Precision Time-Domain Reflectometry: Helping to solve today

1
October 2003

• Precision Time-Domain Reflectometry Helping to
  solve todays difficult signal integrity/transmiss
  ion problems

2
Agenda

• 1. Brief TDR review
• Some new things and some old things seen in a new
  way
• 2. Advanced calibration techniques unique to
  Agilent
• 3. New techniques for improved 2-event resolution
  and impedance accuracy
• 4. S-parameter results from the TDR
• Where to find out more New Application Note
  (see last slide)

3
1. What is TDR?

• Time domain reflectometry
• Analyze the quality of high-speed components and
  channels for transmission quality
• Are there any reflections due to impedance
  discontinuities?
• How big are they?
• Where are they?

4
TDR Launch a fast step into the DUT and measure
anything that reflects back
OSCILLOSCOPE
Er
ZL
TRANSMISSION SYSTEM UNDER TEST
STEP GENERATOR
Typical Step 200 mV, 250 kHzsquare wave with
35 ps rise time
5
What is TDR?

• Launch a fast pulse into the device under test
• Measure what reflects back from the DUT
• The size and polarity of any reflections
  indicates the magnitude of any discontinuity
• The time it takes for the reflection to return is
  used to indicate the location of any discontinuity

Transmission lines with changing impedance
Input pulse
Reflected pulse(s)
6
Displaying impedance in the Time DomainTDR
provides Instantaneous Impedance

• Typical TDR result
• A 50 Ohm cable
• B Launch to microstrip
• C 50 Ohm microstrip
• D 75 Ohm microstrip
• E 50 Ohm microstrip
• F open circuit

F
D
B
A
E
C
Compare to a network analyzer which provides
impedance as a function of frequency
7
2. Some important advantages of the 86100 TDR

• Would you buy a network analyzer without a
  calibration kit? No!
• Without calibration we are forced to rely
  completely on the raw performance of the
  instrument and have no ability to remove error
  causing mechanisms outside the instrument that
  are in the measurement path

DUT
8
Systematic TDR measurement errors can be removed
through simple calibration

• A simple concept By placing known reflections
  on the system, the measurement errors can be
  identified and removed
• Simple to perform Connect a short and a load at
  the reference plane
• Errors caused by cabling, attenuation etc. can be
  removed from the measurement
• Agilent is the only provider to use Normalization.

Test Fixture
Device Under Test
Blue Trace - Normalized Green Trace -
Standard Error 2.3 ohms
9
What arguments might you hear against this?

• We dont need fancy calibrations. We have a
  precision airline inside the TDR
• But what can you do for error mechanisms beyond
  the TDR output?
• Agilent doesnt need to do a calibration either,
  unless there is something beyond the TDR output
  that degrades the results (which, in real life,
  there almost always is)
• Calibration is a weak excuse for bad hardware
• Calibration techniques are a proven route to a
  better measurement
• Can you imagine doing network analysis without a
  good calibration process?

This is all explained in more detail in the new
Application Note (see last slide)
10
3. Some problems the industry faces..

• Data speeds are getting faster in electrical
  circuits
• Devices are getting smaller and more complex
• As edge speeds increase, more high-frequency
  energy is present
• More difficult to control impedance

11
The edgespeed of the TDR step sets two important
measurement levels

• The two-event resolution (how close can two
  reflections be and still be seen as separate
  events)
• Closely spaced reflections can get blurred
  together
• Two-event resolution set by material velocity and
  TDR system risetime

• How accurate is the measurement of the reflection
  magnitude
• As step speeds increase, more high frequency
  content
• Reflections often get worse
• Reflections for a 20 ps edge can be much larger
  than a 35 ps edge

12
A 35 picosecond step is insufficient to see
closely spaced reflections

• With a 35 ps step, all you know is the device is
  there
• If there is more than one reflection, we cant
  tell

35ps
13
High resolution allows your customers to see what
they could never see before
hermetic feedthrough
coaxial-microstrip launch
coaxial feed- through
9ps
V-connector pin-collette
V-connector pin-collette
microstrip transmission line

• At 9 ps step speed, we see 5 separate reflections
• Each event is easily seen and quantified

14
A faster step often yields a higher reflection
magnitude

• At 35 ps, the reflections look very small (52
  Ohms)
• At 9 ps the reflections increase to over 58 Ohms
• The 35 ps result isnt necessarily wrong and the
  9 ps right
• Test at an edge speed similar to how the device
  will be used
• Some examples
• 20 to 35 ps for 10 Gb/s
• 5 to 12 ps for 40 Gb/s

Designers working at the very high data rates or
with very small devices need a very fast TDR
15
How can I test faster than the 35 ps the TDR is
specified at? Two choices

• Electrically speed up the pulse
• Use external hardware to produce a much faster
  edge

• Digitally increase the edge speed through some
  signal processing
• Normalization calibration (discussed earlier)
  can use DSP to enhance the effective edge speed
• Can decrease the risetime to less than 20 ps

16
86100/Picosecond Pulse Labs 4020 Measurement
capabilities
The Picosecond Pulse Labs 4020 modules takes the
35 Picosecond pulse from the Agilent TDR and
increases the speed to under 9 picoseconds Two-ev
ent resolution is improved by a factor of 4!
(1.5 mm air, less than 1 mm in common
dielectrics)
lt9ps
17
Optimizing Measurements
54754A TDR module
86118A

• You will lose your edge speed if you have
• Excess or poor quality cabling to and from the
  DUT
• The scope receiver channel has insufficient BW
• Recommend TDR with the 86118A 75 GHz remote
  plug-in
• Max. bandwidth
• Minimum cabling distances

Sampling Port
4020 Remote TDR Head
Device Under Test
18
Configuring a system

• 86100 A or B mainframe (3.05 FW or above)
• 54754A TDR plug-in
• 86118A 70 GHz plug-in
• Lower BW channels can be used, but edgespeed and
  resolution will be reduced
• Cabling between the DUT and the receive channel
  degrades TDR speed
• Picosecond 4020 TDR or TDT enhancement module

19
Using the 4020 with TDRs that dont use
Normalization
TDR without Normalization 4020 pulse
86100 4020 pulse

• PSPL 4020 works with other TDRs
• Significant pulse aberrations. Cannot be
  calibrated out
• If pulse aberrations are not removed, they can be
  misinterpreted as close-in reflections
• 86100 TDR calibration significantly improves the
  4020 pulse quality
• Normalization also provides an excellent way to
  eliminate fixturing errors

20
4. Frequency domain analysis is critical for
completely understanding device performance
21
Benefits of S-parameter analysis
There is no fundamental difference in the
information content between the time domain and
the frequency domain Eric Bogatin Chief
Technical Officer GigaTest Labs

• Some things are just easier to see in the
  frequency domain
• Resonances
• Frequency response
• Device modeling can be more accurate with
  frequency domain data
• Some critical measurements of differential
  devices are better understood as a function of
  frequency

22
Changing the way high speed digital customers
view VNAs

• Everything you ever wanted to know about your
  device and more

VNA covers all combinations of in, out,
reflections, crosstalk. All combinations
contained in a 16 element matrix for a 4 port DUT.
23
For differential circuits, frequency domain
analysis helps even more
Balanced (Differential) devices can be analyzed
as pairs (Mixed-Mode S-parameters) rather than
single ended

• Differential are circuits becoming more important
  at high speeds
• Differential behavior can be much different than
  viewing each port individually (there is
  transmission line coupling designed in )
• Differential circuits reduce emissions and are
  less susceptible to radiation
• Like making the cross-talk work for you.

Port 2
Port 1
Differential Common

• Less far field emissions (crosstalk)
• More cancellation of incoming interference

24
What About Non-Ideal Devices?
Undesirable mode conversions cause emission or
susceptibility problems

• Differential-stimulus to common-response
  conversion

EMI Generation
Imperfectly matched lines mean the
electromagnetic fields of the signals are not as
well confined as they should be giving rise to
generation of interference to neighboring
circuits.

• Common-stimulus to differential-response
  conversion

EMI Susceptibility
Imperfectly matched lines mean that interfering
signals do not cancel out completely when
subtraction occurs at the receiver. Measured by
stimulating common-mode to simulate interference.
25
S-Parameters describe differential well.Four
quadrants of differential/common mode S
parameters S(response,stimulus,output,input)
26
Everything you ever wanted to know..
Frequency domain s-parameters can be used to gain
insight in frequency domain plots and time domain
views.
Mixed Mode S Parameters Displayed in the Time
Domain
Single ended S Parameters
Mixed Mode S Parameters
27
VNA, TDR, or both?

• All of the S-parameter data available using the
  Physical Layer Test System (PLTS) is now
  available using the 86100 TDR!
• N1930A
• Controls 86100 TDR
• Guided setup and calibration
• Automatic deskew
• Conversion of TDR data to complete S parameter
  results

28
True differential measurements

• Some TDRs make a big deal of producing both a
  negative and a positive step for doing
  differential TDR. True differential
• Agilent produces only positive pulses and then
  uses math to build a differential measurement

A differential system has coupled lines. The
electromagnetic fields will be very different for
two positive pulses. How can you get the right
impedance result if you dont have the correct
voltages present?
Agilents method provides differential, common
mode, cross terms, all with a single, accurate
setup. And this method simplifies the design to
allow almost perfect matching of the two positive
pulses giving the most accurate results..
29
There are very good reasons why we do what we do.

• We use superposition techniques to combine the
  results of multiple separate measurements
• I learned superposition in my first course in
  electronics. I believe it for circuits with
  wires and resistorsbut Im not sure about
  electromagnetics

From the classic text on electromagnetics,
Fields and Waves in Communications Electronics
by Ramo, Whinnery, and Van Duzer, (1965, John
Wiley and Sons) we read It is frequently
possible to divide a given field problem into two
or more simpler problems, the solution of which
can be combined to obtain the desired answer.
The validity of this procedure is based on the
linearity of the Laplace and Poisson equations.
That is   ?2(?1 ?2 ) ?2?1 ?2?2
  and   ?2(k?1 ) k?2?1   The utility of
the superposition concept depends on finding the
simpler problems with boundary conditions which
add to give the original boundary conditions.
Or..
30
Dont worry.its covered in the new Application
Note!

• Easy to read and digest
• TDR is valid only for linear passive devices (or
  active devices configured as linear and passive).
  So our technique is completely valid
• We could have built it with a and pulse.
  Important reasons we did what we did
• Allows almost perfect symmetry in the stimulus on
  each leg
• Asymmetry leads to mode conversion and
  potentially critical measurement errors

Linear, passive device to be tested
86100 channel steps overlaid almost exactly the
same
31
By the way..the VNA analysis is also based on
superposition

• The VNA can only stimulate one port at a time
• The VNA uses sinewaves (doesnt even use
  pulses!)
• Precision results obtained by combining results
  after taking several individual measurements
• No one ever questions the accuracy of a VNA.

True differential is best is a myth, and one
that is keeping you from making the most accurate
measurements.
32
Summary

• Review of TDR
• Why Calibration gives superior results
• Speeding the pulse up significantly for higher
  resolution
• Using frequency domain analysis from TDR data to
  get more insight
• Why true differential is better is a myth that
  may be keeping you from making the best
  measurements

• New literature
• Application Note High-precision Time Domain
  Reflectometry 5988-9826EN
• Flyer 86100 and the Picosecond 4020 lit
  5988-9825EN
• Accessories Flyer lit 5980-2933EN