Signal Integrity

Transcription

1 Signal Integrity

2 Factors influencing Signal Integrity 2

3 Studying Factors Studied the following factors Resin system Fabric Construction Conductor Moisture Temperature Test method 3

4 Resin systems 4

5 Resin systems Neat resin electrical performance Source : IPC EXPO

6 Df Df (Resin System & RC) A : bisphenol A novolac system A FR4 B : bisphenol A+phenol novolac C : TUC mid-tg material D : TUC Hi-Tg material E : Hi-Tg Dicy-cured system B C D E F G H I J Mid-loss Low-loss F : phenol novolac system G : TUC mid-loss material H : anhydride base mid-loss system I : BT/Epoxy base system J : TUC low loss material % 45% 55% 65% RC 75% 6

7 Loss ( db ) Mid / Low Loss Materials Trace Length : 4 inch Dicy Hi-Tg Frequency ( GHz ) Low-loss : TU-872 LK Anhydride Novolac : TU-862 HF Low Dk resin system could be formulated with modified epoxy to achieve the desired low loss property and compatible with FR-4 processing parameters 7

8 Construction 8

9 Dk Construction 1MHz : LCR meter 9

10 Df Construction 1MHz : LCR meter 10

11 Loss ( -db/cm ) Loss Construction 1.4 Construction Impact Delta db 8 inch 12 inch 16 inch 5 GHz GHz GHz GHz HRC LRC HRC LRC mil 5 mil Delta db 8 inch 12 inch 16 inch 5 GHz GHz GHz GHz Frequency ( GHz ) Construction is one of key factor for loss 11

12 Fabric 12

13 Fabric Source : IPC EXPO

14 Low Dk Glass Signal Integrity Comparison SLK SLK SP SLK SP E glass Low Dk glass 1 Low Dk glass 2 GF R/C DK DF 106 6Z7B Z5B ZDB ZMB Z3B D6B D6B D4B DbB DPB DeB DdB

15 Low Dk Glass Signal Integrity Comparison The Dk of CCL reduces approximately by (depends on the RC) and on Df when E glass replaces with low Dk glass on the same resin system

16 Signal Skew Differential circuit Resin rich Resin poor Resin Dk ~ 3.2 Glass Dk ~ 5.6 Special Layout and Routing Designer Rotates Image Lopsided glass weaving 16

17 Signal Skew Fabric Selection 1080 FABRIC Special Layout and Routing Designer Rotates Image Lopsided glass weaving 17

18 Signal Skew Fabric Selection 1080 Grade Fabric Regular Open-Filament Flatten Weaving 18

19 Signal Skew Fabric Selection Window Area Ratio Regular Type Open-filament Type Flatten Weaving Type 2 mil ~30 % ~15 % ~1 % 3 mil ~25 % ~8 % ~3 % 4 mil ~10 % ~3 % ~1 % Optimize fabric can moderate the signal skew effect through: Use Low DK fabric Increase weaving density Even/flatten glass count/weave 100% 19

20 Conductor 20

21 Conduct Loss Roughness HTE roughened foil RSTF smooth foil 21

22 Conduct Loss Roughness G Circuit trace surface roughness level Type Position sample N F Roughness HTE Foil RTF Foil F : Shinny side mil G : matte side mil F : Shinny side mil G : matte side mil 22

23 Loss ( -db/cm ) Conduct Loss Roughness Roughened Foil Smooth Foil Freq/(delta) 8 inch 12 inch 16 inch 5 GHz db db db 10 GHz db db db 15 GHz db db db 20 GHz db db db Frequency ( GHz ) Material : Low Loss Material The difference in loss increases with increasing in frequency 23

24 Df Effective Dk/Df (IBM SPP Method) Effective Df HTE Foil MLS Foil D field solver Frequency ( GHz ) Changing from HTE to MLS foil, Df could reduce by ~ 18% 24

25 Temperature 25

26 Influence of Temp. PN cured High Tg FR4 SPP method Temp increases from ambient temperature to 70C could cause Df deteriorate by ~20% 26

27 Moisture 27

28 Influence of Moisture Resin system is a major factor for moisture pick up 28

29 Influence of Moisture Moisture Pick Up (wt%) 29

30 Influence of Moisture Moisture Pick Up (wt%) 30

31 Oxide Treatment 31

32 Influence of Oxide Treatment HVLP VLP STD Source : 2009 IPC APEX / Cisco / S20_02 The oxide tooth also will affect the SI. Shorter the tooth, lower the interference 32

33 Measuring Method 33

34 IPC Dk/Df & SI Test Method SECTION ELECTRICAL TEST METHODS V 2.5.5A B A C Dielectric Constant of Printed Wiring Materials--7/75 Permittivity (Dielectric Constant) and Loss Tangent (Dissipation Factor) of Insulating Material at 1MHz (Contacting Electrode Systems)--5/86 Dielectric Constant and Dissipation Factor of Printed Wiring Board Material--Clip Method--12/87 Permittivity (Dielectric Constant) and Loss Tangent (Dissipation Factor) of Materials (Tw o Fluid Cell Method)--12/ Dielectric Constant and Dissipation Factor of Printed Wiring Board Material--Micrometer Method-- 10/85 Stripline Test for Permittivity and Loss Tangent (Dielectric Constant and Dissipation Factor) at X-Band-- 3/ C Stripline Test for Complex Relative Permittivity of Circuit Board Materials to 14 GHZ-- 3/ Non-Destructive Full Sheet Resonance Test for Permittivity of Clad Laminates-- 5/ Characteristic Impedance and Time Delay of Lines on Printed Boards by TDR-- 11/92 V Low Frequency Dielectric Constant and Loss Tangent, Polymer Films-- 7/ Permittivity and Loss Tangent, Parallel Plate, 1MHz to 1.5 GHz--11/98 V V V High Frequency Testing to Determine Permittivity and Loss Tangent of Embedded Passive Materials Propagation Delay of Lines on Printed Boards by TDR Test Methods to Determine the Amount of Signal Loss on Printed Boards (PBs) Relative Permittivity and Loss Tangent Using a Split-Cylinder Resonator 34

35 TUC Capability TUC Capability For base CCL material (Network Analyzer ) ( Impendence Analyzer ) ( VNA & Split Post Cavity ) For print circuit board Time delay ( TDR ) SPP, SET2DIL ( TDR, VNA ) Cisco MST ( VNA ) Eye-Diagram 35

36 IPC Test equipment : HP E4980A Frequency range : 20KHz~2MHz From Cp=E0XErX(A/d) to estimated Dk value of material at low frequency. D show material Dissipation factor values. E0= 8.854X10^-12 F/M Er= the Dk of material d: thickness A: Area 36

37 IPC Equipment Used: HP4291B 1MHz-1.8GHz RF Impedance / Material Analyzer HP4291B Test Station HP16453B Dielectric Material Test Fixture This method uses the material capacitance and conductance to determine the Dielectric Constant and Dissipation Factor 37

38 IPC SPC Fixture VNA Equipment used : Vector Network Analizar R&S ZVB 20 VNA available frequency range : 10MHz~20GHz Fixtures working frequency : Split For resonant measurements, the Dielectric Constant is determined from measurements of the resonance frequency and quality factor The Q=f 0 /f, where f 0 is the resonance frequency and f is the frequency difference between 3dB points 38

39 Summary Measurement Equipment Methods Sample Accuracy Frequency Range Thickness Size Dk Df Agilent HP 4291B IPC Mm 25x50 mm 1MHZ-1GHz +/ / Agilent E4980A LCR IPC ~1.6 Mm 50X50 mm 20KHz~ 2MHz +/ / R&S VNA ZVB 20 IPC SPC 0.2~0.5 mm 80X100 mm 1GHz ; 3GHz 5GHz ; 10GHz +/ /

40 Insertion Loss Measurement 40

41 IBM SPP Technique Time-Domain Extraction Technique Measured Simulation TDR Base LOSS 2D field solver Effective Dk/Df up to 50GHz SPP output : S21, Dk, Df, R, L, G, C 41

42 IBM SPP Technique Effective Dk/Df - including Dielectric Dk & Df Trace effect Oxide treatment Foil roughness Skin effect CCL + PCB 42

43 Cisco MST Method Cisco MST Method : For 2+ GHz range Use VNA to extract parameters. Real PCB stripline TV Not a tuned resonator. Calibration structures built into the board ( TRL ) Thousands of discrete measurements (~5MHz intervals) Results yielded : Effectively continues Dk/Df curves Dk/Df extraction algorithms updated to account for Cu roughness Source : Cisco 43

44 Cisco MST Method Source : Cisco 44

45 Intel : SET2DIL SET2DIL Single-Ended TDR/TDT To Differential Insertion Loss Use TDR to measure SDD21 instead of VNA. Measuring SDD21 using only a 2-port TDR measurement. More suitable for High-Volume Manufacturing (HVM). SET2DIL is ½ the length of a standard insertion loss test structure. Can also be used as an impedance test coupon. 45

46 Intel : SET2DIL Source : 2010 DesignCon / Intel SET2DIL output : SDD21 46

47 Measurement results Material Type TU-862 HF TU-872 LK SPC SPP Resin RC % Dk Df Dk Df Insertion loss GHz Rich Poor Rich Poor SET2DIL Material 1GHz Foil Layer 4GHz 8GHz (db/inch) (db/inch) TU-862 HF HTE L TU-862 HF HTE L TU-872 LK MLS L TU-872 LK MLS L Measurement results vary with test method 47

48 Loss Dominants 48

49 Loss Elements (SPP Method) Attenuation = f ( frequency ) = Dielectric loss + Conductor loss = Loss ( base material ) + Loss ( circuit trace ) = Material Dk / Df + Circuit Geometry and Roughness 49

50 Loss Dominator Breakdown element FR4 Overall Loss Dielectric Loss ~1.5GHz Conduct Loss RTF foil 50

51 Loss Dominator Mid-Loss Overall Loss Dielectric Loss ~4.0GHz Conduct Loss RTF foil 51

52 Loss Dominator Overall - HTE Overall - RTF conductor - HTE ~17GHz Dielectric ~12GHz conductor - RTF 52

53 Loss Elements (SPP Method) Before the breakeven point (circle with red), the conductor will carry more weight age in total loss than material. After cross over the breakeven point, material will have higher effect on the total loss than conductor Break even point will be shifted to the right when moving from standard loss to very low loss material Before the breakeven point, improvement should focus on improving the processes like oxide type, copper tooth or copper etch rate. However, after the breakeven point, the total loss can be reduced effectively through selecting a lower loss property material Copper becomes a critical factor when moving toward from standard loss application to very low loss application 53

54 Factors influencing Signal Integrity Product Copper Test Method Signal Integrity Oxide Construction Temp. Fabric Moisture 54

55