Difference between revisions of "TIGER documentation"

From BESIII Ferrara Group Wiki
Jump to: navigation, search
m (DAQ PC)
m
Line 104: Line 104:
 
|}
 
|}
  
== Rate considerations ==
 
 
=== TIGER ===
 
 
CGEM-IT setup:
 
* clock frequency = 160 MHz
 
* 2 Tx links per chip running at 160 MHZ SDR (Single Data Rate)
 
 
Chip maximum bandwidth = 320 Mbit/s
 
 
1 event word = 64 bits --> 80 bits with 8b/10b encoding
 
 
Chip maximum rate = 320 Mbit/s / 80 bit/event = 4 M event/s = 4 MHz
 
 
Channel maximum rate = 4 MHz / 64 channels = '''62.5 kHz per channel'''
 
 
Frameword rate = 4.8 kHz per chip (equivalent to 76 Hz per channel increase)
 
 
=== GEMROC ===
 
 
==== TRIGGER-MATCH ====
 
 
Assumptions:
 
* 25 kHz rate per channel (noise + physics)
 
* 1.5 us time window
 
* 8 TIGER (512 channels)
 
* Two 64-bit header and trailer words for GEMROC packet
 
* Maximum L1 trigger frequency = 4 kHz
 
 
25 kHz per channel --> 1.6 MHz per chip
 
 
Number of hits per trigger = 1.6 MHz * 1.5 us = 2.4 hit/packet
 
 
Average packet size = 2.4 hit/packet * 64 bits (event word size) * 8 (TIGER) + 2 (header+trailer) * 64 bits = 1228.8 bits + 128 bits = 1357 bits = 170 bytes
 
 
Maximum trigger-matched data rate = 170 bytes * 4 kHz = 680 kB/s = 5440 kbit/s per GEMROC
 
 
==== TRIGGER-LESS ====
 
 
Assumptions:
 
* 25 kHz rate per channel (noise + physics)
 
 
25 kHz per channel --> 1.6 MHz per chip
 
 
Trigger-less data rate = 1.6 MHz * 8 (TIGER) * 64 bits (event word size) = 819.2 Mbit/s = 103 MB/s per GEMROC
 
 
=== DAQ PC ===
 
 
1 GbE link
 
 
Maximum bandwidth = 1 Gbit/s = 125 MB/s
 
 
* Can easily accomodate data from more than 100 TM GEMROC modules
 
* Can only accomodate data from 1 TL GEMROC module (only if UDP communication is optimized)
 
  
 
== TIGER padframe ==
 
== TIGER padframe ==
  
 
[[File:TIGER padframe.png|thumb|1000px|TIGER padframe]]
 
[[File:TIGER padframe.png|thumb|1000px|TIGER padframe]]

Revision as of 04:19, 6 June 2019

QDC mode[edit | edit source]

Efine with wrap-around effect explained

The counter for the ADC performing the charge digitization has a wrap-around effect such that very high signals (just below the saturation point) have 1008 < Efine < 1024.

For these values the measurement is still linear, so in order to apply the usual conversion using the calibration line (Efine = const + slope * Qin) we need first to shift them by 1024. Therefore we can define a new variable Efine_new given by:

  • Efine_new = Efine (for Efine < 1008)
  • Efine_new = Efine - 1024 (for Efine > 1007)

In this way, the values of Efine between (1008, 1023) are shifted to the region (-16, -1) and thus correspond to the very high charge signals, with Efine_new = -16 being the new value of saturation. This is explained in the Figure on the right and the Table below.

Qin Efine Efine_new Description
5 385 385 normal region
10 340 340 normal region
15 295 295 normal region
20 250 250 normal region
20 205 205 normal region
30 160 160 normal region
35 115 115 normal region
40 70 70 normal region
45 25 25 normal region
46 16 16 normal region
47 7 7 normal region
48 1022 -2 wrap-around region, still linear -> apply Efine_new
49 1013 -11 wrap-around region, still linear -> apply Efine_new
50 1008 -16 saturation
55 1008 -16 saturation
60 1008 -16 saturation


TIGER padframe[edit | edit source]

TIGER padframe