Atlas SCT Test DAQ Release Notes


Contents

Tagged Versions

The following versions have been tagged in the CVS repository and may be retrieved as follows:


BJG/PWP:2002-11-05 Version 3.30 Release Notes

This PRODUCTION version is built to work with version 3.02/07 of ROOT (WITHOUT experimental GDK driver). It may also be built under Linux to run with ROOT version 3.02/07.

To use this version under Windows:

The built versions will work for the following system configuration, with all VME boards set to the usual addresses:

Users with non-standard file system paths or VME configurations will need to edit the appropriate source files and recompile as usual. (For a system without a d: disk, a user with administrative priviliges may simply mount the c: disk as drive d.)

Additions/fixes/improvements:


BJG/PWP:2002-06-05 Version 3.20 Release Notes

This DEVELOPMENT version is built to work with version 3.02/07 of ROOT (WITHOUT experimental GDK driver). It may also be built under Linux to run with ROOT version 3.02/07.

Additions/fixes/improvements:


PWP:2002-04-08 Version 3.14 Release Notes

Additions/fixes/improvements:


PWP:2002-03-21 Version 3.13 Release Notes

Additions/fixes/improvements:


PWP:2002-02-12 Version 3.12 Release Notes

This DEVELOPMENT version was the first built to work with version 3.02/07 of ROOT (WITHOUT experimental GDK driver). This version was found to be reasonably stable for users of both Windows NT and Windows 2000 operating systems.

Additions/fixes/improvements:


PWP:2002-01-18 Version 3.11 Release Notes

This DEVELOPMENT version is built to work with the current production version of ROOT, version 3.02/07 (WITHOUT experimental GDK driver).

After an extensive period of testing, involving fixes to several memory leaks which had crept into stdll and the test macros, this version appears to be stable. We aim to elevate this version, or a development of it, to PRODUCTION status in the near future - so we would appreciate user feedback now. This is particularly relevant to those users of Microsoft Windows 2000 who have reported stability problems with recent versions of the code.

Additions/fixes/improvements:


PWP:2001-08-31 Release Notes

This is a major release with many new features in addition to the usual cluster of bugfixes. The most important new features are listed below:

The electrical test macros are in accordance with version 3.01 of the document Electrical Tests of SCT Hybrids and Modules.

Two built files are provided, for users with and without CLOAC, which support up to six modules connected to one MuSTARD, one SLOG, 0-3 SCTLV cards of any flavour (except T7 specials) and 0-2 SCTHV cards.
PWP:2001-03-08 Release Notes

This release includes a few minor revisions to bring the documentation more up to date. The documentation link presented on the main menu of ST.cpp has also been corrected to point to our maintained web site.


PWP:2001-03-06 Release Notes

This release includes the following developments:


PWP:2001-03-06 Module Testing Implementation & Status

This new release includes several important new macros tailored for module production testing, all of which can be found under the new menu ABCD Test , accessible from the main menu by pressing the ABCD Test button.

For a more detailed explanation of these tests, please refer to Peter's Electrical Tests of SCT Hybrids & Modules specification document. Two test sequences are provided:

Complete characterisation of a hybrid or module by pressing the Characterisation Sequence button.

This takes around 90 minutes (depending on PC performance), and produces the following output files for each module connected in the system:

Confirmation of basic hybrid or module functionality by pressing the Confirmation Sequence button.

This is an abbreviated sequence of tests including digital bypass and redundancy functionality as well as strobe delay and simple gain measurements. It is intended to be used more frequently than the full characterisation sequence, for example, at various stages of handling during production or shipping.

Both of these sequences invoke tests which are also available individually from additional buttons on the same menu bar:

BypassTest

A simple test of the ABCD Bypass functionality. Standard cases 0 to 10 (as defined in the test specification document) are studied by default: cases 11 and 12 have not yet been debugged. Results are appended to the summary text file.

RedundancyTest

Verification that the module responds to both the primary and redundant clock and command signals. Results are appended to the summary text file.

PipelineTest

A scan of the pipeline, to identify dead cells. Results are appended to the summary text file.

StrobeDelayTest

Firstly, a threshold scan is made to set the threshold of each chip to be approximately 2fC. Next a strobe delay scan is performed with 4fC charge. The edges of the resulting strobe delay peak are fitted and the strobe delay is set to be 25% of the way between the two edges. Results are appended to the summary text file.

ThreePointGain

A calculation of the gain and ENC at 2 fC from a linear fit on 50% points at 1.5,2.0,and 2.5 fC. At this stage of the characterisation the hybrid or module has not been trimmed. Results comprising gain, offset and input noise values are appended to the summary text file together with a list of bad channels. The bad channels are categorised as follows:
  • Lost = dead + stuck + unbonded channels
  • Dodgy = inefficient + low noise + high noise channels
Of course if the module is very noisy, has many unbonded channels or is built with poor quality chips, a very long list of faulty channels will be produced.

At the moment the default values for low/high noise are 1100 ENC and 1800ENC respectively. Whilst these values should be reasonable for stable, well cooled barrel or forward modules with ABCD3T chips, they may be modified by editing the macro ResponseCurvePlot.cpp at the following lines:

  • MIN_INNSE 1100
  • MAX_INNSE 1800
Modifications are under discussion to identify channels with anomalous noise values as those for which the output noise is more than |n σ| different from the average value.

TrimScan

The hybrid or module is now trimmed. For each module connected to the system, two files will be produced in the sctvar/results directory, the trim file and the mask file. If they are moved to the sctvar/config directory and renamed to match the name of the detector configuration file, they will be picked up the next time you initialise the system with that module connected. However the macro itself sets online these values so they will be used during the remaining tests. Only dead, stuck and untrimmable channels are masked.

ResponseCurve

The response curve is determined using a series of 10 threshold scans with charges between 0.5 and 8fC. The gain and input noise at 2fC are calculated using an exponential fit. The same analysis routine is used as for the ThreePointGain measurement; hence all comments about noisy channels are valid here too. When called as part of a Characterisation sequence, the chips will now be trimmed and (some) bad channels will have been masked: this should be kept in mind when comparing the ResponseCurve results to the ThreePointGain results.

Noise occupancy

A noise occupancy scan is performed at 100 kHz trigger rate (at least for users with CLOAC, and once the occupancy is down to the level of a few percent.) The number of triggers increases at low occupancy. For now, the output consists of a PS file plotting occupancy as a function of threshold. The graph is annotated to highlight the noise occupancy value at 1fC. For users with SCTLV, the nominal power consumption is also shown. During analysis the 1fC point is read from the most recent results file which contains trim results.

Timewalk

The timewalk is studied using a threshold of 1fC whilst injecting 10 different charges between 1.25 and 10 fC. If the 1fC threshold setting point is not known in advance, it is determined by making a threshold scan. Timewalk is defined to be the difference between the rising edges of the strobe delay peak for charges of 1.25 and 10 fC. (Studies continue into the possibility of making a fit to the timewalk data obtained for all 10 charges, however this seems to be more easily prone to error, so it is likely that the arithmetic method will persist.)



Other Comments

(Thanks to Daniela Macina for this section)

Although the OPTO Functionality Test for forward hybrids/modules has not yet been implemented, partial support for the Liverpool support card is included in this release. The commands to program the support card operating mode have been coded, although since the RESET line must be driven during programming of the operating mode, this will work for users with SCTLV only. Other users can select the operating mode of the card by means of the inbuilt switch.

At the time of writing, the masking of bad channels is not exhaustive. Extra channels may be masked by hand, by adding them to the mask file.

comments/suggestions/corrections to Bruce, Peter or Gareth