CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca...

15
CLARO QA at Milano-Bicocca http://CLARO.mib.infn.it Angelo Cotta Ramusino, Massimiliano Fiorini, Roberto Malaguti INFN and University of Ferrara, Italy Paolo Carniti, Lorenzo Cassina, Claudio Gotti, Matteo Maino, Clara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca, Italy Mateusz Baszczyk, Piotr Dorosz, Wojciech Kucewicz INP and AGH-University of Science and Technology, Krakow, Poland Some considerations of these slides have been not previously discussed with people in gray, but people in gray cannot be forgotten as they are part of the business, anyway.

Transcript of CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca...

Page 1: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

CLARO QA at Milano-Bicocca http://CLARO.mib.infn.it

Angelo Cotta Ramusino, Massimiliano Fiorini, Roberto MalagutiINFN and University of Ferrara, Italy

Paolo Carniti, Lorenzo Cassina, Claudio Gotti, Matteo Maino, Clara Matteuzzi, Gianluigi PessinaINFN and University of Milano Bicocca, Italy

Mateusz Baszczyk, Piotr Dorosz, Wojciech KucewiczINP and AGH-University of Science and Technology, Krakow, Poland

Some considerations of these slides have been not previously discussed with people in gray, but people in gray cannot be forgotten as they are part of the business, anyway.

Page 2: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 2

A short summary of the system strategy…

For large volume systems the good practice is to design-for-testability.

This is the strategy we tried and try to apply in all our projects.

One example for this is the HV distribution board system of the present RICH: the boards were designed to work with 2 configurations, one of which designed only for testing purpose (HV boards had 100 %).

IEEE TNS, Vol. 56, pp. 2828-2834, 2009;NIMA, vol. A598, pp 173-174, 2009;IEEE TNS, Vol. 53, pp. 1397-1402, 2006.

Page 3: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 3

… of the layout

CLARO

FPGA

PMT

The CLARO layout has maintained the same approach: it is completely modular, allowing for several ways of testing approaches (maybe too many…).

This was the aim since the first prototypes.

NIMA, vol. 652, pp487-490, 2011.

Page 4: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 4

The modularity of the layout vs the testability (1)

Modularity allows easily to test each part individually.

Page 5: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 5

The modularity of the layout vs the testability (2)

A further property: the chip area is small.

As a consequence the yield is expected to be quite large.

FEB’s

Page 6: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 6

The modularity of the layout vs the testability (3)

At the end every part of the so called elementary cell is separable from the rest. This gives the maximum flexibility for the test procedure to be adopted.

Page 7: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 7

QA procedure

1. Go-non-go test: in this case only a few significant parameters are verified;

2. Boundary scan procedure: at the design level a “series path” is implemented that allows to cascade the parts and verify the integrity of the system with few “serial” signals. This is typical of a digital domain and does not apply to our case;

3. Full-test: all the parameters are tested sequentially. This is our case.

QA procedure can be done following several steps

We will concentrate on steps 1 and 3.

Both steps 1 and 3 can be implemented with the same setup or different setups.

Page 8: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 8

QA option 1

FPGA

CLARO FEBPMT

Steve’s board

1. CLAROs are assembled on their FEBs;

2. The FEBs are connected to the Steve’s board (previously tested);

3. The FEBs are subjected to a go-non-go test from the FPGA;

4. The FEBs (not working FEBs are substituted) are characterized from the FPGA.

5. Defected FEBs Boards are re-worked.

No PMTs are used at this stage. Partial final cell

Concerning the CLAROs several options can be considered, listed below.

Page 9: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 9

QA option 2

FPGA

CLARO FEB

PMT

Steve’s board

1. CLAROs are assembled on their FEBs;

2. The FEBs are connected to FE-FPGA (INFN-FE) board or to a MIB-ARM-Cortex -controller board (INFN-MIB);

3. The FEBs are subjected to a go-non-go test from the MIB--controller/FE-FPGA;

4. The FEBs (not working FEBs are substituted) are then characterized with the same MIB--controller/F-FPGA card;

5. Defected FEBs Boards are re-worked.

No PMTs are used at this stage.

No Steve’s board are used at this stage.

MIB‐controller or FE‐FPGA

The system can be designed to test one or more boards at the same time.

Page 10: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 10

QA option 3

CLARO FEB

1. CLAROs are tested individually with a go-non-go test on a burn-in card;

2. CLAROs are mounted on the FEBs;

3. The FEBs are subjected to a go-non-go test from the MIB--controller/FE-FPGA (this step can be eventually omitted);

4. The FEBs (not working FEBs are substituted) are then characterized from the MIB--controller/FE-FPGA;

5. Defected FEBs are re-worked.

MIB‐controller or FE‐FPGA

Step 1 Step 2

Page 11: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 11

QA status (1)

We investigated how to proceed.

Opt 1 is part of the cell characterization and must be done in lab.

Opts 2 and 3 are time consuming and can be done by an operator. We asked to a good company to do that obtaining a quotation.

At the moment we considered only the (very good) company to which we (FE-MIB) refer for our assemblies.

The cost of a specialized operator is slightly less than 50 €/hour, duties included.

CLARO reworking (included re-testing) is 9 €/board.

Examples:

testing single CLARO at 30 sec/chip is about 0,35 €/chip. A more conservative 1 min/chip is about 0,7 €/chip, or 30000 € per 40000 (5000 PMT readout).

Go-non-go on a FEB is about 4 €/FEB at 5 min/FEB test.

Page 12: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 12

QA status (2)

We are now equipped with a working re-working station for last minute problems.

Page 13: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 13

Final considerations

Adoption of either of the 3 options is dependent on the CLARO yield.

We(-MIB) are in favor of opt 1 or its hybrid opt 2.

The next production will be about 250 CLAROs.

Based on the yield of this small production we will be in the condition to decide to follow opts 1 or 2 or opt 3.

Even in case the yield would be excellent, for the final production, in a conservative way, a few hundred of FEBs could be assembled and tested before to decide for the final opt.

Page 14: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 14

Spare

Page 15: CLARO QA at Milano-Bicocca Gianluigipessina.mib.infn.it/Biblio/LHCb/CLARO QA at Milano-Bicocca Gianluigi.pdfClara Matteuzzi, Gianluigi Pessina INFN and University of Milano Bicocca,

Edinburgh, April 13, 2015 g.pessina 1515

Documentation:

Most of the documentation is, or linked starting from, the new CLARO home page:

http://CLARO.mib.infn.it

CLARO was designed following the LHCb upgrade requirements and the characteristics we measured on the MaPMT.

CLARO characteristics and RICH specifications are almost the same.