next up previous contents
Next: Time-stamps in the November Up: GRASP: a data analysis Previous: Example: diag program   Contents

GRASP Routines: Reading/using FRAME format data

0 The LIGO and VIRGO projects have recently adopted a data format standard called the FRAME format for time-domain data. The 40-meter laboratory at Caltech implemented this data format in Spring 1997; data taken after that time is in the FRAME format. The FRAME libraries are publicly available from the VIRGO project; they may be downloaded from the site The GRASP package has been tested up to release 3.72 of the frame library. Contact Benoit Mours
[4] for further information.

The GRASP package includes routines for reading and using data in the FRAME format. Also included in the GRASP package is a translator (see Section [*]) which translates data from the old data format used in 1994 to the new FRAME format. Data distributed for use with GRASP will primarily be distributed in this new FRAME format, and over a period of time we will remove from the GRASP package all of the code and routines which make use of the old format. In order to help make the transition from old format to FRAME format as smooth as possible, the GRASP package currently contains both old format and FRAME format versions of all of the example programs. For example animate and animateF are two versions of the same program. The first reads data in the old format, the second reads data in the FRAME format. If you are new to GRASP, we don't recomend that you waste your time with the old data format; start using the FRAME format immediately.

Data distributed in the FRAME format may not be compatible with future releases of the FRAME library, so if the FRAME libraries are updated you may need to obtain a new copy of the standard 40-meter test data set from November 1994. The data that has been distributed and is currently being distributed makes use of either version 2.20, 2.30 or 2.37 of the FRAME library. We will shortly begin distributing data in version 3.50 of the FRAME format. Only two files in the GRASP package (src/utility/frameinterface.c and src/examples/examples_utility/translate.c) depend upon the version of the FRAME library. We distribute GRASP with versions of these files appropriate for different releases. The files determine the version of the frame library at compilation time, and then include the appropriate code. This code works correctly with any version of the frame library $2.37 \le {\rm version} \le 3.70$. Note that version $\ge 3.50$ of the frame library can read data written by any version back to and including 2.37.

One of the nice properties of the FRAME formats $\ge 3.30$ is that they support a ``compressed" format. This is transparent to the user (except that reading the ``compressed" frames takes a bit longer because the frame library then needs to uncompress the data). Data distributed in version 3.50 of the FRAME format is being distributed in this compressed form and occupies somewhat less space than the old-format original data. As shown in Section [*] the old-format data for the November 1994 runs occupied about 13.6 Gbytes. For comparison, the FRAME-format data occupies less than half of that space:

     14nov94.1.frame 314
     14nov94.2.frame 397
     18nov94.1.frame 503
     18nov94.2.frame 543
     19nov94.1.frame 551   The space occupied
     19nov94.2.frame 535   is shown in Mbytes
     19nov94.3.frame 641
     19nov94.4.frame 605
     20nov94.1.frame 553
     20nov94.2.frame 422
     20nov94.3.frame 755
The total storage space required for FRAME 3.50 data totals only 5.8 Gbytes.

In order to give the 1994 40-meter data a form as similar as possible to the data being taken in 1997 and beyond, the channel names used have been given equivalent ``FRAME" forms. These are shown in Table [*].

Note that new data created in the frame format attempts to address at least a couple of the problems in the ``old format" data. In particular, new frame format data (i.e., post 1996) has sample rate in Hz always being powers of 2, for example, 4,096 Hz or 16 Hz or 16,384 Hz. In addition, each frame always contains a power-of-two number of seconds of data. These conventions will make it easy to ``match up" sample of channels taken at different rates, and to do FFT's of the channels. However the 1994 data does not conform to either of these conventions: each frame of 1994 data contains 5000 samples of the slow channels, and 50,000 samples of the fast channels, during a $5.06666\cdots$ second interval.

Table: Channel assignments for the November 1994 data runs. Channels 0-3 are the ``fast" channels, sampled at about 10 kHz; the remaining twelve are the ``slow" channels, sampled at about 1KHz. The equivalent ``FRAME" format names are also given.
Channel # $\le$ 14 Nov 94 FRAME name $\ge$ 18 Nov 94 FRAME name
0 IFO output IFO_DMRO IFO output IFO_DMRO
1 unused   magnetometer IFO_Mag_x
2 unused   microphone IFO_Mike
3 microphone IFO_Mike unused  
4 dc strain IFO_DCDM dc strain IFO_DCDM
5 mode cleaner pzt PSL_MC_V mode cleaner pzt PSL_MC_V
6 seismometer IFO_Seis_1 seismometer IFO_Seis_1
7 unused   slow pzt IFO_SPZT
8 unused   power stabilizer PSL_PSS
9 unused   unused  
10 TTL locked IFO_Lock TTL locked IFO_Lock
11 arm 1 visibility IFO_EAT arm 1 visibility IFO_EAT
12 arm 2 visibility IFO_SAT arm 2 visibility IFO_SAT
13 mode cleaner visibility IFO_MCR mode cleaner visibility IFO_MCR
14 slow pzt IFO_SPZT unused  
15 arm 1 coil driver SUS_EE_Coil_V arm 1 coil driver SUS_EE_Coil_V
Note: On 18 November 1994 run 1 the power stabilizer was accidentally disconnected until approximately 20:00 local time.

next up previous contents
Next: Time-stamps in the November Up: GRASP: a data analysis Previous: Example: diag program   Contents
Bruce Allen 2000-11-19