Download ppt - Tony Smith LHCb Velo Electronics PRR CERN 16/01/06 1 LHCb Velo Electronics PRR Front End Hybrids and short cables

Tony Smith LHCb Velo Electronics PRR CERN 16/01/06 1

LHCb Velo Electronics PRR

Front End Hybrids and short cables


Scope of this presentation

Detailed construction and mechanical aspects of the front end hybrids are covered by the module review and therefore this presentation contains only a small amount of detail to outline the steps of the construction and testing.

Detailed information of the assembly and testing can be found at http://hep.ph.liv.ac.uk/lhcb/Hardware/Production/production.html

Schematics, layouts and production files can be found in EDMS

http://hep.ph.liv.ac.uk/lhcb/Hardware/Production/production.html


HYBRIDS- construction

CONSTRUCTION of composite SUBSTRATE

TPG central core 400um

woven layer of CF prepreg each side oriented +-45 degrees 125um nominal

woven layer of CF prepreg each side oriented 0,90 degrees 125um nominal

Laminated in vacuum bag at Liverpool

Total thickness ~900um

Frame of Carbon Fibre 3mm wide around edges to prevent delamination

CF and TPG outlines shown in RED

– the substrate is routed from the kit of

these parts to the substrate profile shown in

GREEN –

the circuit is shown in BLUE

SUBSTRATES


HYBRIDS-construction

Four copper layers with Kapton dielectrics bonded with 50um Espanex (modified polyimide) bondply – essentially a standard multilayer flexi circuit.

DESIGN RULES - not very tight for the technology good yield

100um track and gap

300um vias - 500um lands no buried vias

Layer material pressed copper thickness

thickness top bottom

solder resist 30

1 copper foil 05+carrier 5um 5um -------

SPB-050A Bond Ply 50um (was 35) -- --

2/3 SB1250 Cop/Clad flexi 60um 12um 12um

12/50/12 (Adhesiveless) (power)

SPB-050A Bond Ply 50um (was 35) -- --

4 copper foil 05+carrier 5um -- 5um

total 200um (ground)

NB – plating thickness of ~8um Cu is added to top and bottom + few um Ni/Au on top

CIRCUITS


HYBRIDS- construction

Component and trace layout

CIRCUITS layout



Two CIRCUITS are now bonded back to back on the SUBSTRATES using the corner holes for alignment. The cut-outs in the corners allow for the differential expansion of the jig relative to the circuit and avoid stressing of the substrate when heated.

Bonded using 50um bond Ply as in the circuit construction

Total thickness = 900um (substrate) + 2* 50um (bondply) + 2*200um(circuits) = 1.4mm

Circuits are manufactured at Stevenage Circuits Limited UK. The circuits undergo two standard tests using flying probes for open and short circuits before and after plating and only parts which are 100% good are charged for. After lamination a further flying probe test is done, however because of some non linear distortion introduced in the lamination process it is not possible to probe the bond pad fingers.

Instead, only the surface mount pads and vias are used as these are large enough to be probed. This is likely to catch the vast majority of induced shorts (one found so far in 24 circuits (12 hybrid assemblies)

Lamination to form a HYBRID


HYBRIDS

Tests on early prototypes showed that reflow soldering would produce distortions as the whole hybrid would be heated beyond the glass transition temperature of the adhesives without the hybrid being supported. It was therefore decided to Hand solder the components at Hawk Electronics UK.

Bond areas are masked to prevent solder splashes and damage.

Hybrids are tested for short circuits after population, they are visually inspected and any problems rectified. They are then cleaned with prozone defluxer and DI water prior to die and Pitch adapter attachment.

Population


Hybrid – post population testing

Hybrids can be tested visually and quickly for shorts using a known good cable.Each pin on the connectors has an individual LED so probing any one pin on the contacts should light only one led of any pair. Shorts are quickly found by running the probe along the pads. This does not test for open circuits but the most likely failure mode at this point is shorts from soldering.



Pitch Adapter designs for Phi and R All locations are different, drawn in sets of 8 FINE LINE 20um T&G

Pitch Adapters

8 phi and R sets tested and available

Can test faster than module production rate

Half of required PAs delivered

Only 100% good parts pass

Failure rate from cap tests 11%



Four sets of pitch adapters glued to the substrate. Glue is precisely dispensed and weights are applied during cure

Pitch Adapter attachment


HYBRIDS Hybrid + short cable tested together in Vac Tank

Hybrid populated with decoupling caps (2*500v 10nf to Gnd)

Guard and HT shorted together

All lines except HT and Guard on HT connector shorted to ground (includes Power)

Discharges (brief – up to 15nA ) seen on ramping up above 430V

0.4nA leakage at 500V for 1 hour

0.2nA leakage at 500V if taken to 510 and back down again.

holds 500V but need more statistics

K5 clearances are the same or better than K4 so no differences expected.

HT ISOLATION TEST


Cables - testing

Cables can be fully tested visually and quickly for shorts, open circuits and intermittent faults. Each pin on the connectors has an individual LED so continuity is easily assessed. LEDs are powered in blocks with most likely pair to pair or pair to power and ground shorts show up immediately


MODULE testing LASER SYSTEM

The laser system is used to test and debug modules at several stages during construction: after the chip back ends are bonded, after PA’s are bonded and when sensors are bonded. The system makes it possible to stimulate every sensor channel under program control. Data from all tests is uploaded to the production database at each stage.


HYBRID K5 Status -1Double sided hybrids produced and functionally verified in test setup with short cables and dummy cards ( but not HT with long cable?)

No parametric S/N from this (K5) version yet, but construction layout and components are virtually the same as previous version (K4) which was successfully tested in a beam last year. The only substantial change is that of splitting the power plane into four separate planes so that rad hard voltage regulators may be used and this very unlikely to change high frequency performance that could impact on the noise.

Active components (readout chips and temperature sensors and detectors) are qualified as Rad Hard Passives are standard thin film resistors and capacitors with X7R dielectrics as used in ATLAS Circuits are bonded to substrates using modified polyimide as in the manufacture of kapton multilayer circuits which are known to be radiation tolerant. NOTE the radiation performance of the epoxy adhesive used in the CF prepreg is untested and unknown


HYBRID K5 Status -2

current status of the hybrid production is: 7 K5 hybrids populated with discrete components of which1 at cern for electrical test with chips one side and back ends bonded1 at cern for mechanical test – no chips 2 at Liverpool chips both sides, fanouts attached – fully working back ends –awaiting FE bonding – electrical for PRR?2 with chips both sides – partially working back ends – mechanicals for PRR1 unused11 more awaiting population at Liverpool (3 next week 8 more in 2 weeks) 28 more substrates at Stevenage for Lamination a further 100 circuits produced and tested at Stevenage (sufficient for 50 hybrids ) 19 more pieces of TPG available 20 more on order

QA procedures and database tracking of all parts is in place.


HYBRID K5 Known issues -1Minor PCB layout error means that POR RET line is not terminated on the hybrid – this will have no effect on the operation of modules. – see schematic Yield of flat substrates is unstable

From the first batch of 20 substrates produced ` 3 were rejected on twist <> 200um+-200um

1 rejected on thickness 1 rejected on surface quality

from the second batch of 2010 pass twist <> 200um+-200um17 pass with 300um +-300um

hybrid flatness after lamination to selected substrates (200um criteria) is not stable, From last batch of 12 laminated substrates

1 failed electrically – unfixable short circuit after lamination only 4 pass a 200um +-200um flatness criteriaor 7 pass with a 300um+-300um flatness criteria

efforts to resolve this are ongoing – we believe it may be possible that acceptance can be increased by batching into groups with similar twist and adjusting the angle at the top of the paddles


Cables Known issues Prototype cables were not flat (wrinkled) which is thought to be caused by heating the cables without them being clamped along their length when stiffening pieces are attached. The manufacturer is aware of the problem and intend to try to clamp the cables in the next production. This is not an electrical problem but it was initially thought that it may pose problems for attaching to strain relief clamps on the module bases. Subsequent measurements show that although there is some wrinkling it looks worse than it really is and should not cause problems.

The prototype cables were made 5mm too long – however this does not pose a problem (may help with insertion) and production cables will be made the same size to avoid a delay for retooling. Soldering of connectors to cables by hand soldering has shown some open circuits which are fixable but require rework and raise the question of reliability. This is thought to be caused by use of low flux content solder which produces a visually appealing finish but shows poor wetting of contacts.

This has been raised with the population company and will be solved by using different solder better QA or possibly by using a reflow method.


Ground cable Status & Known issues

8 Prototype ground cables were made at the same time as the short cables

From 3d modelling it appears that they will be approximately 1.6mm short. This does not present a problem for initial tests as the holes in them can be enlarged.

It has now been decided to split the production of these cables from that of the short cables so that there is no impact on the short cable production due to the need to retool the dies for cutting.

Redesign and manufacture of correct size parts estimated at 6 weeks