### A VNA Manifesto: A Primer for Practical Mastery

```A VNA Manifesto: A Primer for Practical Mastery
Day 4: Application Topics of S-Parameters
Wild River Technology LLC
www.wildrivertech.com
Alfred P. Neves
[email protected]
phone 503 679 2429
Slide 1
Wild River Technology LLC
De-embedding with T-matrix approach
TRL calibration
Passivity and Causality – practical tests
S-parameter work flow
Rational Compact Modeling
To fix or not to fix a bungled S-parameter
Slide 2
Wild River Technology LLC
T-matrix
Create S-parameter from scaled T-line
Port Extension
Slide 3
Wild River Technology LLC
I1
I2
+
A B
C D1
V1
-
V1
I1
V2
I2
I3
+
A B
C D2
V2
V3
-
I1
A B
=
C D
A B
=
C D
×
1
×
2
V2
I2
V1
V3
I1
A B
=
C D
I3
Slide 4
Wild River Technology LLC
A B
×
C D
1
×
2
V3
I3
Measure a de-embedding structure and get sparameter
Model something and get S-parameter
Use VNA firmware to de-embed with file from active
measurement using T-matrix de-embedding
Matlab is another option for T-matrix approach
Slide 5
Wild River Technology LLC
TRL Calibration, On-Board
Lines
1,2,3
THRU
Slide 6
Open
Wild River Technology LLC
TRL Calibration is finicky
Launch must be decent (low S11, no resonance) –
good launch design
Launch Connector Repeatability from SMA to SMA –
TDR must be very good
Line lengths accurate – layout, etch
Impedance variation across board low – etch, fiberweave, etc., can wreak havoc
We measure the LINE standards group delay, then
use those group delays in Cal Kit, also verify
impedance
Slide 7
Wild River Technology LLC
Establish a Concerted Calibration Verification
Ideal TRL THRU S11=S22=0, S12=S21=1
TRL calibrated THRU
0
20
-10
0
dB(S(2,1))
dB(S(1,1))
dB(S(2,1))
dB(S(1,1))
THRU - SOLT calibration
-20
-30
-40
-20
-40
-60
-50
-80
-60
0
2
4
6
8
10
12
14
16
18
20
0
4
6
8
10
12
freq, GHz
freq, GHz
Slide 8
2
Wild River Technology LLC
14
16
18
20
Verification of Group Delay of THRU includes noise
and moving average, perfect group delay=0psec
Group Delay THRU in psec
avggroupdelay
group_delay_picoseconds
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
0
Slide 9
2
4
6
8
10
12
freq, GHz
Wild River Technology LLC
14
16
18
20
Without delving into all the details, I don’t use TRL calibration for
general signal integrity work.
There are better calibrations to get reference plane near DUT,
especially for multi-port 4,6,.. 12 port structures
Slide 10
Wild River Technology LLC
• Verification includes
– THRU or Insertion Response using insertable
– return loss using wideband terminators
– symmetry, S11=S22
– Reciprocity, S21=S12
– Group Delay
Slide 11
Wild River Technology LLC
Again, use validation structure features to your
Slide 12
Wild River Technology LLC
Slide 13
Wild River Technology LLC
adapter THRU for non-insertable and flush THRU with
insertable Cal. KF-KF adapter has approximately 0.1dB
insertion loss and 50psec delay.
Slide 14
Wild River Technology LLC
Verify Low-Frequency Calibration. Passivity issues!
Slide 15
Wild River Technology LLC
Example of Simple THRU
for TRL:
Simple check of obvious Passivity
Violations. The check of |S21|>1
is NOT sufficient however!
Another example
using on-board THRU
Slide 16
Wild River Technology LLC
Analysis of Calibration
Causality Using Polar Plot,
Causal S-parameters should
only rotate clockwise
Quick Polar view of Insertion
locates non-causal behavior
Slide 17
Wild River Technology LLC
Cal verification using precision airlines
First, examine transmission aberrations, and return loss
I may check this
S21 with no
calibration
enabled
Slide 18
Wild River Technology LLC
Secondly, use Resonant structure like Beatty Standard.
Why?
Slide 19
Wild River Technology LLC
Simple Matlab routine to analyze for Beatty - Symmetry issues
Slide 20
Wild River Technology LLC
For 4 Ports:
After Cal Verification,
Assess S-parameter of
DUT
• Overall return loss
• Symmetry
• Group Delay Distortion
• Reciprocity
• Insertion Loss
variation
• Quick Time Domain
Transform
Slide 21
Wild River Technology LLC
Rational Compact Modeling of Sparameters
Slide 22
Wild River Technology LLC
Be definition it fixes passivity and causality
Is very accurate, low residual MSE error
No interpolation issues, continuous functions
Works in Spice engines directly
Required intermediate step for Time Domain
Simulation
Slide 23
Wild River Technology LLC
Slide 24
Wild River Technology LLC
Stripline Resonator, CMP-28
Slide 25
Wild River Technology LLC
Validate calibration based on DUT
Make measurement of external NIST standard,
such as stepped impedance or 50ohm airline
Try to predict what DUT measurement will look like
based on length, resonances, pathologies, etc.,
Make measurement on on board standard, such as
Channel Modeling Platform
Analyze measurement
Import measurement into SI Tool, obtain quality
metrics – quality, passivity, causality
RCM model
Slide 26
Wild River Technology LLC
Don’t fix S-parameters, the tools on the market don’t
work very well.
If a tool reports problems, check calibration
Sometimes you have to live with and manage
issues when using T-matrix, or partial calibration, or
fancy de-embedding.
See DesignCon2012 tutorial: High-confidence Sparameter Measurement Methodologies for 1528 Gbps, it is available WRT website.
Slide 27