ATLAS Insertion Loss Correlation Study vs traceable VNA
Application Note AP8186
The requirement of the ATLAS insertion loss correlation study was to assess the correlation between the following single ended and differential test methods:
(An additional objective was, in the case of the SPP method, to assess the correlation between the use of hardware Impulse Forming Networks (IFN) and a software emulated equivalent.)
Three PCB boards (test coupons) were manufactured using industry standard Gerber data generated with the Polar CGen Coupon Generator (V14.02.)
Each coupon contained the following test traces:
The test coupon is shown below. The coupon is divided into three sections, SPP, SET2DIL and impedance
The VNA portion of the work was conducted by the UK's National Physical Laboratory (UK equivalent of NIST) using fully traceable VNA equipment.
SPP, SET2DIL, SET2SEIL and impedance testing was performed with Polar ATLAS Insertion Loss Test System software (V14.04) using a Tektronix DSA8300 with 80E04 sampling modules and Polar ATLAS 800 ESD Protection Module. SPP tests were performed both with PicoSecond Pulse Labs 5208 Impulse Forming Networks (IFN) and with Polar software modelled IFNs.
Modelling information was provided by Polar Instruments Si9000e Insertion loss GHz PCB transmission line field solver modelling software (V13.02)
Cross sectioning was performed by Spur Electron Limited, Havant, Hampshire, UK.
Scanning electron microscope inspection cross section images of a test coupon are shown below.
SEM coupon cross section (trace height)
SEM cross section – detailed view (trace width)
Dimensions obtained from the cross sections were used in subsequent modelling in the Si9000e.
Raw data were transferred for comparison and presentation to Microsoft Excel. Spreadsheets were generated for both single ended and differential data and resulting graphs plotted.
The charts show measured data for VNA, SET2DIL and SPP, both with hardware IFNs (SPP in the charts) and with software modelled IFNs (charted as SPP No IFN.) Measured results for a typical coupon correlation test are shown below.
Raw data and regression are shown normalised to loss/10mm v frequency in GHz. Phase results are shown as rad/10mm v frequency in GHz.
VNA results are shown to 30 GHz, SPP and SET2DIL are shown to 20GHz (the sampling bandwidth limit of the DSA8300/80E04 TDR system employed.)
Raw data (loss/10mm v Frequency in GHz)
Regression data (loss/10mm v Frequency in GHz)
Phase (rad/10mm v frequency in GHz)
The charts of combined data show good correlation up to the measured bandwidth of 20GHz between VNA, SPP and SET2DIL. The data sets using the software emulated and hardware impulse forming networks show extremely good correlation, with the data sets in most cases overlying each other directly.