dyconex stat nov

Fine pitch cables from Dyconex: 

Fine pitch cables from Dyconex ordered engineering sample of CDF L00-type cable from Dyconex in May/June 2001 manufactured in a new process technology: thermal direct imaging with infrared sensitive photoresist after material optimization studies by Dyconex, the first two pre-prototypes have been delivered by 9/30 no gold plating and no solder resist applied, cables have unprotected bare copper traces and were untrimmed one cable sent to FNAL the other kept at Zurich for testing have results for these two on visual inspection, opens/shorts, capacitance and resistance last week got another 23 cables, which I brought to FNAL engineering run now over these cables are gold plated, have solder resist and are trimmed Frank Lehner U Zurich 11/12/2001

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex schematics of the cable (L00): 128 channels/traces 13.7 cm of 50mm pitch traces 26.0 cm of 100mm pitch traces fan-in and fan-out region (1.7-2.9 cm) total trace length including fan-in/out between 41.4 – 42.6 cm 2 rows of bond pads on each side two traces for bias lines general Dyconex tried to minimize trace width (lower capacitance) trace width measured typically to 7-8 mm

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex bond pad regions: outer pads are big enough, size is ~50 mm x 210 mm pads of inner row region very small, only ~30 mm x 190 mm Dyconex can try to increase pad size, but not by much. Difficult, since distance pad-trace becomes ~20um bond tests necessary to see if we can wire bond on such small pads two different kind of platings on pads on the 23 new cables: only gold and nickel+gold outer inner

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex performed optical inspection and electrical measurement on first flex cable total of 3 open traces visually found not so bad for first shot! opens confirmed by trace continuity measurements no shorted traces detected, only few regions with slight metal excess open trace excess metal – no short

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex Other typical problems: regular and systematic trace offset by ~5 mm every 2 mm (just a shift of one pixel size during imaging) inherent to imaging process and fixed only by a new plotter (a purchase is planned in one year from now)

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex in fan-in/out regions, critical thinning of traces along the horizontal axis observed trace width should be increased in fan-in/out region

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex spots of adhesive on few traces looks serious, but not harmful – better cleaning/rinsing will be done for further flexes

fine pitch flex cable from Dyconex: 

fine pitch flex cable from Dyconex Capacitance measurements LCR four probe measurement measure capacitance of one trace against all others (total capacitance seen by one trace) At 1 MHz: C~15.3 pF for 42cm => 0.36pF/cm At frequencies above 1MHz capacitance measurement inaccurate due to cable inductance. Result confirmed by Kazu’s measurements at FNAL with 2nd cable

fine pitch flex from Dyconex: 

Capacitance of one trace to the left and right neighbors has been determined to be 12 pF only,i.e. 0.28 pF/cm next-to-next neighbors carry 20% of the total capacitance the 0.28 pF/cm is in fair agreement with analytical calculation (~0.23pF/cm): fine pitch flex from Dyconex

fine pitch flex from Dyconex: 

fine pitch flex from Dyconex the capacitance will be higher if solder resist (coating) is applied calculations show an 20% increase for 2mm coating expect for the new 23 cables higher capacitances need also to disentangle experimentally the part of the capacitance with 50mm and 100mm pitch noise contribution: assume SVX4: 450+43*C_tot and 1.2pF/cm for Si C_tot=C_Si+C_flex~37pF (for 42cm cable) noise~2040e acceptable

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex trace resistance as measured by H. Haggerty R~200-220 W or 4.7-5.2 W/cm expect ~4.3-4.9 W/cm from copper resistance and trace widths and thickness

Status of fine pitch flex from Dyconex: 

Status of fine pitch flex from Dyconex serial noise due to Rs: ENC=13*C_tot*(Rs/t) Rs=Rs_Si+Rs_flex Rs_Si specified to be less than 30W/cm Rs=500-520 W serial noise ~950e However: conservative noise assumption total noise: 2250e wider traces would decrease Rs but enlarge capacitance => stay with thin traces

shielding : 

shielding CDF experience shows that shielding of cable is mandatory to avoid noise pick up ground layer will also reduce cross talk induced by inductive coupling for L0, cables will run in a stack of up to 12 single flex cables stacked cables also exhibit capacitances to each other => have to separate cables however: space constrains are tight maximum 4.6 mm space left before we hit outer sensors subtract 1 mm clearance for wirebonds stacked cable area: 3.6mm left, meaning ~200mm in between cables how ‘flat’ can we stack flex cables? Have to allow for tolerances, bends, warps in the cables individual shield: sandwich the single cable by a 80 mm thick low-e material (foam) underneath the cable and add a mesh copper layer with copper laminated Kapton copper layer can be used for individual +HV supply extremely difficult to make individual shields under such tight space constrains common shielding: wrap stack into a copper foil Guard trace: add additional guard traces on flex

shielding: 

shielding additional GND layer under the Kapton flex will increase capacitances seen by one trace shown is a calculation for two er for a CPW (w/o GND layer) and a CPWCB (w/ GND layer) Capacitance to GND layer not important for heights>200mm (100mm) if er=3.5 (1.5)

Repair on cables: 

Repair on cables it is unlikely to get zero fault cables accept cables with 1-2 faults, either opens or shorts repair on painted (i.e. having solder resist) cable is time consuming. CDF has done it though bonding the traces (~7um wide) in case of opens is extremely difficult: need very thin wire ~<10um ETH Zurich has some experience with ultra-thin wires (AMS) however: bonding on copper difficult propose two pairs of spare traces at left and right border of cable open or shorted traces can then conveniently skipped

How to proceed with fine pitch flex from Dyconex: 

How to proceed with fine pitch flex from Dyconex from prototypes to preproduction: Dyconex would have more process parameters to tune in order to improve the quality of the cable engineering run is over and ‘the remaining 10%’ are making the effort as the Dyconex engineer says especially gold plating is not yet optimized several problems related to plating process bonding is an important issue preliminary price tag for a mass production: $409/pc with one allowed bad trace have to go with at least another prototype sample as soon as we have our own cable layout Dyconex is moving their plant to outside the city limit of Zurich (still relatively close too my place) they expect to be back in operation by February 2002

other games to play: 

other games to play try to go with a 25mm thick Kapton cable => will reduce the capacitance plasma etch Kapton material between copper traces in order to make grooves into the cable => lower capacitance use another dielectric substrate than standard Kapton add halogens? polyethylene or similar. Has lower dielectric constant. However, radiation hardness not really good Liquid polymer – too fancy

Conclusion: fine pitch flex from Dyconex: 

Conclusion: fine pitch flex from Dyconex we have received flexes from Dyconex which are extraordinary difficult to make these cables have been engineered in a rather short time (June – October) and using a new technique I think Dyconex is close to demonstrate us that they are able to do such cables – but we should wait upon the result of the 23 cables image transfer works copper etching works gold plating still a problem for further improvements on cables and if we want to go with Dyconex we should make another prototype run before we can launch this prototype run in February or March we need to converge on the design