Martin Hewitson

1
Data acquisition at GEO600

• Martin Hewitson

2
Schematic overview (Ruthe)
coch
32f/64s chans 1 MB/s
DCU (alchemist)
central
Data collector
To Hannover (lt 1 MB/s)
vDCU (control)
Raw data server
1000us 5 kB/s
f fast (16 kHz) S slow (512 Hz) Us
ultra-slow (1 Hz)
16f 500 kB/s
16f 500 kB/s
3
DCU
Fast channels
Slow channels
4
On Coch

• Data collecting software
• Writes data to disk in proprietary format
• Uses GEO directory structure
• Raw data server
• Accessible from Ruthe
• 1 sec latency

5
Schematic overview (Hannover)
bute
client
frame maker
h(t) maker
fserv 9000
From Ruthe (lt 1 MB/s)
RAID array (3 TB)
hour trend maker
day trend maker
client
fserv 9008
Charlie (Archival)
6
A summary of file types

• Raw data files (1 second duration)
• fs lt 16384 Hz
• One file per DCU per second
• Frame files (1 minute duration)
• fs lt 16384 Hz
• 60 frames per file, one second per frame
• All channels in one file
• calibration information
• Hour trends files (1 hour duration)
• fs 1 Hz
• 60 frames per file, 60 seconds/samples per frame
• Min, max, av
• Day trend files (1 day duration)
• fs 1/60 Hz
• 24 frames per file, 60 minutes/samples per frame
• Min, max, av

7
Current DAQ system - summary

• Commercial solution (mostly)
• Has evolved into quite a stable solution
• 99 reliability
• improvements mostly due to software
• Some problems remain
• Easily damaged input stage
• Sometimes high level of cross-talk is observed
• Some parts are obsolete
• Main drawbacks
• Costly repairs and replacement parts
• No user-serviceable parts

8
Moving forward

• Continue with the current system?
• Need to repair ADC boards once per year
• Spare processor boards no longer available
• Problems still exist with signal sizes and
  channel overload protection
• This needs more investigation
• Cross-talk is difficult to predict and control
• Move to a new DAQ system

9
A new DAQ system

• Aims
• Replace existing system with something we can
  control
• More exactly meet the needs of GEO
• Make something cheap (comparatively), expandable
  and versatile
• A flexible system that can also be used in lab
  environments
• Extends the current software model of parallel
  data streams

10
General approach

• Make a modular system
• Each channel is an independent module
• Each module collects data and sends it over
  ethernet
• Modules can be placed anywhere (near signal
  source)
• Cost per module 300 infrastructure

11
System overview
12
A channel module

• Single channel board designed to fit into its own
  enclosure
• Each module connects directly to a data
  collecting computer via TCP/IP

13
Advantages over current system

• Each channel is independent of all the others
• Cross-talk is minimised
• Effect of failures is minimised
• Only the required number of channels is used
• Individual channels can have independent
  acquisition rates
• Individual channels are easy to repair/replace
• Channels can be placed close to signal source
• Reduces pick-up of noise

14
Open questions

• What is the required operational time of the
  current DAQs?
• Should we be looking beyond GEO?
• GEOHF needs
• Are we happy to continue with the (costly)
  repairs?
• Should we move to a new DAQs?
• Manpower?