Copyright © 2014 Wild River Technology LLC Slide 1 Wild River Technology LLC Alfred P. Neves phone A VNA Manifesto: A Primer for Practical Mastery Day 3: What Constitutes a 'Good' VNA Measurement?

Copyright © 2014 Wild River Technology LLC Slide 2 Calibration Calibration comparison Reference plane defined Device geometries A way out of the “I can’t get a good calibration” nightmare Port definitions VNA setup – IF BW, smoothing, sampling points Example of using simple normalization

Copyright © 2014 Wild River Technology LLC Slide 3 Calibrations are not perfect, since both the standards are not perfect, and the models are also not perfect Residual error is the post calibration error and can be used to calculate measurement uncertainty Residuals are source,load match, and directivity

Copyright © 2014 Wild River Technology LLC Slide 4 THRUSHORTOPEN MATCHED © H. Heck S 11 (Short) S 11 (Open) S 11 (load) S 21/12 (Thru)

Copyright © 2014 Wild River Technology LLC Slide 5 Calibration – Removal of unwanted measurement portion using known standards SOLT Short-Open-Load-Thru TRL Thru-Reflect-Line De-Embedding – post process removal by mathematically removing fixture artifact with known response of fixture T-matrix de-embedding Error Correction Normalization, gating

Copyright © 2014 Wild River Technology LLC Slide 6 SOLT - Short, Open, Load, THRU Unknown THRU - Short, Open, Load, Reciprocal THRU –reciprocal is a reciprocal Unknown THRU for a non-insertable calibration Electronic aided cal - uses classes of impedances or switches and cal kit elements TRL - THRU, Reflect, Line De-Embedding-Port Extension, T-matrix

Copyright © 2014 Wild River Technology LLC Slide 7 From: High-confidence S-parameter Measurement Methodologies for Gbps, DesignCon2012

Copyright © 2014 Wild River Technology LLC Slide 8 Understand two general types of Calibration –Insertable means I can connect measurement cables (includes an adapter if used in calibration) –Non-Insertable requires an adapter, that was NOT part of calibration –Cal Kits elements are defined by their connector type and sex Cal kit THRU, 2.92mm female to female

Copyright © 2014 Wild River Technology LLC Slide 9

Copyright © 2014 Wild River Technology LLC Slide 10 ABC Where is Ref Plane– A and C? B? A or C?

Copyright © 2014 Wild River Technology LLC Slide 11 Reference Plane

Copyright © 2014 Wild River Technology LLC Slide 12 A Flush THRU: 0psec group delay 0db loss 0deg phase Perfect symmetry

Copyright © 2014 Wild River Technology LLC Slide 13 DUT S11 S21 S12 S22Edf Esf Erf Elf Etf1 a1 b1 b2 a2 Error box ref

Copyright © 2014 Wild River Technology LLC Slide 14 An uncalibrated THRU response does not have an established location called Reference Plane. The absolute magnitude and phase may not be useful but they can indicate if a connection has failed or if there is a hardware issue.

Copyright © 2014 Wild River Technology LLC Slide 15 Port Definitions Selecting Fstart, Fstop, Number of Points Marker Function Smoothing, Averaging, IF Bandwidth

Copyright © 2014 Wild River Technology LLC Slide 16 1st Establish port definitions. This is an important step. Two options: DUT Port 1 Port 2 Port 3 Port 4 DUT Port 1 Port 3 Port 2 Port 4

Copyright © 2014 Wild River Technology LLC Slide 17 Modal Decomposition Method Port 3Port 1 Port 4 Port 2 physical ports logical ports Port 1Port 2 DUT Port3Port 1 Port 4 Port 2 physical ports logical ports Port 1Port 2 DUT

Copyright © 2014 Wild River Technology LLC Slide 18 IF BW 1kHz 4 port SOLT Calibration, THRU paths 1 to 2, 3 to 4 No averaging, no smoothing Fstart 25MHz, Fstop 40GHz, 1600 points IF BW 300Hz 4 port SOLT Calibration, THRU paths 1 to 2, 3 to 4 No averaging, no smoothing Fstart 10MHz, Fstop 40GHz, 4000 points VNA set up is done before calibration and is part of that calibration, don’t change later!

Copyright © 2014 Wild River Technology LLC Slide 19 Averaging is a feature that reduces the effects of random noise on a measurement. The VNA computes each data point based on the average of several measurements. Both Averaging and IF BW can be used for the same benefit of general noise reduction. For minimizing very low noise, Averaging is more effective than reducing IF bandwidth. Averaging takes slightly longer than IF bandwidth reduction to lower noise, especially if many averages are required

Copyright © 2014 Wild River Technology LLC Slide 20 Smoothing establishes a moving average over a window and serves a useful function of eliminating spike like noise (group delay) Avoid using smoothing if at all possible since smoothing degrades measurement accuracy depending on the aperture size selected. Smoothing aperture is a "sliding window" or frequency range over which the smoothing algorithm collects data a wider aperture produces less noise, AND less accuracy. We use Smoothing for analysis of y=mx+b linear plots, like Group Delay and Unwrapped Phase

Copyright © 2014 Wild River Technology LLC Slide 21 I use IF BW of 1kHz for most work, 300Hz for finalizing compliance measurements like on backplanes Don’t use smoothing, it gets you into trouble, especially if it alters the Touchstone. Check your VNA firmware I don’t use averaging either. For the limited dynamic ranges typical of SI work I don’t think it is needed.

Copyright © 2014 Wild River Technology LLC Slide 22 Your latest 10Gbpsec SERDES is being characterized and you believe it can handle 25dB of loss in loopback (TX to RX) at Nyquist (5GHz) without DFE and RX CTLE, just TX preemphasis. Use ZVA50 VNA to determine how much transmission line length you need. You do want to not include return loss impact. Will work this out: dB of Loss/inch/GHz

Copyright © 2014 Wild River Technology LLC Slide 23 Measuring 7.25inch at 5GHz marker -2.78dB loss Measuring 22inch at 5GHz marker -8.18dB loss =5.4dB, to a first order launch impact is de-embedded Net length difference = =15.25inches (5.4db of loss)/15.25inches/5GHz for 1GHz= 0.071dB/inch/GHz

Copyright © 2014 Wild River Technology LLC Slide 24 Loss of this microstrip line for this stackup and material is: 0.071dB/inch/GHz Checking result with 49.5inch transmission line (.072db/in/GHz)X49.5inchX5GHz = 17.82db Loss, which is close to a measured 17.98dB

Copyright © 2014 Wild River Technology LLC Slide 25 Exercise: Use Markers to determine frequency distance, start with REF marker Estimate DK of Microstrip based on S11 suck-outs frequency interval

Copyright © 2014 Wild River Technology LLC Slide 26 DUT S11 S21 S12 S22Edf Esf Erf Elf Etf1 a1 b1 b2 a2 General Signal Flow Diagram Next, let’s review what is a reference plane (see Red dashed Lines). Don’t get hung up on the Flow Diagram details yet! The location to which the calibration or de-embedding has been performed (i.e., where the S-parameters are most accurate).

Copyright © 2014 Wild River Technology LLC Slide 27 Port 1 Port 2 connector

Copyright © 2014 Wild River Technology LLC Slide 28 Port 1 Port 2 connector A B C

Copyright © 2014 Wild River Technology LLC Slide 29 Provides calibration when THRU is available Does not correct for some errors such as source mismatch (E SF ) and directivity (E DF ) Will first consider flow diagrams of Short and Thru 0 0 a1 b1 b2 a2 For a reflect signal flow graph, Gamma is either +1 (an open), or -1 (a short)

Copyright © 2014 Wild River Technology LLC Slide 30 DUT S11 S21 S12 S22Edf Esf Erf Elf Etf1 a1 b1 b2 a2 General Signal Flow Diagram

Copyright © 2014 Wild River Technology LLC Slide 31 1 a1 Etf DUT S21 DUT

Copyright © 2014 Wild River Technology LLC Slide 32 Recall that this measurement reference planes are at SMAs S21 includes loss of SMA, 1.75inch trace on each side of non-insertable DUT

Copyright © 2014 Wild River Technology LLC Slide 33 AA BB

Copyright © 2014 Wild River Technology LLC Slide 34 PROCEDURE: Measure DUT, including test fixture Measure THRU (no DUT) In dB scale, subtract THRU from DUT Normalization: corrects for loss and phase delay of test fixture will not correct for resonances, launches and structures need to be matched should only be used when test fixture and DUT have low return loss

Copyright © 2014 Wild River Technology LLC Slide 35

Copyright © 2014 Wild River Technology LLC Slide 36

Copyright © 2014 Wild River Technology LLC Slide 37 Red trace is TRL calibrated Offset resonator, blue is Normalized

Copyright © 2014 Wild River Technology LLC Slide 38 Red trace is TRL calibrated Offset resonator, blue is Normalized

Copyright © 2014 Wild River Technology LLC Slide 39 Port 1 Port 2 connector A B C

Copyright © 2014 Wild River Technology LLC Slide 40

Copyright © 2014 Wild River Technology LLC Slide 41 Questions, discussion? Alfred P. Neves