TRIPt bench measurements

1
TRIP-t bench measurements
Leo Bellantoni D0 AFEII Directors Review
2
The TRIP-t

• The TRIP-t is the custom mixed-signal ASIC at the
  heart of the AFEII board
• Input is the VLPC-amplified pulse, plus timing
  signals
• Outputs 32, multiplexed are
• Digital discriminator pulse for triggering
• Analog O1V signal giving amplitude of VLPC
  pulse
• Analog O1V signal giving time of VLPC pulse
• Analog pulses through 48 bucket pipeline
• Serial programming interface to set operating
  parameters via DACs
• Test injection circuit

3
Design goals

• 100ns integration window typical, adjustable from
  50 to 150ns
• A-pulse adjustable gain saturation 150fC at
  highest gain, 3000fC at lowest
• t-pulse adjustable gain to cover integration
  window
• Noise lt 1fC referred to input
• Analog pulse outputs uniform to ?3, 7 bit
  precision on amplitude, 2ns on time
• 99 of discriminator turnons within ?4fC width
  lt1fC
• Discriminator thresholds 4-20fC typical in
  tracker for preshower detector, want as high as
  practical
• Power lt 320mW

4

• A set of DAC values and timing diagrams for the
  control lines that feed into the chip that are
  close to final have been found and used.

5
Frontend
6
Digital Discriminator outputs
99 ? 2.6s 2.6 x 0.6fC lt 4fC
32 chans
lt1fC
Turn-on curves fit (integrated) Gaussians nicely
7
A-pulse outputs
32 chans
Differential gain, ?V/?Q
Its in the SPICE simulation too
Not linear at low outputs
8
A-pulse outputs
?V/?Q
Problem occurs in 2-FET output buffer at
output of the analog pipeline.
9
A-pulse outputs
Swapping n-channel FETs for p-channel FETs fixes
it in simulation
7.6 bit accuracy
10
A-pulse
0.8 gain uniformity
55ns flat-top (?3 less gain spread) for 100ns
gate
11
t-pulse outputs
32 chans
A 2nd effect of the non-linearity Will set ?V/?t
to keep out of this region (2-20 mV/ns range) -
this is final parameter setting issue Space for
pedestal gain already in database - both AFE
on-board pulser and tracks available
for calibration
Residual for 50 ? t ? 110ns 1ns RMS
12
Time Walk

• ?t / ?Q ? 55 mV/fC, corresponding to ? 44 ps/fC
• T_OUT has large pulse-to-pulse fluctuations below
  20fC ? 2 p.e. over threshold

A MIP at 90? corresponds to about 50fC (8
photoelectrons collected _at_ 40k gain)
13
Design goals

• 100ns integration window typical, adjustable from
  50 to 150ns - Operated at 100ns w/no problems
• A-pulse adjustable gain saturation 150fC at
  highest gain, 3000fC at lowest - Obtained
  90/2900fC ? Workable
• t-pulse adjustable gain to cover integration
  window - OK out to 300ns
• Noise lt 1fC referred to input - Measured lt0.4fC
• Analog pulse outputs uniform to ?3, 7 bit
  precision on amplitude, 2ns on time - Amplitude
  gains lt1, linearity needs to be fixed. Time
  gains OK, 1ns residuals
• 99 of discriminator turnons within ?4fC width
  lt1fC - Looks good, needs more stats
• Discriminator thresholds 4-20fC typical in
  tracker for preshower detector, want as high as
  practical - Operated at several settings between
  5 and 10fC
• Power lt 320mW - Measured 265mW

14
Conclusion

• A prototype TRIP-t has been extensively bench
  tested and good operating parameters have been
  found. Further tests in quantity and with wider
  temperature variations are planned.
• It is fully functional and with a simple
  modification should perform as required. The
  chip designer describes the modification as
  minimal risk.
• Well need 5000 good packaged chips, and plan a
  joint submission with the FPIX chip on 20 May.
  (One version)
• Chip packaging has been worked out, and there
  is even some documentation.