FPIX2 The BTeV Pixel Readout Chip - PowerPoint PPT Presentation

1 / 26
About This Presentation
Title:

FPIX2 The BTeV Pixel Readout Chip

Description:

SINTEF sensors (160x18), Indium bumps. 3 out of 4 have obvious damage ... Contract with VTT for solder bumping should receive hybrids by Christmas. The End. ... – PowerPoint PPT presentation

Number of Views:30
Avg rating:3.0/5.0
Slides: 27
Provided by: DavidCh60
Category:

less

Transcript and Presenter's Notes

Title: FPIX2 The BTeV Pixel Readout Chip


1
FPIX2 The BTeV Pixel Readout Chip
  • David Christian Fermilab
  • Vertex 2003
  • September 18, 2003 Windermere, UK

2
Outline
  • Readout chip requirements.
  • Description of almost-final chip (FPIX2).
  • Bench tests of FPIX2.
  • Bench tests of hybrids (SINTEF sensor FPIX2).
  • Bench tests of irradiated hybrids
    (SINTEF sensor FPIX2).

3
Chip Designers
Abderrezak Mekkaoui Lead engineer (analog
design overall responsibility)
Jim Hoff Digital design
4
Readout Chip Requirements
  • Radiation hard.
  • Optimized for 132 ns between beam crossings (264,
    396).
  • Able to tolerate large sensor leakage current.
  • Very high speed zero-suppressed readout.
  • Lowest level trigger is pixel based ? all data
    from every crossing must be read out.
  • Spectrometer design luminosity 15 MHz
    interaction rate.
  • Pixel readout chip requirement minimal data
    loss at 45 MHz interaction rate.

5
Radiation Hard
  • 0.25m CMOS design started in 1998 verified in a
    series of small test chips.
  • Tested to 87 Mrad with no degradation in analog
    performance and only minor changes required to
    bias conditions.
  • Digital cells insensitive to total dose.
  • No latch-up, no gate rupture.
  • Single event upset cross sections measured,
    typically lt 10-15 cm-2 per bit.
  • Measured with 200 MeV protons about the same for
    MIPs.
  • Expected upset rates 10 pixel kill bits/hr, 2
    DAC register bits/hr, 1 data serializer bit/hr.
  • Will reload kill pattern each fill, monitor
    other registers reset as necessary (can be done
    without interrupting data taking for most errors).

6
Optimized for 132 ns crossing time
  • Timewalk is much less of a problem than at the
    LHC.
  • Can minimize (FE noise ? Disc. threshold
    dispersion) without trim DACs in every pixel.
  • Allows low discrimination levels (lt3000 e-)

7
Tolerate large leakage current
  • Separate control of return-to-baseline time and
    leakage current compensation.

8
Very high speed readout
  • Data kept in pixels until it is serialized and
    transmitted off chip no buffer memory.
  • Occupancy varies more than a factor of ten from
    the corner of the detector closest to the beam to
    the corners furthest from the beam ? Flexible
    serial data output interface allows the use of 1,
    2, 4, or 6 output links (each 140 Mbps).

9
FPIX2 Block Diagram
Pixel Unit Cells (22 columns of 128 rows each)
Core
Fabricated by TSMC (through MOSIS). Received
early 2003. Only bias voltages required are 2.5V
ground. All I/O is LVDS.
End-of-Column logic (22 copies)
Core Logic
DACs
Next Word Block
Clock Control Logic
Data Output Interface
Programmable Registers
Word Serializer
Programming Interface
Steering Logic
MCA/MCB (Readout Clock)
BCO Clock
Input/Output
High Speed Output
10
FPIX2 Layout
Debugging Outputs
Pixel array End-of-Column Logic e Core
128x22 Pixel array
End-of-Column Logic
Registers and DACs
Data Output Interface
Command Interface
LVDS Drivers and I/O pads
Internal bond pads for Chip ID
11
Pixel Unit Cell
12
Pixel Cells (four 50 x 400 mm cells)
12 µm bump pads
Preamp
2nd stage disc
Kill/ inject
ADC encoder
Digital interface
ADC
13
FPIX2 Version A,B and C
  • Three versions fabricated by TSMC, all with the
    same digital logic, but with different analog
    cells.
  • Version A
  • Unchanged from smaller test chips.
  • Version B
  • Same preamplifier as in version A.
  • Second amplifier stage and discriminator
    optimized for TSMC.
  • Version C
  • Preamplifier with higher gain (lower Cf)
    optimized for TSMC.
  • Added unity buffer between 1st and 2nd stage.
  • Modified 2nd stage and discriminators as in
    Version B.

14
Results at a glance
  • Version A works as expected.
  • Version B and C work almost as expected except
    for some oscillation problems when biased at
    nominal conditions. Origin of problem still under
    investigation.
  • Readout and control perform as expected in all
    versions.
  • Readout bandwidth of more than 840Mbps
    achieved.
  • No Xtalk observed. Analog performance
    independent of number of communication lines used
    (1,2,4 or 6).
  • The following test results are for
    non-hybridized chips (0 input cap).

15
Measured Noise
Expected to be 100 e- when connected to a sensor
pixel
Same amp. as Ver. A
16
Measured Threshold Dispersion
First prototype fabricated in CERN process had
sth 190 e-.
Amp. disc. optimized, preamp gain increased
(dynamic range reduced), sth 125 e-.
Disc. optimized for TSMC 231? 176.
17
Data Output
  • Data is driven off of a hit pixel onto the Core
    output bus, which is 23 bits wide. The data word
    consists of the information generated in the
    pixel unit cell (7 bit row number, and 3 bit ADC
    value), plus a 5 bit column number and an 8 bit
    BCO number, which are added by the end of column
    logic.
  • The Data Output Interface latches data from the
    Core output bus on the falling edge of the
    readout clock, serializes the data, and drives it
    off chip.

18
Output Data Format
  • Five bits are used to encode 22 columns. The
    column numbering scheme has no column number
    ending in 00. This ensures that a data word can
    never have 0s in b01 b13. This feature
    distinguishes a data word from a sync/status
    word.
  • Synchronization between the FPIX2 and the Pixel
    Data Combiner Board is established and maintained
    using the sync/status word. Whenever no data
    is available for output, the FPIX2 transmits the
    sync/status word. At least two sync/status words
    are guaranteed to be output every time the column
    number decreases. In addition, 23 bit hit data
    is transferred using a 24 bit word. The PDCB
    uses the word mark bit as a sync check on every
    word transfer.

19
RCLK and SCLK
 
  • The core readout clock (RCLK) is derived from the
    serial clock (SCLK). SCLK is constructed from
    external clocks and is nominally 140 MHz.
  • The frequency of RCLK depends on the number of
    output pairs being used. This relationship means
    that no buffer memory is required in the Data
    Output Interface.

35 MHz 4.6/132 ns
20
70 MHz clock
This result is from a smaller test chip (used for
dynamic SEU measurement).
140 Mbps 64 kbit random pattern after 50 feet of
flat twisted-pair cable.
21
FPIX2 Hybrids
  • 4 single chip hybrids recd from AIT.
  • SINTEF sensors (160x18), Indium bumps
  • 3 out of 4 have obvious damage to wire bonding
    pads.
  • The 4th FPIX2 (ver. A) oscillates unless almost
    all pixels are killed.

No Al on most pads
Al
22
Tesla p-spray sensors
  • Pixel copied from ATLAS.
  • Arrays match FPIX2 (128x22).
  • Modules for BTeV baseline design 4x, 5x, 6x,
    8x.
  • 600m pixels between chips.
  • No ganged pixels.
  • 1x modules some smaller.
  • Contract with VTT for solder bumping should
    receive hybrids by Christmas.

23
The End.
24
Damage to wire bonds
25
(No Transcript)
26
(No Transcript)
Write a Comment
User Comments (0)
About PowerShow.com