3 for the direct readout mode to simplify and
make the synchronization phase on start-up more reliable. The primary changes are
related to the SAMPAs on/off control of the SLVS drivers when the direct readout
mode is enabled, in addition to optimizations in the clock-to-output delay and skew.
The MCH front-end cards are relative small boards, as seen in Fig. 4.12, containing
two SAMPA devices. Up to five boards are connected to a long flex-cable connected
to a remote concentrator card containing the GBTx and GBT-SCA devices. The
testing of the MCH front-end cards proved that the inter-device communication in
the daisy chaining functionality works, as well as the communication with the GBTx.
It also confirms that it is possible to mount and route the device on boards that are no
bigger than their previous boards without greatly affecting noise or signal integrity.
Resulting from the MCH testing some optimizations for v3 has been done to avoid
packet loss in the daisy chaining link at high occupancies, as well as an issue where
a soft reset through the I
2 C link would not return an acknowledge due to the device
being in process of resetting.
During testing of the I
2 C communication between SAMPA and GBT-SCA devices
over the long cables for the MCH on the v3 device, issues were discovered with missing acknowledgement of sent information. Due to the cable having a high capacitance
of 400 pF and termination not being present, the received signals were misinterpreted
by the SAMPA. The termination issues, which sometimes caused a too early response
or missing acknowledge due to reflections, was mitigated by adding ferrites to the
line and configuring the GBT-SCA to directly drive the SCL line instead of using
external pull-up resistors. Subsequently, parts of the issue were still shown to be
present where the design was found to sometimes completely stop responding and
recovery was only possible by resetting the device. Due to slow rising times caused
by the high capacitance of the cable and due to the SAMPA not having Schmitt-
121
4.3.5 System Integration Tests
To integrate the SAMPA with the rest of the front-end and readout electronics, the
TPC, MCH, STAR, and MPD have all designed prototypes of their final front-end
cards, as shown in Fig. 4.12. They have also performed tests of their system to verify
that the new front-end cards with SAMPA devices operate as they require them too
for the final installation. This includes verifying signal and noise integrity, the proper
operation of the slow control, the data readout, triggering and any other parts of the
functionality that they require in their system.
The TPC front-end card contains five SAMPA devices running in direct readout
mode and connected to two GBTx devices which are further connected to a CRU
through optical links. For the TPC, their tests have confirmed that the device operates
reliably in combination with the GBTx and GBT-SCA. This includes confirming
that the I
2 C communication with the GBT-SCA works at full speed with several
SAMPAs on the same bus, the multi-drop clock and trigger distribution are operating
satisfactorily as well as the GBTx’s ability to sample the transmitted signals reliably.
Several front-end cards have also been tested when connected to a small TPC chamber
in a beam test and has been shown to work reliably in this setting as well. Some
optimization has been applied in v3 for the direct readout mode to simplify and
make the synchronization phase on start-up more reliable. The primary changes are
related to the SAMPAs on/off control of the SLVS drivers when the direct readout
mode is enabled, in addition to optimizations in the clock-to-output delay and skew.
The MCH front-end cards are relative small boards, as seen in Fig. 4.12, containing
two SAMPA devices. Up to five boards are connected to a long flex-cable connected
to a remote concentrator card containing the GBTx and GBT-SCA devices. The
testing of the MCH front-end cards proved that the inter-device communication in
the daisy chaining functionality works, as well as the communication with the GBTx.
It also confirms that it is possible to mount and route the device on boards that are no
bigger than their previous boards without greatly affecting noise or signal integrity.
Resulting from the MCH testing some optimizations for v3 has been done to avoid
packet loss in the daisy chaining link at high occupancies, as well as an issue where
a soft reset through the I
2 C link would not return an acknowledge due to the device
being in process of resetting.
During testing of the I
2 C communication between SAMPA and GBT-SCA devices
over the long cables for the MCH on the v3 device, issues were discovered with missing acknowledgement of sent information. Due to the cable having a high capacitance
of 400 pF and termination not being present, the received signals were misinterpreted
by the SAMPA. The termination issues, which sometimes caused a too early response
or missing acknowledge due to reflections, was mitigated by adding ferrites to the
line and configuring the GBT-SCA to directly drive the SCL line instead of using
external pull-up resistors. Subsequently, parts of the issue were still shown to be
present where the design was found to sometimes completely stop responding and
recovery was only possible by resetting the device. Due to slow rising times caused
by the high capacitance of the cable and due to the SAMPA not having Schmitt-