STATUS OF THE DIGITAL BACKENDS :

  1. HARDWARE CORRELATOR
  2. PULSAR RECEIVER
  3. NEW HARDWARE CORRELATOR
  4. SOFTWARE CORRELATOR

HARDWARE CORRELATOR :

RECENT IMPROVEMENTS :

  • Till mid-2002, the working G.M.R.T. digital backends consisted of the single sideband ( USB only ) G.M.R.T. correlator, and the USB G.A.C. along with IA and PA DSP pulsar receivers for the USB
  • In April-May 2002, the G.M.R.T. correlator was upgraded to a dual side-band, 32MHz, synchronized operation correlator with the Indian Polar mode as the default mode of operation
  • Along with the correlator upgrade, the newly designed Delay DPC system was installed, commissioned and debugged
  • The upgrade was also accompanied by a reorganisation and reconnections of the sampler units to the Delay DPC units, as well as a reorganisation and reconnection of Delay DPC outputs to the FFT units. The routing of the A.C. power output from the two UPS units supplying the digital backends was also redone in a manner such that one sideband operation of all backends can be ensured during the failure of one of the UPS units. The grounding and earthing scheme for the digital backend was also revamped

RELIABLE CORRELATOR PERPORMANCE "TOWARDS ZERO BADBASELINES" :

  • After the integration of the dual sideband correlator, the major challenge was the stabilisation of its performance as an error free system. The major problems were due to corruptions in data from individual baselines, caused usually by poor connectivity of cables in the signal distribution path in the MAC, and sometimes by failures of ASICs and other components
  • The performance improvement was affected by implementing a rigorous set of self tests for the correlator, using either internally generated test vector signals or by using noise sources connected to the sampler inputs, which would help in quick localization of the cause of the problem
  • Current Status : The correlator performance is monitored and maintained at the desired ' zero bad baseline ' level through weekly self tests which allow identification and replacement of bad cables and clock lock assemblies, failed hardware components, loose connections and bad grounding connections

RELIABLE CORRELATOR PERPORMANCE "THE FLAT BANDSHAPE PROBLEM" :

  • This was a major, though intermittent, problem that was detected after the integration of the dual sideband correlator, where the performance of some of the FFT cards was found to be unreliable, leading to a flat bandpass being produced for the concerned antenna signals
  • Because of its complex and intermittent nature, it took a lot of test time to localize the cause of the problem. Finally, it was traced to a noise glitch being picked-up in the control word section of the FFT ASIC pipeline, causing the ASICs to lose their configuration
  • The problem was solved by electrical shaping of the control signal lines in the FFT card, to make them immune to random noise glitches

FURTHER IMPROVEMENTS :

  • The full polar mode of the G.M.R.T. correlator remains to be tested and released for observations
  • Implementation of Walsh demodulation in the Delay DPC section of correlator remains to be completed
  • Total power computation and read out from the Delay DPC to facilitate noise on-off calibration also remains to be completed
  • It should be possible to implement some relatively simple algorithm for real time RFI mitigation in the Delay DPC section of the GMRT correlator
  • The FSTC implementation, as currently implemented in the FFT section, suffers from unwanted quantisation effects in the phase gradient tables. The FSTC can be implemented more accurately within the new Delay DPC section by using time domain interpolation FIRs
  • To modify the DSP code of the RRI PMTR pulsar backend to take care of the total intensity mode of pulsar observations
  • To modify the code for the Altera FPGA in the RRI PMTR to better optimize the signal flow for the efficient organisation of the data going into the DSP nodes
  • Software Pulsar Receiver : To acquire the array mode data directly from the GAC onto a PC platform and to do all further real-time processing in software; also include an option to record the data at the raw rate for completely offline processing

PERFORMANCE :

System 1999-2000 Failure Rate 2002-2003 Failure Rate 2003-2004 Failure Rate Spares
Sampler 2 / month 1 / month 1 / 6 months 07
Delay DPC 1 / 6 month 1 / year 1 / 3 months 03
FFT 1 / 3 month 1 / 4 month 1 / 4 months 02
Translate 1 / 6 month 1 / year 1 / year 01
MAC 1 / month 1 / 3 month 1 / 2 months 01
FFT Control 1 / 3 month 1 / 6 month 1 / year 01
MAC Control 1 / 2 month 1 / 6 month 1 / 4 months 01
ASICs 1 / 2 month 1 / 2 month 1 / 2 months 42 + 134


PULSAR RECEIVER :

INSTALLATION & COMMISSIONING OF SECOND PULSAR RECEIVER SIDEBAND :

  • Towards the end of 2003, the second sideband GAC ( for LSB ) was installed and released for use
  • First versions of the RRI PMTR for both USB and LSB were also installed along with the LSB GAC. After debugging and test observations, these systems have become available for regular observations from mid-2004
  • The data acquisition for the RRI PMTR was transferred to the high speed DAS systems used for the correlator and IA / PA bin systems

PERFORMANCE :

System 1999-2000 Year 2002-2003 Year 2003-2004 Year 2004-2005 Failure Rate
GAC USB Installation and commissioning In use for observation Integration with new correlator and in use for observation 01 / Year
IA USB Developmental work, installation and commissioning In use for observations integration with new correlator and in use for observations 01 / Year
PA USB Developmental work Developmental work, installation and commissioning; release for experimental use Integration with new correlator; release for regular observing 01 / Year
RRI PMTR USB Developmental work Developmental work, installation and commissioning Testing and debugging 04 / Year
GAC LSB - - Developmental work, installation and commissioning 01 / Year
RRI PMTR LSB - - Developmental work, installation and commissioning 04 / Year

SPARES :

For Multibit IA and PA Pulsar Receiver :
  • LCC : 1 nos, IIC : 1 nos, PSC : 1 nos, DSP node cards : 3 nos
    For GAC :
  • 4 i/p cards : 1 nos, 8 i/p cards : 2 nos, GAC Buffer Card : 2 nos
  • GAC clk card : 1 nos, GAC dist. card : 3 nos, 144 Line I / O card : 1 nos
    For Polarisation :
  • DSP card : 1 nos, 144 Line I / O card : 1 nos, ISA Buffer Card : 1 nos

PROBLEMS ANALYSIS(SUMMARY OF CALLSHEETS FOR 2004) :

Callsheet Type Jan Feb Mar Apr May Jun July Aug Sep Oct Nov Dec Total
Weekly Maintenance 4 4 3 3 9 6 1 1 4 - - 2 37
Sampler - - - - - 1 - - - - - - 01
Delay DPC - 1 - 1 1 - - 2 - - - - 05
FFT - 1 - - - - 3 - - 2 1 4 11
Translate - - - - - - - - - - - - 00
MAC 4 - - - 3 3 1 1 - 1 - 5 18
ASICs - - - 1 - - - 1 - - - - 02
Informational 2 - 1 - 1 2 - 3 - - - - 09
Corr_config, New Dly Config - 1 1 - - 1 2 2 2 - 1 - 10
Electrical - 1 - - 2 - 1 - - - - - 04
Non Correlator - 4 1 - 2 4 2 1 - 1 - - 15
Cables 1 6 3 3 - 2 2 3 - - 1 5 26
GPS - - - 1 - - - 1 - - - - 02
Pulsar Receiver - - 2 - - - - - - - - - 02
Problems solved by itself ( ! ) 1 2 1 3 1 1 1 4 - - - - 14
Total 12 20 12 12 19 20 13 19 06 04 03 16 156


NEW HARDWARE CORRELATOR :

Status:

  • 2 ANTENNA single board pocket correlator with delay & fstop correction completed & tested for shorter baselines.
  • 8 ANTENNA dual pol. packetized corrletor with delay & fstop correction under test.


SOFTWARE CORRELATOR :

Status at the II phase :

  • 32 channel system (32 ant, 1 pol, 16 MHz) : Running at 1.5 x real-time rate for 16 MHz BW data. Thoughts at that time .....
  • Extension to 32 ant, 2 pols, 16 MHz (GSB-32.2.16) by adding 8 more nodes.
  • Extension to 32 ant, 2 pols, 32 MHz (GSB-32.2.32) by attaching more Compute only nodes to the cluster.
  • Add : 1. 8 more nodes similar to existing ones.
  • 2. 16 nodes of ~ 2x compute capability of present nodes (e.g. Same motherboard with quad core dual processor CPUs, instead of the dual core dual processor CPUs).

GOTO DIGITAL BACKEND