Summary of EOC demonstrator
submissionpresented by A. Kluge
CERN/PH-ESEMarch 31, 2009
A. Klugea, G. Dellacasab, M. Fiorinia, P. Jarrona, J. Kaplona, F. Marchettob, E. Martina,d, S. Martoiub, G. Mazzab, M. Noya, A. Cotta Ramusinoc, P. Riedlera, A. Rivettib, S. Tiuraniemia
a CERN, Geneva Switzerland, b INFN Torino, Italy, c INFN Ferrara, Italy, d UCL Louvain la Neuve, Belgium
Outline
General architecture
submission procedure
submission problems
A. Kluge
EOC column architecture
A. Kluge
Jitter-free pixel signal to TDC in EOC
time-to-digital converter TDC
buffering & read-out processor
amplifier&
discriminator/time-walk-
compensator
reference clock
A. Kluge
45 x 40 pixel final chip
Addr.
Addr.
Addr.
Hit Arbiter Hit Arbiter Hit Arbiter
LVDSH
it Reg1
Addr.
Hit Arbiter
45
4045 45 45 45
Ref CLK320MHz
serializerDLL Digital processing
Hit R
eg2
Hit R
eg1
Hit R
eg2
Hit R
eg1
Hit R
eg2
Hit R
eg1
Hit R
eg2
32
45 x 1 demonstrator
LVDSH
it Reg1
Addr.
Hit Arbiter
45
45
Ref CLK320MHz
serializerDLL Digital processing
Hit R
eg2
32
45 x 1 demonstrator
Hit R
eg1
Addr
Hit Arbiter
45
Ref CLK320MHz
serializer &LVDS
DLL
Hit R
eg2
32
Pileup
EO
C0
EO
C1
EO
C2
EO
C3
EO
C4
EO
C5
EO
C6
EO
C7
EO
C8
Pixel cell frontend
Jan 18-22, 2009A. Kluge
Pixel cell frontend
A. Kluge
Pixel cell frontend
A. Kluge
Preamp; buffered cascode (NMOS input transistor), resistive feedback (200k)
Gain;70mV/fC (25mV/fC at preamp output)
Preamplifier AC coupled to shaper and discriminator stages
Consumption; 190 µA/pixel (70 µA in analog section, 40 µA digital part of comparator, 80 µA line driver )
Time to Digital Converter
Tapped delay line 32 cells, 100ps
Two hit registers One per both
leading and trailing edge
5 bit encoding
DLL
hit rising Hit Register 1
Ref CLK
32 to 5 bit encoder5
t0,t1,t3,…,tN-2,tN-1
32 to 5 bit encoder5
hit trailing Hit Register 2
Optimisation of transmission line drivers – pre-emphasis
Optimisation of transmission line drivers
Pre-emphasis
Amplitude
Transmission of pixel hit
Jan 18-22, 2009A. Kluge
Transmission of pixel hit
Jan 18-22, 2009A. Kluge
hit Arbiter
Aynchronous latch logic which
defines first arriving pixel hit out of 5 in one group
latches hit address unambiguiously
during time-over-threshold time + TDC acquisition time: ~ 10 ns, address of additional (piled up) hits in same pixel group is stored
March 17, 2009A. Kluge
Read-out
mode pin selects between 45 pixel column or 7 pixel column
9 x 320 MHz data stream can be disabled individually to test dependency
on digital noise
read-out is continuous stream of data
Verilog description of ASIC block & deserializer available
March 17, 2009A. Kluge
Several design changes after October-design-review on system level and on building block level.
4 design areas: Analog front-end – pixel cell hit information driver/receiver – pixel cell & EOC TDC & DLL & hit Arbiter – EOC digital read-out block – EOC
Design changes were practically terminated 3 weeks before submission dated
March 17, 2009A. Kluge
4 design areas: Analog front-end – pixel cell hit information driver/receiver – pixel cell &
EOC TDC & DLL & hit Arbitrer – EOC digital read-out block – EOC
March 17, 2009A. Kluge
Layout –EOC 130µm
A. Kluge6.7 mm
2.8
mm
Layout –EOC 130µm
A. Kluge
Ana
logu
e t
est
str
uctu
res
Dat
a g
rou
pin
g &
pix
el a
dd
res
s
Receiver2x23
Receiver2x23
pad
s
EO
C
test pads
test
pad
s
1 folded column of 45 pixels
test pixels
6.7 mm
2.8
mm
test
pix
els
Layout –EOC 130µm
A. Kluge
Ana
logu
e t
est
str
uctu
res
Dat
a g
rou
pin
g &
pix
el a
dd
res
s
pad
s
sensor 4.14 x 5.37 mm2
EO
C
test pads
test
pad
s 1 folded column of 45 pixels
6.7 mm
2.8
mm
ASIC
2 rows ASIC pads layout
Sensor edge
125 µ(90 µ?)
50 µ
50 µ
125µ(90 µ?)
>300 µ
62 µ
88 µ
50 µ
2000
at least 900 µ
Submission issues
Problems with submission of automatically generated digital blocks (in LM) in custom layout (in DM)
Problems with use of library I/O Pads both for design rule checks (ESD) and LVS (Calibre) Elena (&Jan & Xavier) were working for 3!
weeks At the end (simpler) Angelo-custom made
pads were used
March 17, 2009A. Kluge
Submission
Chip has been submitted: final LVS OK (with parameter mismatch due to extraction
of gate around). LVS has been checked also after stream in (with some
problems in LVDS_TX due to hierarchy; solved by flattening of the transmitter cell).
The new pads gives many ESD errors both in assura and calibre; difficult to solve in short time (new pad library needed).
Antenna errors with assura not detected (clean), few antenna errors with caliber (related to input bump bonding pad and triple well (this error appears only when chip option is switched -> most probably bug in caliber deck))
March 17, 2009A. Kluge
Submission
LVS match on full chip
DRC check, waivers on ESD design rules
Layout optimisation is possible – long lines between building blocks
Further simulation allows optimisation of full custom TDC part with automatically generated parts
March 17, 2009A. Kluge
Summary
Ambitious demonstrator functionality compared to initial approach
Difficult implementation as design kit needs optimisation and technology is new to us
These problems must be cleared for final submission
However, all LVS, DRC done
Required pre-submission simulations have been doneA. Kluge