National Research
Council Canada
Conseil national
de recherches Canada
Observing Efficiency and Reliability of
Gemini South MCAO
Glen Herriot
Herzberg Institute of Astrophysics
National Research Council Canada
Observing Losses for MCAO
Airplanes
Weather
Failures
Satellites
Other
Observatories
Purpose
• This analysis estimates the overall
observing efficiency of MCAO, which is
NOT specified in the requirements
documents.
Top Level Requirements
• Lose no more than 10% of scheduled time to failures
• CoDR book assumed that half the nights in a year will be
scheduled for MCAO, (180 x 10 hr).
• But this is a classical scheduled spec. We have taken it to
mean 1800 hours per year, queue-scheduled.
• Maximum object setup time (excluding telescope slew and
acquire PWFS stars) < 120 seconds
• Telescope Slew and PWFS2 acquisition < 120 seconds.
Lasers Must Not Interfere With
Satellites
Fraction of Time Laser Must Be
Shut Down.
Lost time due to Satellite
Interference
9
8
7
6
5
4
3
2
1
0
0.1
0.2
0.3
0.4
0.5
0.6
>0.7
No. of stars vs. fraction of time lost
Worst Case Laser Duty Cycle
Laser Clearing House Permitted Laser Operation
SAO65711 from Lick site
Laser On = 1 or Off = 0
2
1
0
0
1
3
2
Time (hours)
4
5
Slightly Better Than Typical
Case
Laser Clearing House Permitted Laser Operation
RA14:28 Dec 43:24 Lick site
Laser On = 1 or Off = 0
2
1
0
0
1
2
Time (hours)
3
4
Good Case
Laser Clearing House Permitted Laser Operation for Star 231 from the MMT
Laser On = 1 or Off = 0
2
1
0
0
1
2
3
Time (hours)
4
5
6
Typical: 28 ‘On’ events, 41% Off time;
20 minute max. On ; 1 minute median Off
Typical Permitted Laser Operation (Star 46158 from Lick)
Laser On = 1 or Off = 0
2
1
0
0
1
2
3
Time (hours)
4
5
Histogram of Permissible Time
Durations “Laser On”
Distribution of Laser On-Time
1.
00
2.
00
3.
00
4.
00
5.
00
6.
00
7.
00
8.
00
9.
00
10
.0
0
11
.0
0
12
.0
0
13
.0
0
14
.0
0
15
.0
>1 0
5
M
IN
200
180
160
140
120
100
80
60
40
20
0
Time Minutes
(1
cum) slot 
Probability of “Laser On” vs.
Duration > T, ( Avoid Satellites)
m
0.0375
0.0173
2.886·10 -3
100%
Probability vs. Length of time on
1
P(t>T)
0.1
10%
exp(-T/ +bT2-ct3)
0.01
1%
exp(-T/)
1  10
0.1%
3
0.01%
4
1  10
0
0
20
20
40
40
60
80
80
60
Time T -minutes
100
100
120
Minutes
Queue Scheduling Increases
Odds of Finding a Time Slot.
Probability of
1 slot of
length t>T
Probability of
success with
more targets
in queue.
Solve for
Queue length
needed:
 T bT 2 cT 3 
Pr( t  T )  e 

Pr( suc )  1  (1  Pr( t  T ))
ln( 1  Pr( suc ))
N
ln( 1  Pr( t  T ))
N
Queue Length - vs.
Integration Time >T
1 Event
P(t>T)
Int Time
5
37.9%
10
16.5%
15
8.1%
20
4.4%
25
2.7%
30
1.8%
35
1.2%
40
0.9%
45
0.7%
50
0.6%
55
0.4%
60
0.36%
Probability of success
0.5
0.6
0.7
1.5
1.9
2.5
4
5
7
8
11
14
15
20
27
26
34
44
39
52
68
56
74
98
76
100
132
99
130
171
124
164
216
154
204
268
192
253
333
0.8
3.4
9
19
35
59
91
130
176
229
289
359
445
0.9
4.8
13
27
51
85
130
187
252
327
413
513
637
0.95
6.3
17
35
66
110
170
243
328
426
537
667
828
0.99
9.7
26
55
101
170
261
373
505
655
826
1026
1273
Thin Cirrus Clouds - Blind LGS
• Multiple CW LGS
suffer from
“fratricide” unless
sky is photometric.
• Pulsed laser may
tolerate ~20% loss
on both upward and
downward path.
< ~1/3 magnitude
extinction.
Fraction of Year Lost to Clouds
-CTIO 1997-1999
MCAO
CW
Pulse
Useless for
any Science
1997
46%
32%
21%
1998
49%
31%
19%
1999
41%
26%
15%
46%
30%
18%
Average
Useful Science Time Lost to MCAO Caused by Thin Clouds
Fraction of Science time lost
CW
Pulse
Classical
33.3% 13.9%
Queued
18.9%
8.9%
Failures
Mean Time Before Failure
(MTBF)
Hrs
Adaptive Optics Module
528
Beam Transfer Optics
1107
Laser Launch Telescope
25000
SALSA ( satellite, aircraft, laser 5085
Reliability
MTBF
safety system )
Laser ( minimum MTBF for
<10% downtime from failures )
Gemini South MCAO MTBF
>48
42
Overall reliability calculated from
subsystems’ Mean Time Between Failure.
Reciprocal
of sum of
reciprocals

1 
MTBFtotal  

 i MTBFi 
1
• Total MTBF is poorer than any individual subsystem’s
reliability.
•Total MTBF dominated by “weakest link in a chain.”
Subsystem reliability:
e.g., Beam Transfer Optics MTBF
Shutter & Beam Dump
Relay Optics
M6, M7 Slow tip/tilt & steering
Diagnostics
5 Fast steering mirrors
Rotator
Electronics
Overall BTO reliability
Hours
50000
6249
8333
20000
2000
50000
30000
1107
Sensitivity Analysis:
MTBF & Downtime vs. Laser Reliability
hr
350
30
%
300
25
250
20
Overall
MCAO
reliability
200
150
% Downtime
due to failures
15
10
100
50
5
0
0
25
40
48.3
100
200
1000 5000
Laser MTBF Hours
25
40
48.3
100
200
1000 5000
Laser MTBF Hours
Interference With Other
Observatories, Aircraft
• Gemini South neighbours at
CTIO and SOAR have right
of way for natural guide star
observations.
• Estimate ~1 incident per
night where MCAO must shut
down, costing ~2% lost time.
• Interference problem is
Rayleigh “light sabres”
clashing, not the 90 km
beacon itself.
Observing Scenarios for
Satellite Interference Studies.
Scenario
Slew,
Acquire
PWFS
NGS
Astrometry
Laser
Launch,
Acquire
Science
Integration
Queue
Size
N
Lost
Time
%
Full overhead
2.0
2.0
0.47
30
150
12.9
Astrometry +
Partial setup
1.3
-
0.47
30
150
11.8
Partial setup,
no astrometry
0.3
-
0.47
30
150
9.1
Calib’n. while
Laser off
2.0
2.0
0.47
3 x 15
Dwell on
same Object
2.0
2.0
n x 0.47
n x 6.6
TBD? 8.3
?
1
43%
% Time Sky Useful for Astronomy,
But MCAO Not Available.
*** Optimistic
Case:
Availability Calculation
Failures from MTBF Calculation
10.0%
Cirrus Losses for pulsed laser
13.9%
Satellite Interference ***
8.3%
Other Observatory Interference
2%
Total Lost time best case
26%
Do Sky
calibrations,
change filters
and gratings
while laser off.
(~38.5% of
time)
Availability Calculation:
Combining lost-time factors
Pr(losstotal )  1   (1  Pr( lossi ))
i
B. Subtract ‘up-time’ from 1
to get total percentage of
time lost due to all factors
A. Convert each lost-time
factor to a Probability of
Success. Multiply together
to get overall ‘up-time’
percentage.
Lost Time: Overhead, Clouds,
Satellites, Failures
Total Down time
Laser Format
Satellite Interference Scenario
Satellite
Pulse
Loss
CW
(MCAO)
Full Setup + Astrometry
12.9%
30%
38%
Partial Setup + Astrometry
11.8%
29%
37%
Partial Setup, no Astrometry
Part Setup- no astr'y. – cal. if laser off
9.1%
27%
34.9%
8.3%
26%
34.3%
43%
54%
59%
Dwell on Same Object
PDR Agenda
Friday, 5/25
0800 Laser System
0900 CTIO Sodium Studies
0915 Control System
0945 Break
1000 RTC Electronics
1045 Safety System
1100 Availability analysis
1130 Closed vendor sessions
1200 Lunch
1300
1400
1700
1800
Cost and schedule
Committee session
Committee report
Adjourn