Camilla Galloni (9168d12a) at 28 Mar 16:48
Update instructions for applying or not the noisy channel mask in t...
This MR is a follow-up on issues recently discovered in the DCS due to the publication of NaN
values in the (DAQ) DIM services. Since the very first implementation, the DCS always had issues dealing with the large amounts of data produced by the DAQ. This MR takes a different approach consisting in the large overhaul of the DIM publication in order to provide pre-processed data.
The main change consits in the following: For each one of the metrics to be exported, up to N samples are recorded (by default 20 samples taken ~1s apart). The DIM publications are updated at regular intervals (by default every ~15s) taking the average of the recorded history, ignoring invalid values (i.e. NaN or no readout). Additionally, a dead band w.r.t. the currently published is applied in order to smooth out small variations (default to 0.5° for the temperatures and 0.05V for the voltages).
Other changes are technicalities to reach the end results. Also note the improvement on the GE1/1 temperature readings thanks to the better usage of the precision resistor.
The initial version has been deployed in production at P5 for ~1 week. The latest version has been carefully tested on the GE1/1 setup in the 904 integration area.
This MR aims to enhance further the Online Software's compatibility with ME0. In particular, all features for a successful test beam with a CTP7 backend should be implemented.
Also note that the supported hardware versions have been bumped to the ASIAGO and GEB production series. Prototypes (ASIAGO v1) have never been supported, ASIAGO v2 Pre-Production should work in degraded mode (e.g. missing environmental sensors or missing slave LpGBT reset), and ASIAGO v2 Production, including the Pre-Series, are fully supported.
Tested on the prototype/TB ME0 stack with a CTP7 backend, including the initialization of a second GE2/1 CTP7 for the tracking chamber.
This MR is at first a follow-up on !309 which proved to be useful in the initialization of the trigger links (but less in the programming of the GE1/1 OptoHybrid FPGA). In more detail, the following changes are implemented:
Code builds for all stations; original code from !309 tested for weeks/months in production at P5; modified code tested on the B904 integration area setup.
The title should speak for itself.
More than once, we got tricked because of a poor choice of default GBTx RX phase on GE2/1 when all phases were reported as good by the GBTx RX phase scan. This MR aims at fixing this issue once and for all.
The choice of those default phases is based on the aggregation of GE2/1 QC7 results with at least 5 modules per module type. The actual selection was made following the standard phase selection algorithm. In case of ambiguity in the selection of the best phase, the phase closest to the highest number was chosen (ideally, the phase most frequently good should be chosen, but it isn't immediately available). Also, note how some VFAT are always reported as bad on the M7 module type -- this is likely coming from a failed scan/module.
Not strictly required as long as the code compiles... Time will tell if those phases were the best ones.
Camilla Galloni (80b4e55c) at 13 Mar 15:46
Linked to a broken DAC circuit
Camilla Galloni (6f2ec1cf) at 11 Mar 17:43
Add option to not apply channel masking in physics runs
This MR aims at adding a button to remove the channel masking (typically the noisy channel mask) in physics runs. Typically the low threshold run. Tested in the integration side setup.
On the integration site.
Camilla Galloni (ddc7e086) at 11 Mar 17:26
Add option to not apply channel masking in physics runs
SBit lines all masked on 04/03/24
Bad S-bit line on OH10 - VFAT1 - TU5
TRIG.CTRL.TU_MASK.VFAT1_TU_MASK 32
Masked.
Confirmed on 04/03/24.
A second check showed that the sbit line is broken again. Re-masking it.
On 04/03/24
Bad S-bit line on OH6 - VFAT4 - TU0 Bad S-bit line on OH6 - VFAT4 - TU2 TRIG.CTRL.TU_MASK.VFAT4_TU_MASK 5
Confirmed on 04/03/24. Continues to be masked.