SCT_Links_monitoring_April08

Download Report

Transcript SCT_Links_monitoring_April08

SCT Links Long Term Monitoring
• Summary quality installed links
• Techniques for monitoring long term
performance
– Data links
– TTC links
• Improvements for SLHC
Tony Weidberg
Opto WG April '08
1
Data Links Quality
• Data links status before installation in
pit.
– Dead : 0.6%
– Problematic: 0.3%
• Deaths: ESD VCSELs, broken single
fibres on detector, damaged tracks on
Al/kapton tapes.
• Problematic: “slow turn”-on VCSELs
(see next slide).
Tony Weidberg
Opto WG April '08
2
SCT Slow-turn on VCSELs
• Send fixed
data pattern
• Scan DAC RX
threshold
• Check to see
if correct bit
pattern
returned.
Tony Weidberg
Opto WG April '08
3
Data Links Monitoring - 1
• RX Max and RX Min
distributions from the RX
threshold scans.
– Insufficient dynamic range 
RXmax usually >255 so can’t be
determined.
– Best way to monitor optical
output of on-detector VCSELs is
using RXmin
– Long term monitoring of number
of links with low values RXmax
will indicate if some VCSELs are
going to die .
Tony Weidberg
Opto WG April '08
RX Max
RX Min
4
Data Links Monitoring - 2
Scan RX thresholds for data links,
RXmin  optical power received.
Plot ratio RXmin ratio (Pit/RAL
reception)
See decrease of 1 dB as expected from
attenuation in long cable
Histogram RXmin ratio Pit/RAL
Tony Weidberg
Opto WG April '08
5
Data Links Monitoring - 3
• No dramatic changes
but need to start long
term monitoring.
• Ideally we should
have several data
sets before beam to
understand stability
and performance
before radiation.
Tony Weidberg
Opto WG April '08
<RX Min>
2007
data
2008
data
<RX Max>
6
TTC Links Quality
• TTC links before installation in pit.
– Dead : 0.2%
– Problematic: 0.7%
• Deaths: broken single fibres on detector,
damaged tracks on Al/kapton tapes.
• Problematic: low p-i-n current channels
– We initially thought this was a very
widespread problem but it turned out to be
mainly an artifact of oscillations in the
current monitoring circuitry in the power
supply.  replace up to 10% VCSEL arrays
in ROD.
Tony Weidberg
Opto WG April '08
7
TTC Links Monitoring
• Measure the currents for the ondetector pin diodes (Ipin) using
the LV0 power supplies in the
counting room.
• Recent measurements of Ipin
for SCT barrel in the pit.
• Mean <Ipin>=0.24 mA ok
• 7.7% links with Ipin<0.1 mA (we
require Ipin>0.1 mA to minimise
SEU).
• Measure Ipin for all suspect low
channels with simple card with
battery + 1 kW series resistor +
DVM.
Tony Weidberg
Opto WG April '08
0.1 mA
Ipin
8
Correlation (DVM + 1kW) vs LV0
Ashley's Ipin correlation
0.6
0.5
Ipin(Ashley)
0.4
0.3
0.2
0.1
0
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Ipin (LV0)
Correlation ok at high Ipin but poor at low values of Ipin
9
Status Low Ipin Channels
• Measured 1941 channels
• From measurements of 150 suspect
channels only 14 confirmed low (of
which 7 read 0)
– To be checked with optical power meter.
• 0.8% low channel replace ~ 9% of
TXs (12 way arrays).
Tony Weidberg
Opto WG April '08
10
Monitoring BER
• Can monitor BER for data links using
– Preamble (fixed pattern)
– Header (some bits fixed)
– Trailer (fixed pattern)
• TTC links
– BCID and L1ID counters in the ROD.
– ROD receives full BCID and L1ID from L1 Trigger.
– FE chips (ABCD) on module count BCs and L1s and the 4/8
LS Bits of these numbers are transmitted to the RODs.
– Compare LS bits of returned BCID & L1ID with ROD data.
– Any discrepancy  bit error(s). Note that we expect finite
BER from SEU at high luminosity.
• 95% modules show no errors, 5% have errors,
mainly BCID and L1Id; under investigation …
Tony Weidberg
Opto WG April '08
11
Conclusions
• We have tools available and we will be able to track
any long term degradation in performance of data or
TTC links.
– Data links: perform RX threshold scans.
– TTC links: measure pin currents.
– BER for data/TTC links from BC and cross checks.
• Monitoring could be better if more thought at the
design stage:
– More accurate measurements of pin currents for low values
essential for determining whether TTC link performance is
acceptable.
– better dynamic range for RX DAC (data links) so that digital
scans would always determine meaningful values of
RX_Max.
– by measuring individual p-i-n currents in RX.
Tony Weidberg
Opto WG April '08
12