Upload
tarungupta79
View
227
Download
0
Embed Size (px)
Citation preview
7/28/2019 Lecture02 Link Over Environ 2up
1/15
EE290C Spring 2011
Lecture 2: High-Speed Link Overview and
Environment
Elad AlonDept. of EECS
EE290C Lecture 2 2
Keep in mind that your goal is to receive thesame bits that were sent
Most Basic Link
7/28/2019 Lecture02 Link Over Environ 2up
2/15
EE290C Lecture 2 3
Why Wouldnt You Get What You Sent?
EE290C Lecture 2 4
This is a 1
This is a 0
Eye Opening - space between 1 and 0
te
Ve
With voltage noise
With timing noise
With Both!
V0
V1
tb
Eye Diagrams
7/28/2019 Lecture02 Link Over Environ 2up
3/15
EE290C Lecture 2 5
BER
clk
BER = Bit Error Rate
Average # of wrong received bits / total transmitted bits
Simplified example:
(voltage only)
BER = 10-12: (Vin,ampl Voff) = 7n BER = 10-20: (Vin,ampl Voff) = 9.25n
,12
2
in ampl off
noise
V VBER erfc
=
EE290C Lecture 2 6
What About That Wire
Back plane connector
Line card trace
Package
On-chip parasitic(termination resistance and
device loading capacitance)
Line cardvia
Back plane trace
Backplane via
Packagevia
Back plane connector
Line card trace
Package
On-chip parasitic(termination resistance and
device loading capacitance)
Line cardvia
Back plane trace
Backplane via
Packagevia
[Kollipara, DesignCon03]
7/28/2019 Lecture02 Link Over Environ 2up
4/15
EE290C Lecture 2 7
ICs: usually use lumped models for wires
Capacitance almost always matters
Sometimes resistance
Less often inductance
Works because dimensions
7/28/2019 Lecture02 Link Over Environ 2up
5/15
EE290C Lecture 2 9
Links and Lengths
Cables connecting chips on two different
PCBs
Cables are lossy, but relatively clean if coax
Connector transitions usually the bad part
Distance: ~0.5m up to ~10s of m (Ethernet)
Data-rate: 1-10Gb/s
Wavelength in free space =
Wavelength on PCB (FR4) =
EE290C Lecture 2 10
Links and Lengths
High-speed board-to-board connectors
Daughtercard (mezzanine-type)
Backplane connectors
Distance: 8 up to ~40
Data-rate: 5-20Gb/s
Wavelength in free space =
Wavelength on PCB (FR4) =
7/28/2019 Lecture02 Link Over Environ 2up
6/15
EE290C Lecture 2 11
Transmission Lines Quick Review
Delay
Characteristic Impedance
Reflections
Loss
EE290C Lecture 2 12
Reflections
Sources of Reflections : Z - Discontinuities
PCB Z mismatch
Connector Z mismatch
Vias (through) Z mismatch
Device parasitics - effective Z mismatch
Z1 Z2
Z2 Z 1
Z 1 Z2+--------------------
2Z 2
Z 1 Z2+--------------------
DC via Conn via BP
(1) Energy conserved
(2) Voltages equal
7/28/2019 Lecture02 Link Over Environ 2up
7/15
EE290C Lecture 2 13
Skin Effect
At high f, current crowds
along the surface of the
conductor
Skin depth proportional to f-
Model as if skin is thick Starts when skin depth equals
conductor radius (fs)
Figure 2001 Bill Dally
EE290C Lecture 2 14
Skin Effect contd
100100MHz 500MHzMHz 500MHz 1GHz1GHz
W=210umt=28um
=6.6 um =2.08 um=2.95 um
Skin depth
7/28/2019 Lecture02 Link Over Environ 2up
8/15
EE290C Lecture 2 15
Dielectric Loss
High frequency signals jiggle
molecules in the insulator
Insulator absorbs energy
Effect is approximately linear
with frequency
Modeled as conductance term in
transmission line equations
Dielectric loss often specified
in terms of loss tangent
Transfer function =
Table 2001 Bill Dally
DLengthe
EE290C Lecture 2 16
Dielectric Loss contd
FR4 cheapest most widely used
Rogers is most expensive high-end systems
May not matter that much due to surface roughness
8 mil wide and 1 m long 50 Ohm strip line
-40.0
-30.0
-20.0
-10.0
0.0
1.E+06 1.E+07 1.E+08 1.E+09 1.E+10
Frequency, Hz
Attenuation
FR4
Roger 4350
Kollipara DesignCon03
7/28/2019 Lecture02 Link Over Environ 2up
9/15
EE290C Lecture 2 17
Skin + Dielectric Losses
Skin Loss f Dielectric loss f: bigger issue at high f
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.0 E+06 1. 0E+0 7 1. 0E+08 1.0 E+09 1. 0E+10
Attenuation
Frequency, Hz
FR4 dielectric, 8 mil wide and 1m long 50 Ohm strip line
Total loss
Conductor loss
Dielectric loss
Kollipara DesignCon03
EE290C Lecture 2 18
Everything Together: S21
S21: ratio of received vs. transmitted signals
Breakdown of a 26" FR4 channel with 270 mil stubs
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 .0E+0 0 5.0E +08 1.0E+09 1 .5E+0 9 2.0E +09 2.5 E+09 3 .0E+0 9 3.5E +09 4.0 E+09
Frequency, Hz
Transferfunction
PCB traces
PCB traces & connectors
PCB traces, connectors & vias
Entire channel
7/28/2019 Lecture02 Link Over Environ 2up
10/15
EE290C Lecture 2 19
Real Backplane
EE290C Lecture 2 20
Practical PCB Differential Lines
Differential signaling has nice properties
Many sources of noise can be made common-mode
Differential impedance raised as f(mutuals) between
wires
Strong mutual L, C can improve immunity
W S
H
H
SW
r
H
+ -
- Strip Strip-line
7/28/2019 Lecture02 Link Over Environ 2up
11/15
EE290C Lecture 2 21
Coupling Crosstalk
Near-end xtalk: NEXT (reverse wave)
Far-end xtalk: FEXT (forward wave)
NEXT in particular can be very destructive
Full swing TX vs. attenuated RX signal Good news: can control through design
NEXT typically 3-6%, FEXT typically 1-3%
EE290C Lecture 2 22
NEXT: What Not To Do
X
X
X
X
X
X
X
X
Tx Rx Tx
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 100 200 300 400 500 600 700 800 900
Time, ps
Voltage,
V Tx
Rx
XTX
7/28/2019 Lecture02 Link Over Environ 2up
12/15
EE290C Lecture 2 23
NEXT: Better Design
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 100 200 300 400 500 600 700 800 900
Time, ps
Voltage,
V Tx
Rx
XTX
X
X
X
X
X
X
X
X
Tx
Rx
EE290C Lecture 2 24
Connectors Particularly Tough
NEXT FEXT
55 ps (20-80%) 55 ps (20-80%)
80ps (10-90%) 80ps (10-90%)
AB 4.4% 3.7%
DF 3.3% 2.6%
GH 3.3% 2.6%
JK 4.3% 3.5%
Tight footprint constraints
Hard to match pairs and even individual lines
May compensate skew on line card
Also big source of impedance discontinuities
7/28/2019 Lecture02 Link Over Environ 2up
13/15
EE290C Lecture 2 25
Skew Within Link
Need very tight control to maintain constant % of bit
time
1% skew on 30 line 50ps skew Half of a bit time at 10Gb/s
Good news: connectors relatively short (~200ps)
EE290C Lecture 2 26
Reflections RevisitedTX
DATA
RX
DATA
AT
AR
CR
CT
D
B
-8
-6
-4
-2
0
2
4
6
8
10
gh-ghconn. (baseline):NormalizedRaw andeq pulse response:PRlength after
main60
A T,R
A2 T,R
B
C T,R D
-8
-6
-4
-2
0
2
4
6
8
10
gh-ghconn. (baseline):NormalizedRaw andeq pulse response:PRlength after
main60
-8
-6
-4
-2
0
2
4
6
8
10
gh-ghconn. (baseline):NormalizedRaw andeq pulse response:PRlength after
main60
A T,R
A2 T,R B
C T,R D
T
Connector-BP
transitions
7/28/2019 Lecture02 Link Over Environ 2up
14/15
EE290C Lecture 2 27
Reflections Due To Via Stub
0 2 4 6 8 10
-60
-50
-40
-30
-20
-10
0
frequency [GHz]
Attenuation[dB]
9" FR4,via stub
26" FR4,via stub
26" FR4
9" FR4
Stub: extra piece of T-line hanging off main path
Usually leads to resonance (notch) Especially on thick backplanes, vias are a big culprit
EE290C Lecture 2 28
Minimizing Via Stubs Thinner PCB?
counter-
boredblind
via
All expensive: 1.1-2x
Better vias?
7/28/2019 Lecture02 Link Over Environ 2up
15/15
EE290C Lecture 2 29
Summary
Packaging, chip connection, etc. can all have an effect
Entire conferences dedicated to signal integrity (SI)
EE290C Lecture 2 30
Implications
Need to know range of channels you will face Drives design of the link circuitry
Start diving in to that next lecture
Dont be a pure circuit weenie
Simple fixes to channel may go a long way
FR-4 BP, Length: 20", T/S: 30/270 mil
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 .0 0 0 .5 0 1 .0 1 1 .5 1 2 .0 1 2 .5 2 3 .0 2 3 . 5 2 4 .0 2 4 .5 3 5 .0 3 5 .5 3
frequency, GHz
Transferfunction
meas
sim
Roger BP, Length: 1.5", T/S: 30/270 mil
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0.0 0 0 .7 8 1.56 2.33 3.11 3 .89 4. 67 5. 45 6.22 7 .0 0
Frequency, GHz
Transferfunction(s21)
meas
sim