Upload
egbert-ball
View
213
Download
0
Embed Size (px)
DESCRIPTION
Agilent Pulse Fitting Apply Gaussian fit to both functions Measure the time difference between the two fit means => actually a factor sqrt(2) better High timing resolution for a fixed amplitude over a small timing window Gaussian Fit 6.4 psRMS
Citation preview
Update on the New Developments & Analysis from Hawaii
Larry Ruckman & Gary VarnerInstrumentation Development Laboratory (ID-Lab)
University of Hawaii at ManoaJanuary 31, 2008
Agilent Pulse Measurement• Two separate BLAB1
ASIC with a common sampling strobe
• RF split the Agilent pulse with additional cable delay in the 2nd channel
• Tune the AC coupling capacitor to create a fairly Gaussian looking signal
CH1
CH2
Agilent Pulse Fitting• Apply Gaussian fit to both
functions• Measure the time
difference between the two fit means => actually a factor sqrt(2) better
• High timing resolution for a fixed amplitude over a small timing window
Gaussian Fit
6.4 psRMS
Pulse Fitting Disadvantages• Spend a lot of time
developing an amplitude varying fit function
• Requires software• Sluggish online
processing duty cycle• Offline processing
requires a very large data storage
• Offline processing requires a lot of CPU time
Cross-Correlation Theorem
CH1
Peak = 15
CH2
Peak = 45
F[CH1*(v)CH2(v)]
Peak = 30
Agilent Pulse Cross-Correlation Method
• Also, high resolution• Same timing jitter as
the fitting method• Hitting a timing
resolution limit from time base drifting
• Waveform splining was used to remove quantization effects for this plot
6.4 psRMS
Why use the Cross-Correlation Method?• Universal method to find the
time between the peaks of any two functions
• Cross-Correlation can be done with firmware=> “an online solution”
• Currently working on a developing firmware to determine the required FPGA resources and readout speed for this method
Cross-Correlation with Firmware
• FFT, INV_FFT, FIR Filter, and Complex Multiplier are free IP cores from Xilinx
Waveform DATA
FIR Filter (optional)
FFT Memory Stored Fourier test
function
Complex Multiplier
INV_FFTCorrelation Waveform
DATA
Peak Measurement
Charge Measurement
Time Measurement
Online “Spline”
• Artificially extend INV_FFT input array
• Set Re & Im part of array > Nyquist frequency to zero
• Possible to implement in firmware
1024 array @ 100 ps steps
8192 array @ 12.5 ps steps
System Block Diagram
• Up to 7x64 channels per cPCI card• Up to 4 cPCI cards per CAMAC card
MCP
BLAB2
BLAB2
BLAB2
BLAB2
MCP
MAIN cPCI
CARD
x7 cPCI
Crate
(Linux)x1
CAMAC
CARDCAMAC
Backplane
fDIRC Fiber Optic cPCI
• 8x 1.25Gbit Fiber Optic channels
• 8x 16Mbit SRAM• Initiator Transfer
through PCI bus• Store all waveforms
with a cPCI crate using Sci-Linux
• Linux PCI driver is currently under development
fDIRC Fiber Optic CAMAC
• 4x 1.25Gbit Fiber Optic channels• USB 2.0 optional computer interface• Enough IOs for all CAMAC pins
Action Items
• BLAB2 ASIC development• Online cross-correlation firmware development• Fiber optic firmware development• Linux cPCI driver development• Setup CAMAC data transfer protocol