Transcript Presentazione di PowerPoint

New approach to CPC design
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
1
It is my 1-st presentation
from LNF/INFN (Frascati)
The scientific work is still under development
and the new ideas have to be tested…
Triggered by P.Campana
Why results on cross-talks obtained with
small chamber in May 02 are somehow better
than with the large M0 prototype tested in October 02?
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
2
I’d like to add:
?
?
2 problems observed in beamtests
have to be explained and suppressed:
- rather high cross-talks from wires to pads;
- double and multiple TDC spectra
Problems mentioned here were observed already
in M2R1 and other prototypes built at CERN,
as well as in Ferrara’s prototypes;
50% crosstalks observed in some conditions in M3R3
in October 2002 beam-tests at operational HV
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
3
Cross-talks
9 December 2002
?
A.P.Kashchuk (LNF/INFN), Frascati)
4
Cross-talks
along the wires (longitudal)
perpendicular to wires (transverse)
Longitudal crosstalks
are less studied
and much less suppressed
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
5
According to specification (LHCb 2000-061, W.Riegler)
Cross-capacitance Cwp will be increased with pad size, as shown:
M1
M2
R3
M3
M4
M5
LNF M0
Pad size (cm) &
Number of pads
Cdet (pF)
Cwp (pF)
CPC:
2,5
(48x4)
CPC:
2.5,12.5
(24x2)
CPC:
2.7,13.5
(48x2)
CPC
5.8x14.5
(24x2)
CPC:
6.2x15.5
(24x2)
30.3
4.75
50
15
56
21.5
90
40
140
46
Cwp=0.475hw (pF) where h, w are pad height and width (cm)
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
6
Let’s measure:
(very good agreement with table shown above)
Cwp=20.5pF (small chamber)
Charge on pad
(fC)
Cwp measured in large M3R3 is absolutely similar
60000
40000
20000
0
y = 20,493x - 446,41
0
500
1000 1500 2000 2500
Eq.Volt.Step on strip (mV)
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
7
The original idea was following:
Cross-talks from wire strips to pads
(longitudal) will be reduced with grounding
wire strips through HV-capacitors
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
8
RC-model (Rwire =90 Ohm/m)
Only capacitive coupling
is taken into account
in this model
Wire strip with 4 wires
Signal from particle
Assumed that
width of wire strip
is equal to cathode pad
Cathode
pad
current
source
HV-capacitor
grounded on one side,
as shown
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
9
Wire strip as a transmission line, i.e. LC-model
(first proposed by LNF group)
Cathode
pad
current
source
Each wire can be considered as a transmission line
Wires in strip are connected in parallel:
L reduced, C increased (product LC is the same)
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
10
Recently it has been found:
the wire strip is ringing
(response of wire strip in small LNF prototype made with injector)
HV-capacitor 680pF directly grounded
f=95 MHz
The ringing frequency depends on inductance in series
to HV-capacitor (what the reason?)
f=60 MHz
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
11
Cross-talk profile from wires to cathode pads in the large
M3R3 prototype
Volt. step on strip
(strip is floating)
20ns/div
Pad-1 (15%)
Central Pad
Ratio 2-nd/1-st peak 20%
Ringing 18 ns
Pad+1 (20%)
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
12
Fine waveform structure at voltage rise time 1.5ns:
One can see 4 ns oscillation
High frequency is due to LC of the transmision line itself
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
13
Response of wire strip in small LNF prototype
made with Current Injector
Ringing 13ns
Strip is grounded through 680pF,
with adding inductance
period is increased
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
14
Equivalent circuit
1.Inductance blocks
HV-capacitor effect
Wire strip
Terminated end
Cathode pad
High cross-talk:
Cwp
Qpad  Qstrip 
Cstrip

2.Another parasite effect
f ringing 
9 December 2002
1
2  Ltrace  Cequivalent
Compare to ideal case:
Cwp
Q pad  Qstrip 
CHV
Assumed
Cstrip  CHV But!!!
Cwp  C pad  CHV
Cequivalent  Cstrip
A.P.Kashchuk (LNF/INFN), Frascati)
15
What LC-model shows?
with HV-capacitors grounded
at inductance 3nH in series (perhaps, can be acheaved)
1-side strip termination with 0 Ohm
1.0uA
2-side (peak less factor 2)
1.0uA
0.5uA
0A
0A
-0.5uA
-1.0uA
0s
10ns
20ns
30ns
I(R7)
Time
Peak=1uA
Ringing= 8ns
9 December 2002
40ns
50ns
-1.0uA
0s
10ns
20ns
30ns
40ns
50ns
I(R7)
Time
Peak=0.5uA
Ringing= 3ns
A.P.Kashchuk (LNF/INFN), Frascati)
16
Stray inductance (printed traces) in series
to HV-capacitors and full capacitance of the wire strip
mainly specify the ringing frequency
Green – 3nH
Red – 300nH (can be if width of traces 0.25mm, see M3R3)
1-side strip termination with 0 Ohm
2-side termination
2.0uA
2.0uA
0A
0A
-2.0uA
0s
20ns
40ns
60ns
I(R7)
Time
Green peak=1uA
Red peak=1uA
Ringing=30ns
9 December 2002
80ns
100ns
-2.0uA
0s
20ns
40ns
60ns
80ns
100ns
I(R7)
Time
Green peak=0.5uA
Red peak=1uA
Ringing=20ns
A.P.Kashchuk (LNF/INFN), Frascati)
17
Correct strip termination with 377 Ohm
can not be used
1.0uA
1- signal at
far end
to capacitor

R0 
 377 Ohm

0A
-1.0uA
-2.0uA
0s
10ns
20ns
30ns
40ns
50ns
I(R7)
Time
1.0uA
0A
2- middle
-1.0uA
-2.0uA
0s
10ns
20ns
30ns
40ns
50ns
30ns
40ns
50ns
I(R7)
Time
1.0uA
0A
-1.0uA
3- near end
2-side termination
through 680pF
At 1-side termination
amplitude will depend
on signal position along
the strip (see next slides)
-2.0uA
0s
10ns
20ns
I(R7)
Time
No ringing, waveforms are independed to position of the
signal source, but the highest crosstalks will be in this case
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
18
Inductance in series to HV-capacitors
(1-side termination with 680pF and 50 Ohm)
Same schematics
No ringing , but not enough cross-talk attenuation
Peak independ on Lstray due to 50 Ohm
3
2
1
2.0uA
1.0uA
2.0uA
2.0uA
1.0uA
1.0uA
0A
0A
0A
-1.0uA
0s
10ns
20ns
30ns
I(R7)
Time
40ns
50ns -1.0uA
0s
10ns
20ns
30ns
40ns
-1.0uA
0s
50ns
Time
Green=3nH
Red=100nH
9 December 2002
10ns
20ns
30ns
40ns
50ns
I(R7)
I(R7)
Time
Scale +/-2uA
10ns/div
A.P.Kashchuk (LNF/INFN), Frascati)
19
Inductance in series to HV-capacitors
(1-side termination with 680pF and 0 Ohm)
Same schematics
Better attenuation, but ringing at R=0
1
2
3
2.0uA
2.0uA
2.0uA
0A
0A
0A
-2.0uA
0s
10ns
20ns
30ns
40ns
-2.0uA
0s
50ns
10ns
20ns
30ns
I(R7)
I(R7)
Time
40ns
-2.0uA
50ns
0s
Time
Green=2nH/Ringing 200MHz
Red=100nH/Ringing 70MHz
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
10ns
20ns
30ns
40ns
50ns
I(R7)
Time
Scale +/-2uA
10ns/div
20
Inductance in series to HV-capacitors
(2-side termination with 680pF and 20 Ohm)
Same schematics
Good cross-talk attenuation factor at 20 Ohm
High inductance leads to ringing even at R=20 Ohm
and drastically reduces cross-talk attenuation
3
2
1
2.0uA
0A
2.0uA
2.0uA
0A
0A
-2.0uA
-2.0uA
-2.0uA
0s
10ns
20ns
30ns
I(R7)
40ns
50ns
0s
10ns
20ns
30ns
40ns
50ns
I(R7)
Time
0s
10ns
20ns
30ns
40ns
50ns
I(R7)
Time
Time
Red=100nH/Ringing 100MHz
Green=3nH/No ringing
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
Scale +/-2uA
10ns/div
21
Inductance of printed trace (example):
h
L  4 10   l
w
7
if
w=0.25mm trace width (in M3R3 prototype)
h=1.5mm pcb thickness
l=3-10cm length of trace (in M3R3 prototype)
then L=100-1000nH Tringing  20 60ns @ C=100pF
if
w=1.2cm
then L=5nH can be achieved
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
22
Ringing on wire strip can double/multiple signals
and TDC spectra
Threshold defined experimentally:
for wire readout –7fC
for cathode (single) – 5fC
FEE noise 50e/pF is not the first reason for threshold choice,
mainly cross-talks define threshold, at efficiency 95%/gap
Dynamic range of signals in CPC is large (100)
Average signal 50fC
So, high probability for after-pulsing can be found
at bad wire strip termination and imperfect layout in CPC
at any frequency of ringing (it depends on design)
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
23
If Cstrip=Cwire is high?
2.0uA
0A
+/-2uA
-2.0uA
0s
20ns
40ns
60ns
80ns
100ns
80ns
100ns
I(R7)
Time
200nA
Excellent result
0A
+/-0.2uA
-200nA
-400nA
0s
20ns
40ns
60ns
I(R7)
Time
No ringing and high cross-talk attenuation
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
24
Conclusion
!
CPC design can be improved following the way:
No ringing must be on the wire strips at perfect design
1.No wire segmentation is needed in CPC design one of the effective way
No HV-capacitors and resistorrs (cheaper and much easy design), wires are connected
to one HV-resistor.
Minimisation of the trace inductances has to be done in the Combined readout
chambers, which dumps effect of low impedance.
2.Double Cathode Readout scheme, perhaps, can be used in some cases below M3R3
also effective way at large Cwires (it allows increase threshold at fixed HV)
already tested in M1R1 with excelent results (because very low Cwp and good attenuation
of the cross-talks from wires)
9 December 2002
A.P.Kashchuk (LNF/INFN), Frascati)
25
Summary
Voltage zero must be on the wire strips at perfect CPC design, i.e.
M1
M2
M3
Cwire  
M4
M5
CPC
5.8x14.5
(24x2)
CPC:
6.2x15.5
(24x2)
LNF M0
R3
Pad size (cm) &
Number of pads
CPC:
2,5
(48x4)
CPC:
2.5,12.5
(24x2)
Double Cathode readout (CRO) below M3R3. No wire segmentation
Cdet (pF)
Cwp (pF)
9 December 2002
30.3
4.75
50
15
CPC:
2.7,13.5
(48x2)
?
56
21.5
A.P.Kashchuk (LNF/INFN), Frascati)
Single CRO. No wire segmentation
90
40
140
46
26