full waveform metrology: characterizing high-speed
TRANSCRIPT
![Page 1: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/1.jpg)
Full waveform metrology: Characterizing high-speed electrical
signals and equipmentPaul D. Hale
Andrew DienstfreyJack Wang
National Institute of Standards and Technology
Boulder, Colorado
The following slides are a product of the United States government and are not subject to copyright
![Page 2: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/2.jpg)
What are waveforms?
Transducer
Time
Physical stimulus
f (t)
IEEE Standard 181-2003, IEEE Standard on transitions, pulses, and related waveforms– Signal: A physical phenomenon that is a function of time– Waveform: A representation of a signal
• For example, a graph, plot, oscilloscope presentation, discrete time series, equation, or table of values.
– A waveform is obtained by some estimation process.
2
![Page 3: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/3.jpg)
Who cares?
Waveform generation and measurement used in all areas of science and technology
~760 oscilloscopes owned by NIST in 2004Wireless, wired, and optical communications Remote sensingChemical and materials properties~1 oscilloscope for every 4 NIST staff
>$1B annual market$500M market for high end oscilloscopes
Transducer
Time
Physical stimulus
f (t)
3
![Page 4: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/4.jpg)
Waveforms in the modern world
• Waveforms constitute the currency of modern, data-intensive communications.
• Digital signals require time-domain measurements
4
![Page 5: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/5.jpg)
5
High-speed sampling oscilloscopes
• Bandwidth: 20 to ~100 GHz
• Response pulse width ~ 4 to17 ps
• Need to deconvolve detector/scope response for pulses <16 to 70 ps [3% error, rss]
![Page 6: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/6.jpg)
6
Equivalent time sampling
Input to sampler:
TSampling period T > 10 μs
~ ~~ ~ ~ ~ ~ ~
![Page 7: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/7.jpg)
7
Why full waveform metrology? Oscilloscope response calibration
Old way: One number (parameter) to describe response
Bandwidth and transition duration
-0.25
00.9
0.1
Time
Ste
p re
spon
se
Transition duration(10 % to 90 %)
Shape?
One or only a few parameters do not uniquely describe response function
![Page 8: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/8.jpg)
0 0 0
0 0 1
0 1 0
0 1 1
1 0 0
1 0 1
1 1 0
1 1 1
Constructing an ‘eye’ pattern
8
![Page 9: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/9.jpg)
9
Time0
1
Digital signal test Different responses yield different results!
Forbidden region
Passes test0
1
Fails test
![Page 10: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/10.jpg)
10
A real measurement
Generator passed Generator failed
Same source, two different oscilloscopes, same bandwidth
Which measurement is correct?
![Page 11: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/11.jpg)
11
Full waveform metrology
1 1.1 1.2 1.3Time (ns)
-20
0
20
40
Volta
ge(m
V)
-8
-4
0
Mag
nitu
de(d
B)
0 4 8 12 16 20Frequency (GHz)
Time Domain
Frequency Domain
F
![Page 12: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/12.jpg)
12
Full waveform metrology
Vector signal analyzer (VSA)
Large-signal network analyzer (LSNA)OscilloscopeBit Error Ratio Tester
(BERT)
Pulse generator Vector signal generator (VSG) Lightwave component
analyzer (LCA)
Whole waveform metrology
Unified approach to• Calibration• Calculating and comparing uncertainty in time and
frequency domains• Deliver calibrated sources and waveform measurements
![Page 13: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/13.jpg)
Considerations for high-speed measurements: Time calibration
• Sample time errors are significant • Includes both random and systematic
contributions
13
![Page 14: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/14.jpg)
14
Timebase distortion
1 2 3 4 5 6 7 8 9 10……
1 2 3 45 6 78 9 10 11 12 13
Evenly spaced time samples
Actual sample timing
![Page 15: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/15.jpg)
Jitter
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
0
0.05
0.1
0.15
100 individual sweepsAverageTrue waveform
Time domain
Time (ns)
Vol
tage
0 20 40 60 80 10020
15
10
5
0
100 individual sweepsAverageTrue spectrum
Frequency domain
Frequency (GHz)
Pow
er (d
B)
FWHM: 7.4 ps → 7.7 ps Power @ 60 GHz: 0.7 dB error
2 212~ exp( )JLoss σ ω−Jitter averaged acts as lowpass filter⇒
15
![Page 16: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/16.jpg)
Waveform generator
Referenceoscillator
90° Hybrid coupler
S1
S2
S3
Time-base delay generator
SamplerdrivepulseSamplers
τ(1)
τ(2)
τ(3)
τ(tr)
τ(S1)
τ(S2)
τ(S3)
f
Oscilloscope
D
• Measure reference sine waves simultaneously with signal• Software fits sine wave data to model• Find time error at each sample (horizontal lines)• Correct for time error
Trigger
Timebase correction
![Page 17: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/17.jpg)
17
Timebase correction in action
3.9 3.95 4 4.05 4.1Time, ns
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Vol
tage
, V
Software available at http://www.nist.gov/eeel/optoelectronics/sources_detectors/tbc.cfm
![Page 18: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/18.jpg)
Considerations for high-speed measurements: Voltage calibration
• Voltage present at device terminals is a function of the load impedance– Dimensions of device are greater than a wavelength – Impedance of source and test equipment is not 50 Ω– Test fixtures and cables are lossy and may induce reflections
18
![Page 19: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/19.jpg)
Thévenin equivalent circuit
ZS
VT (ω)
Reference plane
ZLVL (ω)
LL T
L S
( )( ) ( )( ) ( )
ZV VZ Z
ωω ωω ω
⎛ ⎞= ⎜ ⎟+⎝ ⎠
Effect of impedance on measurement
19
![Page 20: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/20.jpg)
Definition of source wave bg
• bg is the amplitude of the wave that the source delivers to a matched load.
• e.g., power meters are calibrated to measure P = ½| bg |2
gΓ
1 0b =
1 ga b=
0Γ =
20
![Page 21: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/21.jpg)
Estimate bg from measured b2
2 2 La b= Γ
2 1b a=
( )
1 2 2 L 2 1
2 g 2 L g
g L g 2
and
1
b a b b ab b b
b b
= = Γ =⇒ = + Γ Γ
∴ = − Γ Γ
LΓ
1 g 1 ga b b= + Γ
1 2b a=
gΓ Oscilloscope
21
![Page 22: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/22.jpg)
Adding instrument response
2 2 La b= Γ
2 1b a=
LΓ
1 g 1 ga b b= + Γ
1 2b a=
gΓ Oscilloscope
h’Output
( )
( )
1 2 2 L 2 1
2 g 2 L g
L gg 2
L g1 1g 2
and
1
1 where
b a b b ab b b
b bh
b h b hh
− −
= = Γ =⇒ = + Γ Γ
− Γ Γ∴ =
′
− Γ Γ= =
′
oscilloscope response functionsystem response function
hh′ ≡≡
May want to solve for oscilloscope response h instead of bg22
![Page 23: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/23.jpg)
0 10 20 30 40 50Frequency (GHz)
-5
-4
-3
-2
-1
0
1
Res
pons
e (d
B)
Echo in frequency domain
~raw measurement
Estimate of h after correcting for mismatch and source response
Note that Γ
is measured and we do not have a model for any functions discussed in this talk
23
![Page 24: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/24.jpg)
0.6 1 1.4 1.8Time (ns)
-0.04
0
0.04
Raw measurementEstimated response (offset)
Echo is 28x below peak
Echo is reduced to 175xbelow peak
Time-domain echo
24
![Page 25: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/25.jpg)
Dissect two repeat measurements : Time domain
800 1200 1600 2000Time (ps)
0
0.4
0.8V
olta
ge (n
orm
aliz
ed)
Measurement 2Measurement 3
No obvious differences on this scale
25
![Page 26: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/26.jpg)
Dissect two measurements in the time and frequency domains
0 100 200 300Frequency (GHz)
-60
-40
-20
0
Res
pons
e m
agni
tude
(dB)
Measurement 2Measurement 3
-High dynamic range-No obvious differences on this scale
26
![Page 27: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/27.jpg)
Discrepancy due to contact resistance
0 10 20 30 40 50Frequency (GHz)
-5
-4
-3
-2
-1
0
Res
pons
e m
agni
tude
(dB)
Measurement 2Measurement 3
~0.5 dB difference at resonant frequency
27
![Page 28: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/28.jpg)
Time domain expanded
800 1200 1600 2000Time (ps)
-0.04
-0.02
0
0.02
0.04
Vol
tage
(nor
mal
ized
)Measurement 2Measurement 3
Low noise
Echoes
‘Stuff’ still happening here that’s larger than the noise level
28
![Page 29: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/29.jpg)
Difference between measurements
800 1200 1600 2000Time (ps)
-0.01
0
0.01
0.02
Volta
ge (n
orm
aliz
ed)
Difference voltageDifference is only ~1.4% of peak value but causes ~0.5 dB error at resonance frequency.⇒Small differences in one domain may mean big differences in the other domain.⇒We need errors <<1 %
29
![Page 30: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/30.jpg)
Considerations for high-speed measurements: Voltage calibration and dynamic response
• Instrument response bandwidth is comparable to the bandwidth of the unknown signal or system– Δftest /Δfsignal ~1/2 to 5
• Deconvolution of combined effects of instrument response, loss, and reflections is typicaly required
• Deconvolution often requires regularization
30
![Page 31: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/31.jpg)
Linear systems approach
( ) [ ]( )
( ) ( )
( ) [ ]( )
0 =
Using Fourier transform:
( ) ( ) ( )( ) ( ) ( )
y t x a t noise
a s x t s ds noise
y t x a t Y f X f A fX f Y f A f
∞
= ∗ +
− +
= ∗ ⇔ =
⇒ =
∫
31
![Page 32: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/32.jpg)
Least squares minimization
{ }
( )
2
1
LS
1
minn∈ℜ
−∗ ∗
−
−
=
=
xAx y
x A A A y
A y
⇒ “Normal equation”
Because our system is invertible
32
σ= +y Ax n
![Page 33: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/33.jpg)
0 100 200 300 400 500Frequency, GHz
-60
-40
-20
0
20
Mag
nitu
de, d
B
0 0.05 0.1 0.15 0.2Time, ns
-0.1
0
0.1
0.2
0.3
0.4Vo
ltage
, arb
.
A=System responseX=InputY=Output with noise
Convolution plus noise( )( )
*
3
2
10
ττ
=
=
xa
S
F33
![Page 34: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/34.jpg)
LS solution: Deconvolution without regularization
0 100 200 300 400 500Frequency, GHz
-60
-40
-20
0
20
Mag
nitu
de, d
B
A=System responseX=InputY=Output with noiseXLS=simple solution
34
Least squares
![Page 35: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/35.jpg)
LS solution is unstable
0 0.05 0.1 0.15 0.2Time, ns
-20
-10
0
10
20
Volta
ge, a
rb.
A=System responseX=InputY=Output with noiseXLS=simple solution
35
Least squares
![Page 36: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/36.jpg)
Control instability
• Introduces free parameter(s)• Want systematic, rigorous theory• May want to automate stabilized
(i.e., “regularized”) deconvolution
0 100 200 300 400 500Frequency, GHz
-60
-40
-20
0
20
Mag
nitu
de, d
B
A=System responseX=InputY=Output with noiseXregularized
36
![Page 37: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/37.jpg)
Constrained minimization
{ }
{ }{ }2
2
:
2 22
min
minn
M
λ
≤
∈ℜ
−
⇒ − +
x Lx
x
Ax y
Ax y Lx
Tikhonov framework
Lagrange multiplier
37
![Page 38: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/38.jpg)
Resulting inverse operators
( ) ( )
( ) ( )
21
1* 2 *
22 2 2
1* 2 * *2 2
M
M
λ λ
λ λ
−
−
= ⇒ ≤
= +
= ⇒ ≤
= +
L I x
x A A A y
L D D x
x A A D D A y
“Energy in x(λ) is bounded”
“Roughness ofx(λ) is bounded”
38
These terms serve as a low-pass noise filter.Choice of λ determines effective cut-off.
![Page 39: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/39.jpg)
L-curve
39
-6 -4 -2 0log||Ax(λ)-y||2
-12
-8
-4
0
4
log|
|D2x
(λ)||
2
2log y
212log −D A y
{ }{ }2 222min
nλ
∈ℜ− +
xAx y D x
λ increasing
Balance achieved in this region
2( ) 0λ − →Ax y
22log ( ) 0λ →D x
![Page 40: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/40.jpg)
-3.2 -3 -2.8 -2.6 -2.4 -2.2log||Ax(λ)-y||2
-3.4
-3.2
-3
-2.8
-2.6
-2.4lo
g||D
2x(λ
)||2
3 methods for λ
selection
L-curve method:Maximum curvature point
Discrepancy principle:( ) ( )2 2E Nλ σ σ− = =Ax y n
( ) 2*
*
min
.
x x
xλ
λ − - This is the "gold standard"
in our study because we know However, it can not be used in a measurement c
Optimal :
ontext.
40
![Page 41: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/41.jpg)
Simulation space
( )( )
( )( )
*
2 2* *
22
* *
2* *
2
Vary ratio of durations:
Vary signal to noise ratio:
Scales with for fixed amplitude
Scale to keep for every 1
NE
N
ττ
σσ
τ
σ
=
= =
•
•
⇒ =
xa
y y
n
x x
x y
T
S
T
S
41
![Page 42: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/42.jpg)
Simulation space
• Want to stress deconvolution algorithms– System response = 4th-order Butterworth– Input 2nd-order Bessel-Thompson
• Sample rate fixed, although this might also be an interesting parameter
• Simulated 1000 waveform instances for each (T , S)
42
![Page 43: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/43.jpg)
Conclusions from simulations
• RSS is a bad idea– Measurement can ‘speed up’ waveform– Pulse duration relationships cannot be basis for accuracy claims
• Tikhonov framework with these selectors is stable• Two selectors appear roughly equivalent
– L-curve slightly better, but more waveform shapes need to be studied
– Problems with L-curve not observed
43
![Page 44: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/44.jpg)
44
The calibrated waveform
• A waveform is a set of ordered pairs W={(tj , xj ) | j=1,…, J}• tj and xj are subject to error
• For numerical analysis, the waveform is interpolated to a vector X=(x1 , … , xN )T, where xn = x(tn ) and tn = t1 +nΔt
• Each element of X
is considered a random variable• Correlations may exist between some or all elements of X• The calibrated waveform is a vector Y
= f (X)
• How do we express the uncertainty in vector X?• How do we propagate uncertainty from vector X to vector Y?
![Page 45: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/45.jpg)
45
Transformation to ΣY
22 2 T y x
dfdx
σ σ⎛ ⎞= → =⎜ ⎟⎝ ⎠
Y XΣ JΣ J
What is the uncertainty in ?( )1, ( )TNy y f= =Y XL
( ) ( )( )
( )( )( )( )( ) ( )( )( )
i iij
j j
T
T
T
f
f yJx x
E
E
= ≈ + − +
∂ ∂= =
∂ ∂
= − −
≈ − −
≈
0 0
0 0
X X
Y 0 0
0 0
X
Y X X J X X
X
Σ Y Y Y Y
J X X J X X
JΣ J
K
Jacobian
Covariance
• f (X) can be a linear transformation, some quasilinear function, or a complicated algorithm
![Page 46: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/46.jpg)
46
Covariance matrix Σ
( )( )( ) ( )( ) ( )( )( )( )( ) ( )( )( ) ( )( )( )
( ) ( )( )( )( ) ( )( ) ( )( )( )
( )
21 1 1 1 2 2
22 2 1 1 2 2
2 2var
cov ,
where ( ) and ( ) pdf
i i i i
i j i i j j i j ij
E x E x E x E x x E x
E x E x x E x E x E x
x E x E x
x x E x E x x E x
E x xp x dx p x
σ
σ σ ρ
⎡ ⎤− − −⎢ ⎥⎢ ⎥
= − − −⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦
= − =
= − − =
= =∫
XΣ
L
L
M M O
![Page 47: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/47.jpg)
47
Stepping through oscilloscope calibration• Timebase correction• Interpolation to evenly spaced time grid• Transformation of time-domain waveform to frequency-
domain representation: VS = FvS
• Deconvolve system response:
• Note: Regularization may be required for inversion• Transformation back to time domain: a = F -1A• Scalar pulse parameters, i.e., pulse transition duration,
pulse amplitude, etc. σ2P = JP
ΣA
JPT
1
SPD S1-PDb
−⎛ ⎞
=⎜ ⎟Γ Γ⎝ ⎠A V
Oscilloscope calibration: Propagation of uncertainty
![Page 48: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/48.jpg)
Correlations distribute uncertainty where we expect it
48
0.6 1 1.4 1.8Time (ns)
-0.5
0
0.5
1
1.5R
espo
nse
Raw measurement (offset)Estimated responseUncertainty due to signal noiseUncertainty due to system response functionTotal uncertainty
0
0.001
0.0020.01
0.011
0.012
Unc
erta
inty
![Page 49: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/49.jpg)
49
Full waveform metrology includes:
1. Traceability to fundamental physics (EOS)2. Calibrated time t3. Calibrated voltage v(t)
– Account for reflections and loss– Dynamic instrument response
4. Covariance matrix-based uncertainty analysis– Analysis of the waveform at each point in the
measured epoch, along with uncertainty at each point
– Allows propagation of uncertainty through a linear transformation
– Fourier transforms, pulse parameters, etc.
![Page 50: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/50.jpg)
50
ReferencesNIST publications are available at the searchable web site:• http://www.eeel.nist.gov/publications/publications.cgi
Other useful web pages include:• http://www.nist.gov/eeel/optoelectronics/sources_detectors/measurements.cfm• http://www.nist.gov/eeel/electromagnetics/rf_electronics/high_speed_electronics.cfmOscilloscope calibration• D. Williams, P. Hale, K. A. Remley, “The sampling oscilloscope as a microwave instrument,”
IEEE Microwave Magazine, vol. 8, no. 4, pp. 59-68, Aug. 2007. • D.F. Williams, T.S. Clement, P.D. Hale, and A. Dienstfrey, “Terminology for High-Speed
Sampling-Oscilloscope Calibration,” 68th ARFTG Microwave Measurements Conference Digest, Boulder, CO, Nov. 30-Dec. 1, 2006.
• T. S. Clement, P. D. Hale, D. F. Williams, C. M. Wang, A. Dienstfrey, and D. A. Keenan, “Calibration of sampling oscilloscopes with high-speed photodiodes, ” IEEE Trans. Microwave Theory Tech., Vol. 54, pp. 3173-3181 (Aug. 2006).
• D. F. Williams, T. S. Clement, K. A. Remley, P. D. Hale, and F. Verbeyst "Systematic error of the nose-to-nose sampling oscilloscope calibration," IEEE Trans. Microwave Theory Tech., vol. 55, no. 9, pp. 1951-1957, Sept. 2007.
• A. Dienstfrey, P. D. Hale, D. A. Keenan, T. S. Clement, and D. F. Williams, “Minimum-phase calibration of sampling oscilloscopes,” IEEE Trans. Microwave Theory Tech., Vol. 54, pp. 3197-3208 (Aug. 2006).
![Page 51: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/51.jpg)
51
References
Time-base errors, jitter, and drift• Timebase correction (TBC) software and documentation at
http://www.boulder.nist.gov/div815/HSM_Project/Software.htm • J. A. Jargon, P. D. Hale, and C. M. Wang, “Correcting sampling oscilloscope timebase
errors with a passively mode-locked laser phase-locked to a microwave oscillator,” IEEE Trans. Instrum. Meas., vol. 59, pp. 916- 922, Apr., 2010.
• C. M. Wang, P. D. Hale, and D. F. Williams, “Uncertainty of timebase corrections,” IEEE Trans. Instrum. Meas., vol. 58, no. 10, pp. 3468-3472, Oct. 2009.
• P. D. Hale, C. M. Wang, D. F. Willaims, K. A. Remley, and J. Wepman, “Compensation of random and systematic timing errors in sampling oscilloscopes,” IEEE Trans. Instrum. Meas., Vol. 55, pp. 2146-2154, Dec. 2006.
• K. J. Coakley and P. D. Hale, “Alignment of Noisy Signals,” IEEE Trans. Instrum. Meas. Vol. 50, pp. 141-149, 2000.
• T. M. Souders, D. R. Flach, C. Hagwood, and G. L. Yang, “The effects of timing jitter in sampling systems,” IEEE Trans. Instrum. Meas., vol. 39, pp. 80-85, 1990.
![Page 52: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/52.jpg)
52
ReferencesMisc. applications of calibrated oscilloscope measurements
• A. Lewandowski, D. F. Williams, P. D. Hale, C. M. Wang, and A. Dienstfrey, “Covariance-matrix vector-network-analyzer uncertainty analysis for time- and frequency-domain measurements,” to appear in IEEE Trans. Microwave Theory Tech., 2010
• P. D. Hale, A. Dienstfrey, C. M. Wang, D. F. Williams, A. Lewandowski, D. A. Keenan, and T. S. Clement, “Traceable waveform calibration with a covariance-based uncertainty analysis,” IEEE Trans. Instrum. Meas., vol. 58, no. 10, pp. 3554-3568, Oct. 2009.
• H. C. Reader, D. F. Williams, P. D. Hale, and T. S. Clement, “Characterization of a 50 GHz comb generator,” IEEE Trans. Microwave Theory Tech., vol. 56, no. 2, pp. 515-521, Feb. 2008.
• K. A. Remley, P. D. Hale, and D. F. Williams “Magnitude and phase calibrations for RF, microwave, and high-speed digital signal measurements,” RF and Microwave Circuits, Measurements, and Modeling, CRC Press, Taylor and Francis Group, Boca Raton, 2007.
• D. F. Williams, H Khenissi, F. Ndagijimana, K. A. Remly, J. P. Dunsmore, P. D. Hale, C. M. Wang, and T. S. Clement, “Sampling-oscilloscope measurement of a microwave mixer with single- digit phase accuracy,” IEEE Trans. Microwave Theory Tech., Vol. 54, pp. 1210-1217 (Mar. 2006).
• K.A. Remley, P.D. Hale, D.I. Bergman, D. Keenan, "Comparison of multisine measurements from instrumentation capable of nonlinear system characterization," 66th ARFTG Conf. Dig., Dec. 2005, pp. 34-43.
• P. D. Hale, T. S. Clement, D. F. Williams, E. Balta, and N. D. Taneja, “Measurement of chip photodiode response for gigabit applications,” J. Lightwave Technol., vol 19, pp. 1333-1339, Sept. 2001.
![Page 53: Full waveform metrology: Characterizing high-speed](https://reader033.vdocuments.us/reader033/viewer/2022061101/629be7626ef6513ea47e5ccc/html5/thumbnails/53.jpg)
53
References
Electro-optic sampling• D. F. Williams, A. Lewandowski, T. S. Clement, C. M. Wang, P. D.
Hale, J. M. Morgan, D. A. Keenan, and A. Dienstfrey, “Covariance- based uncertainty analysis of the NIST electro-optic sampling system,” IEEE Trans. Microwave Theory Tech., Vol. 54, pp. 481-491 (Jan. 2006).
• D. F. Williams, P. D. Hale, and T. S. Clement, “Calibrated 200 GHz waveform measurement,” IEEE Trans. Microwave Theory Tech., Vol. 53, pp1384-1389 (April, 2005).
• D. F. Williams, P. D. Hale, T. S. Clement, and J. M. Morgan, “Calibrating electro-optic sampling systems,” IMS Conference Digest, p. 1473, May 2001.
• D. F. Williams, P. D. Hale, T. S. Clement, and J. M. Morgan, “Mismatch corrections for electro-optic sampling systems” 56th ARFTG Conference Digest, 141, Dec. 2000.