3/19/2016 DIGITAL BACKENDS SIGNAL FLOW, PROBLEMS & SOLUTIONS. BY Shri Irappa M. Halagali Date : 21 / 10 / 2010.
Group members Prof. Yashwant Gupta Shri B. Ajit kumar Shri I. M. Halagali Ms. Mekhala V. Muley Shri I. S. Bhonde Mr. S. C. Choudhari Mr. G. J. Shelton Mr. K. D. Buch Mr. H. S. Reddy Irappa M. Halagali
CONTENTS OF THIS TALK 1. INTRODUCTION TO DIGITAL BACKENDs and WEB PAGE. 2. BLOCK DIAGRAM of the DIGITAL BACKENDs : HARDWARE CORRELATOR & PULSAR Receiver 3. SIGNAL FLOW, PROBLEMS & SOLUTIONS IN EACH SECTION OF THE SYSTEMs. 4. MAIN PROBLEMS & SOLUTIONS. 5. Analysis of CALLSHEETS. 6. QUESTION - ANSWERS.
DIGITAL BACKENDS WEB PAGE.
DIGITAL BACKENDs BLOCK DIAGRAM HARDWARE CORRELATOR & PULSAR RECEIVER.
Analog Inputs to DIGITAL BACKENDS
Analog inputs to DIGITAL BACKENDS Total Inputs to Digital backends are ; 30 Dish x 2 polarisations x 2 sidebands = 120 nos. Central Square Dishes : 14 West arm Dishes : 06 East arm Dishes : 05 South arm Dishes : & 175 Polarization
SAMPLER / ADC RACKs Digital backend has 2 Sampler / ADC RACKS Each RACK has 3 BINS. Each Bin has 10 CARDS Each card has 2 analog inputs. Each rack has 2 power supplies. One for Analog and other for Digital section of the cards to isolate the ground.
SAMPLER / ADC RACKs.
SAMPLER / ADC CARD Analog to Digital conversion PROBLEMS : INPUT CONNECTOR BECOMMING LOOSE. FIXED PERMANENTLY. IC ad589 BECOMMING BAD OFTERNLY. STOPPED POWER RESET TO RACKs WHILE CHANGING THE CARDS TO FIX PROBLEMS.
Delay – DPC RACKs. Digital backend has 2 Delay – DPC RACKs. Each RACK has 3 bins. Each BIN has 5 cards. Each card takes 4 Inputs. So each card handles both polarizations of USB & LSB.
Delay – DPC RACKs.
Delay – DPC CARD Delay correction & Data preparation. PROBLEMS : Fringe problem in BOTH SIDEBANDs and BOTH POLARIZATIONS. Replacing of converter ICs or POWER RESET to racks.
FX RACKs. Digital backend has 3 FX racks per SIDEBAND. Each rack has FFT, TRANSLATE & MAC BINs. FFT BIN has 10 FFT & 2 FFT control cards. TRANSLATE BIN has 5cards. MAC BIN has 11 MAC & 1 MAC control card. Each rack has one power supply of +/-5v,200/50A
FX RACKs. FX0 - DISH C0 - C4 & C5 - C10. FX1- DISH C11- C14-W1 & W2-W6. FX2-DISH E arm & S arm. Accumulation & Integration by MAC cards as per matrix chart.
FFT CARD 512 point Fast Fourier Tranform. PROBLEM : No fringe in one pol, one sideband. ASIC in the corr'g pipeline of the DB may be bad. No fringe in both pol. Dry solder or loose contact in the SMD connectors of the MB and DB. ASICs on DB may be bad.
Translate card Tranlate card makes four copies of the signals. PROBLEM : Self problem in one or two antenne. Chip resistors dry solder or loose contact of card in the bin.
MAC card Multiplication & Accumulation. PROBLEMS : Bad baselines : Corresponding ASIC, LATCH or CONVERTER IC may be bad or loose contact or problem in the cables. Or fuse may be blown off.
DAS & Controlling. DATA ACQUISITION from the FX racks of USB & LSB is done by the DAS card inside the CORRACQA & CORRACQB PC's. Controlling of the system like CONFIGURING, DYNAMIC UPDATION OF DELAY and FSTC is done from the CORRCTL PC.
Cards, ASICs and CABLEs Total cards used in the corr. system : = 277 TOTAL ASICs USED in the HARDWARECORR. SYSTEM is : FFT BINs & MAC BINs Sampler I/P – 120 nos. Delay DPC I/P – 120 nos. FX INPUT nos. FX racks -- 1 length : 84 nos. 6 length – 36 nos. 7 length : 54
CLOCK, STA INIT & PPM signals. 32 & 32.5 MHz clock signals generated by DDS unit. STA INIT signal of ms is generated by STAINIT card using clock. ONE MINUTE PULSE is generated by RUBIDIUM UNIT.
AC POWER Distribution & Earthing. Two UPS of 20 KVA used. UPS1 for USB. Load is of 35 Amps. UPS2 for LSB & common load. Total load is of 56 Amps. UPS1, UPS2, DC & CAGE earths.
Bad Baselines Total Baselines = n{(n(n-1)/2} Where n is the number of dish. Total number of baseline for 30 dish is 465. Bad baselines are mainly due to loose contact or bad cables or ASIC/ICs/FUSE blown off on MAC CARDS.
Delay – DPC BIN not configuring Problem in the corresponding BUS controller card in the Delay – DPC racks. Either loose contact of cables between BUS control card & BINS or bad converter ICs.
Instable or no fringe in both Sidebands This is mainly due to either configuration or dynamic delay updation problem in the FPGA of the Delay – DPC cards. Power reset to rack.
Pulsar Receiver Backends GAC (GMRT ARRAY COMBINER) combines the output from the FFT in the CORRELATOR. The phased Array output is formed by coherently combing the voltage signals from 30 inputs. Useful for observation of small sky area with high sensitivity. The Power output of the GAC is processed by IA section, which is useful for large scale pulsar search observations.
Corrupted Data output after few Blocks On general purpose interface PCB, CLOCK signal track was wired between FRC connector to the IC, but the original track wasn't cut. Even though there was no signal on the original track, it was picking up noise and data was getting corrupted after few blocks. By cutting the track, this problem fixed permanently.
Major breakdowns of system. In 1999 : Power supply failure : It was working for few hours and then it was becomming off. So after 2-3 attempts, it made series of damages to the system. So system was down for 3 days. In 2006 : Flat Bandshape : The problem found on the FFT card. Where few control word bits were misbehaving. Problem fixed permanently by putting filter capacitor. In 2008 : Phase Jumps : It found the phase of the DDS unit. Fixed it permanently.
Extreme cares to be taken 1. While changing the cables pins of the patch card may gets broke into the SIP sockets of the backplane. Whole backplane needs to be changed. 2. If there are any indication of power supply failure, change it without trying anything inside the rack. Otherwise it may cause series of cards/components failure. 3. Remove rings, watch while working on the rear side of the FX racks.
Analysis of CALLSHEETS PROBLEMs YEARs LOOSE CONTACT CABLES/CARDS. 2. SYSTEM CONFIG./ REBOOT TO RACKS/PCs. 3. CARD/COMPONENT/ CABLEs BAD/FAILED. TOTAL
* THANKS *