PRR – March 31st 2006 D.BretonD.Breton

ECAL/HCAL Frond-End Board:ECAL/HCAL Frond-End Board:tests at production site.tests at production site.

Production flow.Aging.Flying probes.Boundary scan.

PRR – March 31st 2006PRR – March 31st 2006

Production flowProduction flow All the PCBs and components are checked upon reception at the factory. SMC components are mounted first. Right after the SMC oven, there is a visual control of each side of the board

plus a BGA Xray test for one out of 10 boards. The press-fit connectors are mounted with a dedicated tool. The other components are finally mounted and soldered on the wave. The analog input connectors have to be resoldered manually. There is a final visual control before powering the boards and checking the

power consumption. If this is OK, the boards are sent to the aging oven. The power consumptions are tested again at the exit of the oven to look for

potential component failure. Then the boards undergo the flying probe test … … followed by the Boundary Scan test. If everything is OK, the boards are sent to LAL. All the former step results are noticed in a report file. The production boards will be produced by lots of 32 boards every week.

PRR – March 31st 2006PRR – March 31st 2006

AgingAging Manufacturer will perform an

aging of groups of 16 powered boards in the same crate.

Goals :

• Detection of bad soldered components.

• Infant mortality. First trial with 8 boards

showed no disastrous effect: all boards were functional before and after the aging.

Power consumption is controlled before and after the aging.

Power supplies

PRR – March 31st 2006PRR – March 31st 2006

Aging ovenAging oven

Air fan

Tempprobeon thecentral

ADC

PRR – March 31st 2006PRR – March 31st 2006

Temperature cyclesTemperature cycles 8 cycles of 9 hours between 0 and 60 degrees. The upper temperature was fixed after measurements made on the central ADC (the

hottest component on the board): its package was at 90 degrees when air was at 60. The goal is not to pass over 150 degrees on the ADC die.

PRR – March 31st 2006PRR – March 31st 2006

Flying probesFlying probes Automatic test with a ‘ Takaya ‘ robot :

• Measures the impedance values of nets and components and checks the interconnections.

• Can diagnose a bad soldered BGA thanks to the ‘open checker’ tool.

• Footprints for probes have been implemented under the BGAs.

PRR – March 31st 2006PRR – March 31st 2006

Access to vias below the Access to vias below the BGAs BGAs

The varnish has been removed from all the vias not connected locally to another component..

PRR – March 31st 2006PRR – March 31st 2006

The flying probesThe flying probesFlying needles

Open checker:Hall effect

probe

PRR – March 31st 2006PRR – March 31st 2006

The FEB on the TakayaThe FEB on the Takaya

Onetest on

each side

PRR – March 31st 2006PRR – March 31st 2006

Boundary scan.Boundary scan. Partial Boundary Scan Test:

• Ability to drive 2 ACTEL chains independently.

• GluePGA and SeqPGA have to be programmed first (takes 10 min).

• Test of Actels’ interconnections => all BGAs.

PRR – March 31st 2006PRR – March 31st 2006

The JTAG chainsThe JTAG chains2 J ta g ch a in s : A x celera to r a n d P ro A sic

D ed ic a ted J tag p in s D ed ic a ted J tag p in sT C K T D I T D OT M S T C K T D I T D OT M S

G lu e p g aA P A 1 5 0

S eq p g a

A P A 3 0 0

IS P ( P roA sic p ro gram m in g C on n ecto r )

S ilicon E xp lorerjtag con n ecto r

F e p g aA x 2 5 0

D ed icat ed J t agp in s

T ckT m s

T d iT d o

T C K

T M S

T D I

T D O

TD I[n ]

TD I[n :0 ]

TD O [n]

TDO [n :0 ]

TD I

TD O

Tr ig p g aA x 5 0 0

T ckT m sT d i

T d o

T C K

T M S

T D I

T D O

TD I

TDO

TCK

TM S

Lem o

D ed icat ed J t agp in s

P rob e s

' h om e-m ade 'P ro bes

P rob e s

EC S

TDO [n -1 :0 ]

EC S

TD I[0 ]

TD O [0]

Pro

gram

mab

leO

utpu

t

P r o g r a m m a b leO u tp u t

V ersion 1 /0 3 /2 006

PRR – March 31st 2006PRR – March 31st 2006

Test setupTest setup

AX chainInput:scan +

explorer

APA chaininput:prog +scan

PRR – March 31st 2006PRR – March 31st 2006

Top viewTop view

BoundaryScan

controller

Neighbourcheck

from FEPGAon backplane

Testbackplane

PRR – March 31st 2006PRR – March 31st 2006

Test and repair report (1)Test and repair report (1)

Repère / Code défautLM 001 x RAS Sans JTAG

LM 002 27/02/2006 x J7 / BPV OK Testé avec ancien programme GLE

LM 003 08/03/2006 xJ1, J7, J9 / BPV - CC sous M21 (à

changer)OK Testé avec ancien programme GLE

LM 004 27/02/2006 xC132 / ADF - M63 / AHS - J7, J9 / BPV -

M63 / AHS OK Testé avec ancien programme GLE

LM 005 08/03/2006 x CC sous MN21 (changer) OK Testé avec ancien programme GLE

LM 006 08/03/2006 x J9 / BPV - CC sous MN21CC sous J6 cause blindage Action

corrective sur le lot - Testé avec ancien programme GLE

LM 007 x M17 / ANC (changer le 27/02/06) PB test infra JTAG - en expertise

LM 008 27/02/2006 RAS OK Testé avec ancien programme GLE

LM 009 27/02/2006 RAS OK Testé avec ancien programme GLE

LM 010 27/02/2006 J4 / BPV OK Testé avec ancien programme GLE

LM 011 27/02/2006 RAS OK Testé avec ancien programme GLE

LM 012 27/02/2006 J9 / BPV OK Testé avec ancien programme GLE

LM 013 27/02/2006 RAS OK Testé avec ancien programme GLE

LM 014 08/03/2006 RAS OK Testé avec ancien programme GLE

LM 015 27/02/2006 RAS OK Testé avec ancien programme GLE

LM 016 27/02/2006 J1 / BPV OK Testé avec ancien programme GLE

Test Sondes mobilesLivraison Test JTAGRAYON XOF

N° de série

CNR/F10/01

PRR – March 31st 2006PRR – March 31st 2006

Test and repair report (2)Test and repair report (2)

LM 001

LM 002

LM 003

LM 004

LM 005

LM 006

LM 007

LM 008

LM 009

LM 010

LM 011

LM 012

LM 013

LM 014

LM 015

LM 016

OFN° de série

CNR/F10/01

+3,3V +5V -5V +3,3V +5V -5V

X M21 et M30 inversé (Changé Zevac)

XM21 et M30 inversé (Changé Zevac) + Pose isolant sous blindage

M21 et M30 inversé (Changé Zevac)M21 et M30 inversé (Changé Zevac) + Pose isolant sous blindage

X M21 et M30 inversé (Changé Zevac) + Pose isolant sous blindageM21 et M30 inversé (Changé Zevac)

X

X Essai mesure de température en statique sur U2B à 89,8°C

X

XPrise de température lors du déverminage dynamique 72H sur l'ADC du milieu (carte au centre du rack) + pose isolant

sous blindage

X

DIVERS

Mesure consommationAvant Après

Déverminage

PRR – March 31st 2006PRR – March 31st 2006

Coding of the defaults.Coding of the defaults. Codes défauts utilisés dans les relevés de défauts

A : Article (composant). ANC - Non conforme vis à vis de la nomenclature. Le composant ne correspond pas à l'article prévu à cet emplacement,(désignation, valeur, tolérance, fabricant, boîtier) Pour 2 composants mis l’un à la place de l’autre noter 2 articles ANC. Un composant en trop à un emplacement normalement vide est aussi ANC. Pour une reprise de réglage utiliser le défaut ANC en indiquant "Réglage" à la place du repère de composant. Pour une carte non vernie noter ANC avec comme repère le nom de la carte. AHS - Hors service (Tout composants défectueux ). - résistance ou autre composant coupée électriquement - CI non fonctionnel AMI - Mal implanté (surélevé, mal plaqué, de travers, patte pliée, mal cambré,...) AMT - Manquant (Manquant pour un emplacement donné) Le composant AMT peut-être un chevron de soudure, une modif filaire,…) AIN - Inversé ( polarité: diode, condo polarisé,... sens: circuit intégré, connecteur,..). ADF - Détérioré en fabrication (mauvaise manip, salissure, brûlure, déformation) Utiliser ce code lors de défaut sur le cuivre occasionné par la fab ou l’opérateur. AHT - Composant hors tolérance B Brasage (soudure au fer ou en machine :vague, four,...) BPS - Pont de soudure coté surfusion BPV - Pont de soudure coté vague BNS - Non soudé ou manque de soudure coté surfusion BNV - Non soudé ou manque de soudure coté vague BNC - Non soudé ou manque de soudure dû à la colle Z CI, carte, support ZCP - coupure de piste du au fabricant du cuivre. ZCC - court circuit du au fabricant du cuivre. RAS Rien à signaler (Le défaut trouvé par le testeur n'aboutit pas à une intervention sur la carte) - Problème dû au moyen de test, - Article changé par erreur. - Dépassement de la tolérance jugé acceptable par le

réparateur.

PRR – March 31st 2006PRR – March 31st 2006

Main problems encounteredMain problems encountered When the first series of 24 PCBs was launched for the preseries of 16

boards, only 7 passed the checking procedure. The problem was the width of some conductors which shrinked below

70µm (which is the absolute limit). So only 7 boards could be mounted, and the 9 remaining PCBs had to

be reproduced. Then the 2 APAs were mounted in place of each other except on the first

board (shared responsibility) => we ended up with only ONE good board => Jacques almost got depressed ...

=> BUT this was a good practice for replacing BGAs ! Unsoldering and resoldering a BGA isn’t an easy job !

One prototype board was damaged during such an operation. There were shorts after the operation in a few early cases. New specific tools were developped for the ZEVAC machine. Seems operational now …

Shortcut below one of the press-fit HM connector shields and one line => solved by inserting a little piece of capton below the shield to isolate.

PRR – March 31st 2006PRR – March 31st 2006

SummarySummary The production chain is now perfectly defined and operational. Our experience with the factory has now become long and is very

effective and satisfactory. All the boards will undergo:

A 72-hour aging A flying probe test A partial boundary scan test.

All test softwares were developped by the factory. The failure coverage is at the level of ??% ... Up to now, no board declared « OK » by the factory test was found

bad at LAL, except for certain component failures which they cannot detect without a functional test (2 occurrences).

we’re confident for the production overall quality.

We’re in the starting blocks !!!