Extraction of Broadband Error Boxes for Microprobes and Recessed Probe Launches for Measurement of Printed Circuit Board Structures

Transcription

1 Extraction of Broadband Error Boxes for Microprobes and Recessed Probe Launches for Measurement of Printed Circuit Board Structures, Renato Rimolo-Donadio, Christian Schuster Institut für TU Hamburg-Harburg, Germany

2 Outline Introduction Extraction of the Microprobe Error Box Extraction of the Recessed Probe Launch Error Box Extraction of combined Probe + RPL error box Cascading of the Error Boxes and Correlation to Their Fully Extracted Counterpart Conclusions 2

3 Measurement Challenges in SI Memory Controller Via Ball Bonding Connector Bonding wire IC Via Typical high speed memory link Wideband measurement of long, lossy interconnects in a frequency range from near DC to 40 GHz 3

4 Surface Launch Techniques SMA launch on printed circuit board Surface probe launch Surface Connectors Access Vias MICRO-PROBE 5 mm 5 mm STRUCTURE UNDER TEST STRUCTURE UNDER TEST 4

5 Recessed Probe Launch Ground Vias MICRO- PROBE MICRO- PROBE STRUCTURE UNDER TEST Ground pads with U strap Recessed probe launch [9] Signal trace No access vias Probe placement closer to the embedded structure under test 5

6 Goals of this and future studies Extraction of error boxes of microprobes and recessed probe launches Improvement of the launch Explore limits of de-embedding Create appropriate models for future use Using: Two-tier calibration technique Different calibration algorithms and calibration substrates SOL, SOLT and TRL 6

7 Two-Tier Calibration Reference Plane A Reference Plane B VNA Cable Fixtures DUT Usually coaxial - Adapters/Connectors - Probes - Dedicated Launches 1 st tier - Removal of measurement errors due to the measurement equipment (reference plane A) 2 nd tier - Extraction of the error parameters for the fixtures (reference plane B) 7

8 Used Calibration Algorithms Implemented in Matlab VNA algorithm 1-port SOL 2-port TRL 2-port SOLT 8

9 Measurement Equipment Anritsu VNA 37397D, TET Laboratory at TUHH Microprobes connected to the VNA Calibration bandwidth: 40 MHz to 20 GHz, f = 40 MHz 9

10 Outline Introduction Extraction of the Microprobe Error Box Extraction of the Recessed Probe Launch Error Box Extraction of combined Probe + RPL error box Cascading of the Error Boxes and Correlation to Their Fully Extracted Counterpart Conclusions 10

11 Extraction of the Microprobe Error Box Reference Planes Coaxial SOLT (1st tier) A B B A Cable End Probe CS* Probe Cable End Error Box 1 Error Box 2 Reference Planes SOL on CS-5 (2nd tier) CS*= calibration substrate 11

12 Calibration Standards and Probes Microprobe 50A DS-Style GGB Industries [10] CS-5 Calibration Substrate Signal tip Typical line standard on CS-5 Al 2 O 3, ε r = µm pitch GSG-probe tip [9] Ground tips 12

13 Microprobe Error Boxes The two probes behave different Their reflection doesn t exceed the -20 db limit Their transmission is lower than -1 db ~11% in all cases 13

14 Outline Introduction Extraction of the Microprobe Error Box Extraction of the Recessed Probe Launch Error Box Extraction of combined Probe + RPL error box Cascading of the Error Boxes and Correlation to Their Fully Extracted Counterpart Conclusions 14

15 Extraction of the Recessed Probe Launch Reference Planes SOLT on CS-5 (1st tier) A B C C B A Cable Probe RPL CS* RPL Probe Cable Error Box 3 Error Box 4 Reference Planes TRL on PCB (2nd tier) CS*= calibration substrate 15

16 Calibration Board TV1 GGB Microprobes 225µm pitch Stripline Ground Vias TV1 courtesy of IBM, Yorktown USA U shape (open instead of short was used) Typical Recessed Probe Launch [9] Line standards for calibration in the frequency bandwidth between 3.8 GHz and 19 GHz Dielectric: Megtron 5, ε r 3.9 PCB: 28 layers and ~ mil thickness ( ~4 mm) 16

17 Extracted RPL Error Parameters For the TRL calibration were used thru line 90 mil long, 50Ω line 220 mil long and high reflection standard The reflection exceeds the desired -20 db limit The transmission loss is below db ~ 4% 17

18 Outline Introduction Extraction of the Microprobe Error Box Extraction of the Recessed Probe Launch Error Box Extraction of combined Probe + RPL error box Cascading of the Error Boxes and Correlation to Their Fully Extracted Counterpart Conclusions 18

19 Extraction of a Fully Model Reference Planes Coaxial SOLT (1st tier) A C C A Cable Probe RPL CS* RPL Probe Cable Error Box 5 Error Box 6 Reference Planes TRL on PCB (2nd tier) CS*= calibration substrate 19

20 Microprobe + RPL Error Boxes TRL calibration was performed with the same standards on the same test board The reflection exceeds the -20 db limit The maximal transmission loss is -1.3 db 20

21 Outline Introduction Extraction of the Microprobe Error Box Extraction of the Recessed Probe Launch Error Box Extraction of combined Probe + RPL error box Cascading of the Error Boxes and Correlation to Their Fully Extracted Counterpart Conclusions 21

22 Cascading of Error Boxes Comparison Probe RPL Probe RPL T T Probe + RPL Probe T RPL Full error box obtained from Cascaded error boxes obtained from separate probe and RPL two-tier calibrations one two-tier calibration 22

23 Comparison of Error Boxes The cascaded error boxes of the probe and the RPL correlate to their full error box The difference of 0.2 db in the transmission is acceptable for the first approximation 23

24 Outline Introduction Extraction of the Microprobe Error Box Extraction of the Recessed Probe Launch Error Box Cascading of the Error Boxes and Correlation to Their Fully Extracted Counterpart Conclusions 24

25 Conclusions Two-tier calibration techniques applying SOL, SOLT and TRL calibration algorithms have been used for de-embeding of microprobes and RPLs The detrimental impact of the recessed probe launch dominates the impact of the microprobe with regard to the measurement bandwidth The correlation to the full error box indicates that the calibrations are applicable either in individual steps or in one combined step from 3.8 GHz up to 19 GHz Next step: Extension of the calibration bandwidth applying multiline TRL 25

26 Thank you for your attention! 26

27 References [1] Y. Kwark, C. Schuster, L. Shan, C. Baks, J. Trewhella, The Recessed Probe Launch A New Signal Launch for High Frequency Characterization of Board Level Packaging, IEC DesignCon Conference, Santa Clara, CA, [2] R. Rimolo-Donadio, C. Schuster, X. Gu, Y. Kwark, M. Ritter, Analysis and Optimization of Recessed Probe Launch for High Frequency Measurements of PCB Interconnects, DATE 2008, Munich, Germany [3], Xiaoxiong Gu, Young H. Kwark, Mark B. Ritter, Renato Rimolo-Donadio, Christian Schuster, Bandwidth Study of Recessed Probe Launch Variations for Broadband Measurement of Embedded PCB Structures, GeMic 2009 [4] T. Kutscha, K. Narita, T. Kaneko, T.Saeki, H. Tohya, Resonance Stub Effect in a Transition From a Through Via hole to a Stripline in Multilayer PCBs, IEEE Microwave and Wireless Components Letters, vol. 13, no. 5, May [5] G.F. Engen and C.A. Hoer, Thru-reflect-line: An improved technique for calibrating the dual six-port automatic network analyzer, IEEE Trans. Microwave Theory Tech., vol. 27, no. 12, pp , [6] Doug Rytting, IEEE MTT/ED seminar on Calibration and Error Correction Techniques for Network Analysis, OGI Center for Professional Development, 9/30/04 [7] J. A. Jargon and R. B. Marks, Two-Tier Multiline TRL for Calibration of Low-Cost Network Analyzers, 46th ARFTG Conference Digest, pp. 1-8, Dec [8] U. Arz, H. C. Reader, P. Kabos, and D. F. Williams, Wideband frequency-domain characterization of highimpedance probes, in 58th ARFTG Conf. Dig. San Diego, CA, Nov [9] Pictures from the Laboratory in Yorktown, IBM USA [10] GGB Industries, 27