ACIS Door Failure Summary

Failure Description

At 1245GMT, June 18, 1998, we were preparing to run the AXAF Stray Light test. We gave the command to open the door per the Long Form Functional Test Procedure, paragraph 4.7.5 using Side A of the PSMC drive electronics; the Detector Assembly was thermally stable at -60C. The "closed" indicator switch transitioned off as the actuator reached 80C, but neither the "transition" nor the "open" indicator switches toggled. The last actuator temperature reading prior to an automatic thermal shutdown showed 133C (the shutdown is set to trigger at 135C). By subsequently looking at the radiation pattern of the Fe(55) door source, we could determine that the door had not moved any substantive amount.

At this point in time we had been under constant vacuum for a total of 34 days, including some exposure to cold survival temperatures. During testing at Lincoln Labs, XRCF and BASD we had usually opened the door within a day of pulling a vacuum, and had never before gone through cold survival temperatures first. Last cycled under ambient pressure on November 18, 1997, the door had remained closed during the second BASD thermal/vacuum test in January 1998 and throughout ISIM integration onto the Observatory at TRW, which included an Observatory-level acoustic test prior to going into the vacuum chamber.

In a recovery attempt, we put ACIS into the bakeout mode for 13 hours, reheated the "open" actuator to 70C, and gave the "close" command. Although the microswitch indicators indicated normal motion, including transition, the "closed" indicator failed to toggle and the actuator drive shut down automatically upon reaching an over temperature condition. Upon repressurizing the chamber two days later we verified that the door was sealed.

Background Information and Surmises

Starsys has run tests for LMA in the past which characterised the maximum temperature which could be safely reached before the (internal) rupture disk would break as a function of shaft travel. Full travel of 0.75", for instance, was safe up to 150C. Travel restricted to 0.2", however, was safe only to 95C. (The "transistion" indicator toggles at 0.3" of travel.) If the mechanism was stuck so that essentially no travel was possible, then at temperatures far below our cut-off limit of 135C the rupture disk would have broken. (One needs to set a limit of greater than 130C to assure that the mechanism can operate under all full-travel conditions.) Starsys did not think that we could reset the "closed" indicator without first performing a full-motion travel of the now-presumably-failed actuator, which is consistent with our experience commanding the door closed.

The engineering unit Detector Assembly had been run through 250 cycles for life testing; about a third of those cycles had been at -60C (the rest at ambient and an intermediate temperature). The maximum shaft torque required to open the door was measured to be 18 inch-pounds; the actuators are capable of delivering 75 inch-pounds.

There were, initially, three possible failure mechanisms offered.

The bottom line as one approaches disassembly from the ISIM is that we have no good theories as to probable cause.

Flight DA Door Operation History

Location Date DA Temp Direction MaxAcctTemp
Lincoln 3/2/97* 25C Open 111C

3/6/97* -60C Close 133C

3/6/97* -60C Open 108C

3/7/97* 25C Close 124C

4/8/97 (--) Open/Close (--)

4/9/97 -60C Close 121C

4/9/97 -60C Open 110C
XRCF 4/17/97 25C Open 113C

4/26/97 25C Close 119C

5/8/97 -60C Open No Tlm Data

5/18/97 25C Close 123C
BASD 10/20/97 25C Open - A 106C

10/28/97 -60C Close - A 110C

10/28/97 -60C Open - B 111C

10/29/97 25C Close - B 110C
BASD-Ambient 11/14/97 20C Open 102C

11/14/97 20C Close 104C
* These first entries are actually the EU Detector Assy.

Failure Analysis

Borescope observations of the crack between the door and its seating plane showed a visible separation on that 80% of the periphery which could be inspected in situ. However, ohmmeter measurements indicated that there was electrical continuity between the door and the housing, lobbying strongly for a cold weld.

At LMA the door opened at ambient at a torque of 19 inch-pounds, which is right on the expected value; at first inspection, no mechanical anomalies were noted.

Furter testing with engineering units under a variety of mechanical tolerance extemes, cold, vacuum, and high moisture content in bearings and seals failed to uncover any combination of conditions which would raise the required torque to open the mechanism above 25 inch-pounds. The actuator assembly as delivered by Starsys is guaranteed to provide a minimum of 70 inch-pounds. Subsequent testing on both the door-close actuator and the shards of the burst disk from the door-open actuator established that the actual torque delivered by the unit before failure was a touch greater than 90 inch-pounds.

By reviewing the data logs we can see that the door-close actuator was not actuated during the 6/18 incident, so a "mystery command" from the S/C -- though seen by others on several occasions -- was not implicated. A related event, however, had happened in early February. At that time the S/C bus voltage was being ramped up very slowly (to protect S/C fuzes) and resulted in tripping several pulse commands in the PSMC. One could hypothesize that such an event might trip both enable and door-open commands, trying to open the door against an atmospheric pressure of about 500 pounds holding it closed. The entire event would be over -- with a blown disk -- before the S/C telemetry system had even been turned on! Such an event does have a shadow, however, in the Starsys actuator temperature. This temperature reaches a peak of 10C above ambient (in a vacuum) about 30 minutes after the power is removed. Unfortunately in February the TRW EGSE had not yet had the full telemetry database loaded, so this temperature was not recorded. Starsys also believes that we would not have subsequently had the proper "door closed" indication on the 18th had the disk been blown on some prior occasion.

Data Analysis

An analysis of the time-temperature profiles of various prior door transitions yields the following interesting table. The time increments are in units of Science Major Frames, which are 32s long. (Under "Switch Events" the systax /Closed/ means the Closed switch transition to "not closed".) One needs to know a couple of additional things about the system to properly interpret this table. The first is that the power is always applied at the beginning of a 32s Major Frame (by procedure), and the switch positions are sampled by the RCTU early in each Major Frame; the temperatures, by contrast, are sampled late in each Major Frame -- this fudge has not been factored into the presentation below. The second thing is that it takes the actuators about 2s per degree C to warm up at this voltage; the 6/18 event started at -40C, so it took about 2 Major Frames longer to reach the same absolute temperature as the 10/28/97 runs, for example.

Door Operation Time/Temperature Profiles


A Open A Close B Open B Close A Open A Close
Date: 10/20/97 10/28/97 10/28/97 10/28/97 6/18/98 6/19/98
BusVoltage 30.2VDC 30.3VDC 30.3VDC 30.9VDC 29.4VDC 31.2VDC
Housing T: 25C -60C -60C 25C -60C 25C
Starsys T: 20C -12C -10C -5C -40C -7.5C
MF # Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp
-1
36.1
-2.1
7.6
23.8
-2.1
19.5
0 Power On 48.3 Power On 23.8 Power On 27.4 Power On 30.6 Power On -2.1 Power On 30.6
1
67.9
48.3
50.0
50.0
19.5
53.1
2 /Closed/ 79.3
63.5
65.7
65.7
38.5
60.9
3 Transition 89.3 /Open/ 73.9 /Closed/ 75.8 /Open/ 76.7
53.1 Transition 77.6
4
104.3
81.9
86.9
91.6
63.5
90.0
5 Open 106.3
96.2
99.9
107.4
72.0
105.1
6


109.6 Open 111.1 Closed 110.3 /Closed/ 79.3
119.4
7

Closed 108.1




87.7
132.5
8








99.2

9








108

10








118

11








127

12








132

13








133

One could ask if the Starsys door mechanism housing was suffering the same sort of thermal lag when heated (e.g.: when we went into bake out and warmed the housing to 25C). Although the ISIM interface seems to have been colder at TRW than at Ball, no unusual lag seems evident at the following shows; time increments are 120 Science Major Frames or about 1 hour.

Bake Out Temperature Profiles

Date: 10/26/97 1/17/98 6/19/98
Time DA Starsys DA Starsys DA Starsys
0 -62 C -33 C -70 C -35 C -62 C -40 C
1 hr. -35 -33 -40 -35 -35 -38
2 hr. -13 -28 -28 -33 -15 -35
3 hr. 3 -23 -10 -28 0 -33
4 hr. 15 -15 5 -23 13 -25
5 hr. 18 -8 15 -15 18 -20
6 hr. 20 -3 18 -10 18 -15
7 hr. 20 -3 18 -5 18 -13
8 hr. 20 0 18 -3 18 -10

For completeness one can sample what the valve actuators have done; it is the identical Starsys mechanism after all. But there is no news here; alll is perfectly normal.

Valve Operation Time/Temperature Profiles


A Open A Close A Close A Open B Close B Open
Date: 10/20/97 10/28/97 5/28/98 5/28/98 6/6/98 6/6/98
BusVoltage 26.9VDC 30.2VDC 32.3VDC 31.7VDC 30.5VDC 29.9VDC
Starsys T:

10.7C 10.4C -22.7C -22.7C
MF # Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp Sw.Event Act.Temp
-1
38.6
7.7
30.7
27.5
19.6
23.8
0 Power On 38.6 Power On
7.7 Power On
30.7 Power On
27.5 Power On
19.6 Power On
30.7
1
54.6
14.4
42.8
40.8
27.5
44,8
2
64.6
46.6 /Open/ 59.9 /Close/ 57.3
46.6
58.6
3 /Close/ 72.0 /Open/ 61.1
72.9 Transition 70.0 /Open/ 61.1 /Close/ 67.9
4 Transition 79.3 Transition 72.8
89.6
77.6
71.0
76.7
5
84.4
77.6 Close 98.4 Open 90.8
80.2
86.9
6
93.9
88.5
110
104
90.8
96.9
7 Open 103
108




102 Open 107
8
104 Closed 108



Closed 113


Viton(R) is a registered trademark of DuPont Dow Elastomers
Return to ACIS at TRW page.
There have been 78 accesses to this page since 9/14/2022.
Comments & Complaints: goeke@space.mit.edu
Last modified: Tue Oct 6 14:56:03 EDT 1998