probes and setup for measuring power-plane impedances...
TRANSCRIPT
Istvan NovakPlane impedance measurement with VNA 1
DesignCon99, Feb. 99
Probes and Setup for Measuring
Power-Plane Impedances with
Vector Network Analyzer
Istvan NovakPlane impedance measurement with VNA 2
DesignCon99, Feb. 99
Outline• Introduction• Z, Y, and S parameters• Self and transfer impedances• VNA• One-port impedance measurement• Two-port impedance measurement• DUTs• Measured self and transfer impedances• Correlation to simulations• Resources• References
Istvan NovakPlane impedance measurement with VNA 3
DesignCon99, Feb. 99
Requirements in Power DistributionIn digital and mixed analog/digital systems:• Core and signaling voltages drop• Noise margin goes down• Core and total I/O current go up• Bandwidth goes up
Requirement:Few milliohms over hundreds of MHz bandwidth.
Solution:In multilayer boards, power and ground are
distributed over (solid) planes.
Istvan NovakPlane impedance measurement with VNA 4
DesignCon99, Feb. 99
Z, Y, and S Parameters
2221212
2121111
iZiZv
iZiZv
+=+=
= iZv
2221212
2121111
vYvYv
vYvYi
+=+=
= vYi
• Zero volt/current is hard to set• Calibration plane is critical
2221212
2121111
aSaSb
aSaSb
+=+=
= aSb
• Zero wave is easier to set• Calibration plane is not critical
Istvan NovakPlane impedance measurement with VNA 5
DesignCon99, Feb. 99
Which Parameters Do We NeedAt high frequencies, S parameters are easier to
measure, but• Digital designers deal with voltages and
currents• Good power-distribution network is a voltage
source >>> Z parameters needed
Ii
i j
Vi Vj 0
0
|
|
=
=
=
=
j
j
ii
jji
ii
iii
i
vZ
i
vZ
Istvan NovakPlane impedance measurement with VNA 6
DesignCon99, Feb. 99
Measuring Power-DistributionNetwork
• TDR
• LCR Bridge
• VNA Γ or S11, and S21
Istvan NovakPlane impedance measurement with VNA 7
DesignCon99, Feb. 99
What is a VNA• Tuned sinewave generator• Directional couplers• Tracking receiver(s)
Source
Receiver
DUT
Directionalcoupler
b1 a1
b2
Port1
Port2
01
221
01
111
2
2
|
|
=
=
=
=
a
a
a
bS
a
bS
Istvan NovakPlane impedance measurement with VNA 8
DesignCon99, Feb. 99
One-Port Self-ImpedanceMeasurement
50
5011 +
−=
in
in
Z
ZS
11
1111 1
150
S
SZZ in −
+==
The impedance is measuredbetween the ground and power
planes at the selected point
Istvan NovakPlane impedance measurement with VNA 9
DesignCon99, Feb. 99
Errors of One-Port Self-ImpedanceMeasurement
• VNA accuracy is lower at highreflections
• Connecting discontinuity is inseries of low-Z DUT
Istvan NovakPlane impedance measurement with VNA 10
DesignCon99, Feb. 99
VNA Error in S11 Measurement
• |S11| uncertainty of HP8720D is 1.5% at|Γ|~1 in the 50-2000MHz range
• Impedance uncertainty is 0.375 ohms• For low measurement errors, ZDUT must be in
the ohms range
But we want to measure fractions of an ohm
Istvan NovakPlane impedance measurement with VNA 11
DesignCon99, Feb. 99
Errors Due to Discontinuities
• 50 mils of pigtail connector/cable discontinuityis Lp ~0.4nH
• 0.4 nH is Zp ~ 2.4 ohms at 1GHz• Problem: Zp is in series to ZDUT
Lp
ZDUT
Port1 of VNA andcable: 50 ohm
VS
Istvan NovakPlane impedance measurement with VNA 12
DesignCon99, Feb. 99
Two-Port Self-ImpedanceMeasurement
Cable and Port2of VNA: 50 ohmZDUT
Cable and Port1of VNA: 50 ohm
VS
• S21 instead of S11 is measured• S21 uncertainty is less• Zp is in series to 50 ohms instead of ZDUT
Istvan NovakPlane impedance measurement with VNA 13
DesignCon99, Feb. 99
Two-Port Self-Impedance Reading
First-order calculation:Assume that• Lp ~ 0• ZDUT << Z0
ZDUT
Port1 and Port 2 ofVNA: 25 ohm
VS
ZDUT = Z11 = S21* 25 [ohm]
Istvan NovakPlane impedance measurement with VNA 14
DesignCon99, Feb. 99
Transfer Impedance Measurement
Cable and Port2of VNA: 50 ohm
DUT
Cable and Port1of VNA: 50 ohm
VS V1 V2
Istvan NovakPlane impedance measurement with VNA 15
DesignCon99, Feb. 99
Transfer Impedance Reading
First-order calculation:Assume that• Lp ~ 0• Z11 << Z0
• Z22 << Z0
• Z21 << Z0
Z21 = Z12 = S21* 25 [ohm]
Istvan NovakPlane impedance measurement with VNA 16
DesignCon99, Feb. 99
S21 Uncertainty
• |S21| uncertainty of HP8720D:
<1dB in the |S21| > -60dB range
<3dB in the |S21| > -70dB range
• Impedance uncertainty:
1dB (10%) for ZDUT > 25milliohms
3dB (40%) for ZDUT > 8milliohms
Istvan NovakPlane impedance measurement with VNA 17
DesignCon99, Feb. 99
Measurement Setup
DUT
Vector Network Analyzer: HP 4396 or HP 8720C VNAProbe: 2 pieces of 12-inch long semirigid coax
Istvan NovakPlane impedance measurement with VNA 18
DesignCon99, Feb. 99
Dual Semirigid Probe
Cross-sectional view of theDual Semirigid Probe
NOT TO SCALE
80 mils: 1.5% of λ at 1GHzCoax pigtail: Lp
Semirigidcoax 1
Semirigidcoax 2
31 milsParallel
Cu planes
Coax sleeve soldered all around to upper plane
Istvan NovakPlane impedance measurement with VNA 19
DesignCon99, Feb. 99
Device Under Test (1)31 mil plane separation
10-inch by 10-inch FR4 two-sided PCBBare PCB or 5.1 ohm/inch DET
Signal input and output: 12” Dual Semirigid Probe
Test points: on a 1-inch by 1-inch gridTop view:
Side view without and with DET:
805 SMD5.1 ohm per inch
Istvan NovakPlane impedance measurement with VNA 20
DesignCon99, Feb. 99
Device Under Test (2)2 mil plane separation
10-inch by 10-inch FR4 PCB
Bare PCB or 1 ohm/half-inch DET
Probes: 12” Dual Semirigid Probe
Test points: on a 1-inch by 1-inch gridTop view:
Side view with DET:
805 SMD1 ohm / half inches
Istvan NovakPlane impedance measurement with VNA 21
DesignCon99, Feb. 99
0.1G 1G 5G0.5G
Log frequency [Hz]Log frequency [Hz]
2.5
0.25
0.025
0.0025
Log impedance magnitude [ohm]Log impedance magnitude [ohm]
Residual Probe Response (1)
• DUT is not inplace
• Dual SemirigidProbe, 50-milpigtails,disconnected
Istvan NovakPlane impedance measurement with VNA 22
DesignCon99, Feb. 99
0.1G 1G 5G0.5G
Log frequency [Hz]Log frequency [Hz]
2.5
0.25
0.025
0.0025
Log impedance magnitude [ohm]Log impedance magnitude [ohm]
Residual Probe Response (2)
• Probe sleeves aresoldered to 31-mil DUT
• 50-mil probe tails inplace, but notconnected
ZL of 2-mil planes
ZL of 0.2-mil planes
Istvan NovakPlane impedance measurement with VNA 23
DesignCon99, Feb. 99
0.1G 1G 5G0.5G
Log frequency [Hz]Log frequency [Hz]
2.5
0.25
0.025
0.0025
Log impedance magnitude [ohm]Log impedance magnitude [ohm]
Measured Self Impedance of BareBoard at Center
• 31-mil DUT,edge open
• Dual SemirigidProbe at centerof planes
Istvan NovakPlane impedance measurement with VNA 24
DesignCon99, Feb. 99
0.1G 1G 5G0.5G
Log frequency [Hz]Log frequency [Hz]
2.5
0.25
0.025
0.0025
Log impedance magnitude [ohm]Log impedance magnitude [ohm] • 31-mil DUT,edge terminatedwith 40x5.1ohms
• Dual SemirigidProbe, at centerof planes
Measured Self Impedance ofBoard with DET at Center
Istvan NovakPlane impedance measurement with VNA 25
DesignCon99, Feb. 99
Measured vs. SimulatedSelf-Impedance of DUT
• 10” x 10” x 31mil FR4 with DET• Measured with HP8720C VNA• Simulated with 1-inch grid at
center node:Freq [Hz] Zmagn[ohm]
1.00E+08 0.31431.58E+08 0.4413
2.51E+08 0.65743.98E+08 1.2336.31E+08 1.744
1.00E+09 2.645 50M 100M 0.5G 1G 5G
2.5
0.25
0.025
Log magnitude of impedance [ohm]Log magnitude of impedance [ohm]
Log frequency [Hz]Log frequency [Hz]
Measured at center node
x x x x x x
Measured
Simulated
Istvan NovakPlane impedance measurement with VNA 26
DesignCon99, Feb. 99
0.1G 1G 5G0.5G
Log frequency [Hz]Log frequency [Hz]
2.5
0.25
0.025
0.0025
Log impedancemagnitude [ohm]
Log impedancemagnitude [ohm]
Measured Transfer Impedance
• 31-mil DUT, edgeterminated with 40x5.1ohms
• Two separate semirigidprobes, at middle of sideand at corner
-20
-40
-60
S21 [dB]S21 [dB]
Istvan NovakPlane impedance measurement with VNA 27
DesignCon99, Feb. 99
Measured vs. Simulated TransferImpedance of DUT
-40
-60
S21 [dB]S21 [dB]
-50
0.1G 1G0.5GLog frequency [Hz]
Zoomed responses:
MeasuredSimulated
Istvan NovakPlane impedance measurement with VNA 28
DesignCon99, Feb. 99
Self Impedance, 2-mil DUT
Self impedance magnitude [ohm]
0.001
0.01
0.1
1
1.E+05 1.E+06 1.E+07 1.E+08 1.E+09Frequency [Hz]
• 2-mil DUT, edgeterminated with80x(1ohm + 100nF)
• Dual SemirigidProbe, at center ofplanes
Istvan NovakPlane impedance measurement with VNA 29
DesignCon99, Feb. 99
Equivalent Circuit of ProbesConnection
Lp1 Lp2
Cable and Port2of VNA: 50 ohmDUT
Cable and Port1of VNA: 50 ohm
VS V1 V2
Lp1 Lp2
Coax and Port2of VNA: 50 ohmZ11=ZDUT
Coax and Port1of VNA: 50 ohm
VS
Self impedance:
Transfer impedance:
Istvan NovakPlane impedance measurement with VNA 30
DesignCon99, Feb. 99
S21 Conversion to Self Impedance
2121
2
2121
121 1
125
21
1
2 S
jS
Z
ZZS
ZSZ p
ii −+
∗∗≈+
−=
ωτ
Where Z1 = 50+jωLp1
Z2 = 50+jωLp2
τp = Lp/50
Istvan NovakPlane impedance measurement with VNA 31
DesignCon99, Feb. 99
S21 Conversion to Transfer Impedance
+
+∗
∗+
+∗∗
≈+
+
+
=
2
22
1
112
21
21
2
12121
2
22
1
11
121
11
2501
125
21
11
2
Z
Z
Z
Z
S
jS
Z
ZZSZ
Z
Z
Z
ZSZ
p
ji
ωτ
Where Z1 = 50+jωLp1
Z2 = 50+jωLp2
τp = Lp/50
Istvan NovakPlane impedance measurement with VNA 32
DesignCon99, Feb. 99
Recommended Resources
Hewlett Packard Vector Network Analyzers:• HP 8720 VNA• HP4396 VNACircuit simulator software:• Avant! HSPICE
Istvan NovakPlane impedance measurement with VNA 33
DesignCon99, Feb. 99
Conclusions
• Power-distribution network is characterized by self andtransfer impedances
• One-port measurements cannot handle low impedances
• 2-port VNA measurement introduced
• Probes: Dual semirigid coax with soldered pigtail
• Transmission-line grid is used to simulate parallel planes
• Good agreement between measured and simulated selfand transfer impedances was found
Istvan NovakPlane impedance measurement with VNA 34
DesignCon99, Feb. 99
References
[1] Dale Becker, Larry Smith, "Power Distribution Design for High Performance Systems,"Short course at the 1998 Topical Meeting on Electrical Performance of ElectronicPackaging, October 26-28, 1998, West Point, NY
[2] Understanding the Fundamental Principles of Vector Network Analysis, Hewlett PackardApplication Note 1287-1.
[3] I. Novak, "Reducing Simultaneous Switching Noise and EMI on Ground/Power Planes byDissipative Edge Termination," Proceedings of the 1998 Topical Meeting on ElectricalPerformance of Electronic Packaging, October 26-28, 1998, West Point, NY, pp.181-184.
[4] E. Leroux, Peter Bajor, “Modeling of Power Planes for Electrical Simulations,”Proceedings of the 1996 Wroclaw EMC Symposium, June 25-28, 1996.
[5] Henry Wu, Jeffrey Meyer, Ken Lee, Alan Barber, "Accurate Power Supply and GroundPlane Pair Models," Proceedings of the 1998 Topical Meeting on Electrical Performance ofElectronic Packaging, October 26-28, 1998, West Point, NY, pp.163-166.
[6] Personal communications with Rich Hoft, Hewlett Packard, Burlington, MA.