Issue link: https://iconnect007.uberflip.com/i/396835
14 The PCB Design Magazine • October 2014 this context means analysis-to-measurement correlation is observed, not just for one or two structures (test coupons for instance), but rath- er for a broad range of typical interconnects: single-ended and differential, stripline and mi- crostrip, simple planar and with the vertical transitions or vias, etc. Such comparison should be done consistent- ly both in frequency (magnitude and phase of S-parameters) and time (TDR and eye diagram) domains. In other words, the system- atic validation or benchmarking is needed to make sure that the board is manufactured as designed, measurements are taken properly and, finally, that the interconnect analy- sis software provides accept- able accuracy. It is a whale of a project, if you do it your- self from scratch. Though, the process can be facilitated if you start with a readily available validation platform such as CMP-28/32 from Wild River Technology [2] . The platform was designed to illustrate and facilitate sys- tematic analysis-to-measure- ment validation process at 28–32 Gbps and to demon- strate that interconnects for 28–32 Gbps can be predict- ably designed. Use of such a platform saves time and lowers the risks by benchmarking against known and already measured and simulated structures. The CMP-28 platform [2] and Simbeor electromagnetic signal integrity software [3] are used here to illustrate signal integrity software validation process for 28 Gbps interconnects. The validation process can be divided into three steps: 1. Measure S-parameters with VNA up to 50 GHz and qualify them with formal quality met- rics and, optionally, compare with S-parameters measured by an expert. 2. Identify or confirm broadband dielectric and conductor roughness models. 3. Simulate all test structures with the iden- tified material models and verified board design adjustments and compare with the measured data in frequency and time domains. Validation platform A validation platform is a very important tool for signal integrity software benchmarking or for- mal pre-qualification. Accuracy and limitations of the software can be easily identified with the analysis to measurement comparisons on a typical set of interconnect structures. A validation plat- form can be either developed in-house or purchased from a vendor. One of the industry- first validation platforms was the physical layer reference design board (PLRD-1) from Teraspeed Consulting Group [4] . Use of the PLRD-1 revealed the need and enabled devel- opment of the industry-first broadband dielectric and con- ductor roughness models in Simbeor software. Another ex- ample of validation platform is the CMP-28/32 channel mod- eling platform from Wild River Technology [2] . Both CMP-28 and 32 versions contain 27 mi- crostrip and stripline intercon- nect structures. All structures are equipped with either 2.92 mm (CMP-28) or 2.4 mm (CMP- 32) connectors to facilitate accurate measurements with a VNA. The CMP-28/32 plat- form is shown in Figure 1 and will be used here to demonstrate the systematic approach to the analysis to measurement correlation. The CMP-28/32 platform contains multiple single-ended and differential line segments, suitable for identification or confirmation of material models, and also serve as the simplest validation structures. It also contains practical stripline and microstrip link paths with vias and crosstalk. In addition, it has a set of resonant structures to validate analysis of t-lines with dif- ferent widths and do validation for highly re- flective interconnects [2, 5] . SINk OR SWIM AT 28 GBPS continues feature The CMP-28/32 platform contains multiple single-ended and differential line segments, suitable for identification or confirmation of material models, and also serve as the simplest validation structures. It also contains practical stripline and microstrip link paths with vias and crosstalk. " "