View
34
Download
4
Category
Preview:
DESCRIPTION
Status of Derandomizer electronics. E. Atkin / A. Kluev (MEPhI). Outline. Status of the de-randomizer electronics development Progress on chip design and plans. De-randomizer (remake of the old slide of 2005). - PowerPoint PPT Presentation
Citation preview
Status of Derandomizer Status of Derandomizer electronicselectronics
E. AtkinE. Atkin / / A. Kluev (MEPhI) A. Kluev (MEPhI)
Outline
• Status of Status of the de-randomizer the de-randomizer electronics developmentelectronics development
• Progress on chip design and Progress on chip design and plansplans
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
AnalogAnalog derandomizerderandomizer is a unit performing neuron-like is a unit performing neuron-like processing, but in an analog field.processing, but in an analog field.
It is a deadtime free analog unit withIt is a deadtime free analog unit with nn-inputs-inputs andand
mm-outputs,-outputs, n>m.n>m.Thus it allows to reduce the number of following ADCs.Thus it allows to reduce the number of following ADCs.
The The derandomizationderandomization procedure implies the skipping of empty channels and thus procedure implies the skipping of empty channels and thus is indivisibly bound with data is indivisibly bound with data sparcificationsparcification..
Efficient processing of the randomly appearing signals by blocks, having a dead Efficient processing of the randomly appearing signals by blocks, having a dead time, needs the choice of a proper architecture.time, needs the choice of a proper architecture.
De-randomizer
(remake of the old slide of 2005)
Status
• 442 derandomiser prototype in 2 derandomiser prototype in UMC 0.18 um showed UMC 0.18 um showed functionalityfunctionality
• 12812816 one is under 16 one is under developmentdevelopment
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Crosspoint switch
Small delay between two signals
(partially overlapped)
Two signals at the same time
(full overlapping)
1-comparators2-inputs3-outputs
1
2
3
1
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Crosspoint switch
4 signals (comparator outputs) 2 crosspoint switch outputs (after amplification)
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Development of the de-randomizer 128Development of the de-randomizer 1281616
A. Klyuev(PhD student), M. Trubetskoy(diploma student)A. Klyuev(PhD student), M. Trubetskoy(diploma student)
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Structure of analog de-randomizer
(remake of the old slide of 2006)
0x0y 1x0y
0x1y 1x1y
0xMy 1xMy
Nx0y
NxMy
Nx1y
Analog input
n channels
Analog output
m channels
Address bus
Control
Arbitration logic
Crosspoint
switch (NxM)
Analog switch
(address NM)
n
CMP CMP
0
Analog
derandomizer
“busy” signals from
Output channels
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
t t
The choice of de-randomization factor (d) strongly depends on working conditions of the system. Let's consider 2 types of working conditions:A. The most probable event – pulses occur NOT simultaneously, uniformly in all channels and have equal duration. The 1st prototype is based on this variant. The FSM for this case exists and can be scaled to other d (128->16, 64->40, 59->19, etc. It is better when the number of the channels is divisible by 2 however).B. The most probable event – many pulses occur simultaneously (T>>∆t). It is really typical for the detectors, taking into consideration the cluster nature of their pulses. For this case we have to modernize the existing FSM, so it will be able to commutate one channel per clock.
1->12->23->4(3rd output channel occupied at this time)4->55->6Etc.
A B
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
Two types of considered working conditions
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Structure of new arbitration logic
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Simulation of arbitration logic (RTL model)
clkrst
in_0
in_17adr_0
adr_15
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Simulation of arbitration logic (synthesized netlist for UMC 0.18)
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Structure of the last variant of arbitration logic
CBM Collaboration meeting (September 25 - 28, 2007) @ FZD, Dresden
Simulation of the last variant arbitration logic (synthesized)
Progress on chip design
• Familirization with the Familirization with the N-XYTERN-XYTER (Simulations (Simulations of critical blocks) (is under study by 2 diploma of critical blocks) (is under study by 2 diploma students). The test results of N-XYTER would students). The test results of N-XYTER would be useful.be useful.
• Development of a new release of Development of a new release of derandomiserderandomiser electronics (128 electronics (128 16 16 structure) (talk of A.Kluev)structure) (talk of A.Kluev)
• Study aspects of improvements of Study aspects of improvements of radiation radiation hardnesshardness (talk of A.Simakov/V.Shunkov) (talk of A.Simakov/V.Shunkov)
• Development of low power rad-tolerant Development of low power rad-tolerant ADCADC (talk of A.Gumeniuk)(talk of A.Gumeniuk)
Main tasks since previous meeting:
Plans
• Develop the following building blocks and Develop the following building blocks and participate with them in the UMC 0.18 um MPW participate with them in the UMC 0.18 um MPW run for: run for:
the CBM-XYTER:the CBM-XYTER:1)ADC1)ADC2) front-end block 2) front-end block
the back-up solution:the back-up solution:3)1283)12816 derandomiser 16 derandomiser 4)some rad-hard test structures 4)some rad-hard test structures
• Improve facilities for:Improve facilities for:chip design and measuringchip design and measuringrad-hard testsrad-hard tests
We would like to:
Recommended