Light weight readout cable simulations for inner barrel pixel readout Cs. Soos, J. Christiansen, M....
-
Upload
amanda-holmes -
Category
Documents
-
view
219 -
download
2
Transcript of Light weight readout cable simulations for inner barrel pixel readout Cs. Soos, J. Christiansen, M....
Light weight cable simulations for inner barrel pixel readout 1
Light weight readout cable simulations for inner barrel pixel readout
Cs. Soos, J. Christiansen, M. Kovacs [email protected]
21/MAY/2015
2
Outline
Inner pixel readout reminder
Light weight cable options
36 AWG twisted shielded pair
Copper flex flat cable
Aluminium flex flat cable
Future work and conclusions
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Layout sketches and modularity
R=45mm
R= 320mm
Inner: 14 x 2*2 modules
Outer: 26 x 2*4 modules
2x2
2x4
2cm2cm
Pixel sensorPixel chip
PCB
CO2 pipe r=30mm
Thickness of pixel modules not to scale.
Beam pipe: 45mm
Pixel ROCs
Pixel sensor
Power + HV
PCB with passive components Readout
Heat distribution substrate
CO2 pipe
Links for layer1: 500KHzLinks for layer 2: 1MHZ
4 x 1 pixel module
2
2 8
4(1)2
2 Power9-12W
~2m8 x 1.2Gb/s4(1) x 320Mb/s9-12W
8cm
2.2 cm
1
Links for layer 4: 1MHzLinks for layer 3: 500KHz
Power
8cm
4.4 (4 active) cm
1
1
1
4 4
1
1
1
1
4 4
88
4 x 2 pixel module
16-24W~2m
8 x 1.2Gb/s8(1) x 320Mb/s16-24W
Disk layout to be updated
Jorgen Christiansen,
January 2015
Light weight cable simulations for inner barrel pixel readout21/MAY/2015 3
4
Cable options
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Cable optionWire size, diameter
Wire resistance
Mass for ~3500 cables
% in signal pair
% in shield/Gnd
% in insulator
36AWG Twisted pair, Cu, with shield
125um 2.7 ohm 5.8 kg 27% 40% 33%
36AWG copper pair, Cu, no shield
125um 2.7 ohm 3.5 kg 45% - 55%
Twisted pair Cu with Polyimide insulation
125um 2.7 ohm 1.8 kg 92% - 8%
Twisted pair,Cu cladded Alu, Polyimide insulation
125um Alu5um Cu
4.0 ohm 0.7 kg 83% - 17%
Kapton flat cable, Cu 35um gnd plane
140x35um2 6.9 ohm 4.0 kg 15% 55% 30%
Kapton flat cable, Cu 10um gnd mesh
140x35um2 6.9 ohm 1.5 kg 40% 10% 50%
Kapton flat cable, Alu35um gnd plane
140x35um2 11.5 ohm 2.0 kg 10% 32% 58%
Kapton flat cable, Alu 10um gnd mesh
140x35um2 11.5 ohm 1.0 kg 20% 5% 75%
* Red coloured ones were simulated Jorgen Christiansen, January 2015
5
S parameters
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Scattering parameters matrix is describing electrical behaviour of linear circuits. The matrix can be composed of several ports, in our case it is a 4 port element matrix.
By obtaining S parameters of a linear network, we can simulate it in most of the simulators.
S parameters can be measured by different instruments for example a Vector Network Analyser.
S parameters can be extracted from CAD designs using 3D solvers, for example transmission lines from a PCB design can be simulated before production.
The obtained S parameters can be exported to a file and can be used by chip designers or PCB designers to simulate and validate a design before production.
4 port network
Port 1
Port 3
Port 2
Port 4
6
FFE – Feed Forward Equalization
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Feed Forward Equalization is used to improve signal quality Used in high speed signal drivers when conventional drivers are not sufficient
In general it adjusts the waveform being injected into the channel to compensate for frequency-dependent losses suffered during propagation
The basic idea is to replace a single driver with a series of drivers, each one is delayed by a set amount from the previous one. These driver are called taps.
In Ansoft Designer the tap weights can be optimized and automatically calculated by an algorithm called Zero-Forcing Equalizer (ZFE)
7
36 AWG twisted shielded pair
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Ansys designer built-in twinax cable model was simulated Model is parametrized to match the geometry of an existing cable
A 2.7m length cable samples S parameters have been measured to compare results
A Vector Network Analyzer was used to obtain S parameters of the cable sample
8
36 AWG twisted shielded pair S parameters
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Built-in model’s S parameters with TAND= 0.008 loss dielectric and 2 m length
Insertion loss
Return loss
4 port network
Port 1
Port 3
Port 2
Port 4
9
36 AWG twisted shielded pair S parameters
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Built-in model’s S parameters with TAND= 0.008 loss dielectric and 2,7m length
4 port network
Port 1
Port 3
Port 2
Port 4
Insertion loss
Return loss
10
36 AWG twisted shielded pair S parameters
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
S parameters measured with VNA of a 2,7m twinax (blue) cable
Insertion loss
Return loss
4 port network
Port 1
Port 3
Port 2
Port 4
11
36 AWG twisted shielded pair Eye diagrams
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Eye diagrams simulated on 1.2Gbps, measurement vs simulationEye diagram simulation using FFE on built-in cable model, L=2.7m TD=0.008, 1.2Gbps
Eye diagram simulation on measured S parameters, L=2.7m TD=0.008, 1.2Gbps
Eye diagram simulation on built-in cable model, L=2.7m TD=0.008, 1.2Gbps
Eye diagram simulation using FFE on measured S parameters , L=2.7m TD=0.008, 1.2Gbps
12
36 AWG twisted shielded pair Eye diagrams
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Eye diagrams on built in model, 1.2Gbps VS 2.4Gbps VS Feed Forward EqualizationEye diagram simulated on built-in model, L=2 m TD=0.008, 1.2Gbps with FFE
Eye diagram simulated on built-in model, L=2 m TD=0.008, 1.2Gbps
Eye diagram simulated on built-in model, L=2 m TD=0.008, 2.4Gbps
Eye diagram simulated on built-in model, L=2 m TD=0.008, 2.4Gbps with FFE
13
Copper flex flat cable with 8 differential pairs
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Real size layout with Zero Insertion Force connector One cable contains 8 differential pairs, cable width is 8.5mm
Easy and light connector options
140µm line width, 180µm gap, 35µm copper thickness
S parameter model extracted by Ansys Siwave
More conductors can be added if needed
Reference planeSolder mask
Solder maskDielectric
180µm
140µm
14
Copper flat flex cable with 8 differential pairs
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Extracted S parameters of one differential pair from 2m copper flex cable design
Return loss
Insertion loss 4 port network
Port 1
Port 3
Port 2
Port 4
15
Copper flat flex cable with 8 differential pairs
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Eye diagrams on copper flex cable, 1.2Gbps VS 2.4Gbps VS Feed Forward Eq.Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 1.2Gbps
Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 1.2Gbps with FFE
Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 2.4Gbps with FFE
Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 2.4Gbps
16
Aluminum flat flex cable with 8 differential pairs
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Extracted S parameters of one differential pair from 2m aluminium flex cable
4 port network
Port 1
Port 3
Port 2
Port 4
17
Aluminum flex cable with 8 differential pairs
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Eye diagrams on aluminium flex cable, 1.2Gbps VS 2.4Gbps VS Feed Forward Eq.Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 1.2Gbps
Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 1.2Gbps with FFE
Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 2.4Gbps with FFE
Eye diagram simulated on extracted S parameters of flex circuit design, L=2 m TD=0.008, 2.4Gbps
18
Conclusions and future work
21/MAY/2015 Light weight cable simulations for inner barrel pixel readout
Carry out simulations on meshed plane flex cables and unshielded twisted pairs.
Future work:
The simulation results are well in line with measurement results.
Lightweight cables can be designed on flexible substrate with connectors embedded.
The simulated eyes are open on 1.2Gbps for almost all cable options without FFE.
In some cases signal quality requires Feed Forward Equalization on higher speeds.
Significant signal quality improvement can be achieved by using Feed Forward
Equalization.
Conclusions:
Feed-Forward EqualizationEqualization is one of the principal methods for improving the signal integrity of a channel. In general, equalization adjusts the waveform being injected into the channel to compensate for frequency-dependent losses suffered during propagation. For example, since many channels attenuate high-frequencies more than low frequencies, a simple form of equalization boosts high frequencies as the signal leaves the driver.Feed-forward equalization (FFE) is an extended form of frequency enhancement.The basic idea of FFE is to replace a single driver with a series of drivers, each one delayed by a set amount from the previous one. The delay is most commonly the unit interval (UI). These drivers are called taps. Each tap drives with a given strength, called the tap weight. The tap weights are set so as to reduce intersymbol interference (ISI). FFE can be applied either at the driver side or at the receiver side, although driver side FFE is more straightforward and easier to understand. This discussion applies to the driver side FFE.The algorithm for automatically calculating the tap weights in QuickEye and VerifEye is the Zero-Forcing Equalizer (ZFE). The ZFE algorithm is invoked when the FFE weights are not given but a nonzero number of taps is specified. The algorithm starts with the response to the channel of a pulse with width equal to one UI. (Neglecting jitter, the response of a channel can be considered to be composed of the response to appropriately-placed positive and negative half-height pulses.) A system of equations is set up, with the tap weights as the variables. The goal is to make the total response zero at the time points corresponding to the center of the eye for ISI for a number of bits equal to the number of FFE taps specified. Solving this system of equations yields the desired tap weights.In QuickEye and VerifEye, the tap weights are then applied to the step response calculated using the transient analysis engine of Nexxim. This equalized step response becomes the input to the rest of the eye diagram calculation.Figures 7 and 8 show QuickEye analyses without equalization and with FFE.
Decision-Feedback EqualizationAs in FFE, the number of DFE taps determine the number of unit intervals over which equalization is to operate. The decision-feedback equalizer keeps the results of the decision on the state of previous bits, then applies the weighted tap values to the previous bit waveforms to minimize ISI for the transition to the current bit state. The DFE weights can be automatically calculated using an algorithm similar to the algorithm used for FFE. While the FFE taps work towards canceling ISI both before the UI (precursors) and after the UI (post-cursors), DFE taps are limited to canceling post-cursors, because of the need to make a decision on a bit before its effect can be dealt with.Figure 11 shows the QuickEye analysis of the same high-speed serial channel shown in Figure 8, but with 4 taps of DFE.