Waterson - MWA-LFD Project

download report

Transcript Waterson - MWA-LFD Project

Receiver Systems
20 January 2009
Mark Waterson, ANU
Digital Receiver (RRI)
ADFB - Analog to Digital & Filter Bank (2)
AGFO – Aggregation & Fiber Optic (1)
Node M&C
Arcom Single Board Computer (SBC)
Diamond Systems GPIO card
ATIM – Antenna Tile Interface Module
ATIF – Antenna Tile Interface Filter (8)
ASC – Analog Signal Conditioning (2)
Chassis, Fans, & Power Supplies
Thermal sensor interface cards (2)
Current Hardware Status*
N1 - operational at MRO
N4 – operational at MRO
N2 – repaired, enroute to Haystack
N3 – operational at RRI
* issues with BF coms affect all
* ATIF Walsh channels not fully tested
Current Issues:
Packaging & cooling
Clock system
Control Software (or lack thereof)
Beamformer interface
Rev2 pcb modifications
Digital Receiver development
Packaging & Cooling – PSI (Perth)
Preliminary design in progress
Clock system design/prototype
(CfA, ??) Preparing quotes for
Rev2 boards
ASC gain/filter, connector (mfw)
ATIM reconfigure (PSI/mfw/?)
Backplane? (PSI/RRI)
“Core” control software
Basic “bit-flipping” routines exist
“Robust” framework defined but needs work
Some DigiRec status features still need work
Interface between Central M&C and node
ICD defined, but not implemented completely
Ed Morgan presently working (on Node2 )
Beamformer Control Interface
Power control & distribution
(Lightning protection)
“ATIF/BTIF” boards – existing, need fixes
“Data-over-Coax” – proto being built
ATIF/BTIF Interface
Artwork faults appear to be solved in
current-generation boards – OK
Filter components were incorrect on some
boards, we think all fixed
48v DC current draw
issue with ATIM component identified, will
need mod to board
Fix/mod for Beamformers designed
BF Long-cable communication
Haystack never able to reliably drive long cables, even at 100Hz
ANU never able to reliably read back temperature or checksum
ANU has checked signal fidelity, appears clean (with correct parts) at
100KHz & 200m
Work at site this week:
MIT-Haystack tester blew up
Some BFs tested with internal read-back board
Short & long-cable waveforms recorded
Readback data saved from SBC coms
>> no consistent, proper operation observed!
Repeatedly observed BF not resetting on message start, conflicts with ICD
Errors with both long and short cables
Some ATIF dependency
DID read back correct temperature & saw internal “ok” with SBC software
Data and clock
Long cables,
measured at BF end
ANU Bench tests:
Clock & data waveforms
checksum & temperature
Lab testing (ANU)
=> BF cpld seems to be missing data bits
(or clocks)
Causes checksum & temperature errors seen
Will cause tile mis-pointing
Haystack to confirm, look at BF cpld
Walsh transition drive
Waveforms confirm expectation
Impact of driver faults
Walsh-sw transitions
Walsh switching
ATIF common-mode offsets
MAX232 IC failures, usually seems to be drive
side, (but hard to be sure)
Mostly Receiver end, again not enough data
Common-mode voltage difference between BF and
Receiver exceeds chip supply, turns on/burns out
protection diodes
Added ground conductors + internal tie reduced
problem (2nd Nov trip), may be solution
ATIF re-design?
Real problem – drive & receive circuits do not share
same “ground” reference, so not possible to
define/clamp voltages
Solution: isolation – optical or transformer
Opto-isolated 232 drivers
more-or-less same footprint as existing driver
Well-understood drop-in circuit fix
ATIF/BTIF board revision required
Transformer –
already available in “manchester” version of existing design
Requires mods to add encode/decode to data path (design done)
Capacitive – e.g. Eric/Roger’s serial data system?
BF Interface – now what?
Soon (32T):
Complete & test “D-O-C” interface
Revision of protocol to echo back BF data suggested:
Allows SBC to confirm pointing data bits
Simple changes to both cpld code
Not compatible with D-O-C interface…
Settle ATIF (common-mode) and ATIM revisions
Build & verify interface testers using field hardware
(SBC, ATIM, ATIF) so everybody uses same platform
to test
Haystack visit to resolve?
Decision dates?
ASC High-pass filter
Existing commercial highpass filter causes
Custom design needed to optimize MWA
~ not trivial –
Fc = 300MHz, <<3dB desired
Fstop = 330MHz, >30dB++ attenuation req’d
-> Lark, Minicircuits, …? Possible vendors
footprint change impacts board layout
LFCN-225 filter plot
LFCN-225 filter data
Receiver Development – Now
Complete software – science can’t be
done now
Revise hardware –
BFcoms DNW
Production revisions –
ASC, ATIM, BF interface changes
Field package design
fiber data/clocks, thermal system, etc needed
to get to CDR
Decisions, decisions…
Cooling system capacity
trade cheaper cooling solution (=>loss of
high-temperature ops time) for more tiles?
Assembly & test effort
Trade servicability (design effort/$) for test
ease/ modularity
Interfaces, operating temperature, etc vs cost
& complexity
Existing ATIF problems vs.
Return path to case (faraday cage) for
transient protection should have been
through standoffs at ATIF board corners BUT – boards mounted on connector only!
Cable shield not connected to connector
body -> no faraday shielding…