SMA Technical Memo Number - 122

 

Temperature Measurements in the Mauna Kea Fiber Optic Conduits

Eric Silverberg

May 19, 1998

 

Abstract

Several months of data obtained with temperature sensors in the SMA Mauna Kea conduits indicate that the thermal stability is very good except near the smaller hand holes. Diurnal fluctuations in the middle of the long conduit runs are not larger than 0.01C (RMS). Even without insulating the hand holes, the predicted phase change in a Sumitomo temperature compensated fiber is less than 1 micron of path length per hour. Insulation in the hand holes will slightly improve the overall values. The commonality of many of the conduit paths will make the variations between distant foundations even less.

Introduction

The distance from the control building to the most distance antenna foundation on Mauna Kea is approximately 750 meters, through the path traveled by the optical fibers. The 2-3" diameter conduits are buried about 1 meter under the surface, except where entering or leaving the smaller hand holes. Four hand holes are traversed along the way to foundation 20, some large some quite small. It was thought desirable to make a direct measurement of the temperature profiles along this path to determine if any additional precautions would be needed to maintain satisfactory phase stability within the fibers.

Instrumentation

Temperature measurement instrumentation was designed by John Maute and implemented with the aid of several members of the electronic staff at the SMA site in Westford, Massachusetts and at Mauna Kea. It consists of 32 temperature sensors mounted on several lengths of multi-conductor twisted pair cable. The sensors were mounted at specific points along the cable to evenly space the readings in the conduits and to properly monitor the profiles in the vicinity of the hand holes. A schematic drawing of the circuit is shown in Figure 1.

 

The sensors are monitored with a multiplexer connected into the A-D card of a portable computer. A CVI software package was modified to allow reading and recording the 32 sensors at any interval chosen by the operator, and to run unattended for long periods of time.

 

The Data

Calibration- Three attempts were made to calibrate the thermistor sensors. The entire ensemble was placed in a temperature chamber and readings of all the sensors recorded at 6°C and 20°C. In addition, the entire group was compared after having stabilized overnight in the fiber vault. The temperature in the vault at the time was about 10°C. Unfortunately, the three sets of data do not give consistent results, with discrepancies >0.5°C common in the derived offset. Figure 2 shows the three sets of constants, derived by assuming that the average of the 32 sensors was correct in each case. The two runs in the chamber appear to be more consistent, despite the wider range of temperatures. It seems possible that the location of the reels in the vault did not allow some of them to equilibrate as well as they should. Since it is difficult to expect the two chamber runs to have agreed as shown without some truth to the readings, the two have been averaged to derive calibration constants for the 32 sensors (A HREF="122fig03.gif">Figure 3). However, even a 1-degree uncertainty in the calibrations has little impact on the conclusions, since the conduits are always operating relatively cool and close to the 5.5°C inflection point of the Sumitomo fiber.

Installation - The sensor cable was installed on the 14th of November. Weather and problems with the splicing connectors did not allow the system to become fully operational until Dec 12th. At that time the sensors were deployed according to the drawing in Figure 4. The ends of the conduits were blocked to prevent drafts from flowing along the paths and prevent stabilization of the fibers.

Data - The calibrated temperature data for 4 days after installation are shown in Figure 5. The first point to note is the huge amount of radiative heating and cooling on the exposed sensor near pad 20. Most of the temperature profiles are very flat. Figure 6 shows a subset so that several of the interesting records can be identified. You will note that most of the thermal variations occur in sensors near hand holes, with handhole 6, the smallest, almost following the ambient. This hand hole is only 5'6" x 3' x 2'-3" in size, with the fibers entering the hand hole 22" below the surface of the ground. Apparently the thermal mass of the box is so small that it only partially dampens the diurnal swings. Note the small "blips" in the record from hand hole 1. These blips appear to be the time when the hand hole goes into the shadow of the antenna maintenance building at sunrise. Aside from that, many of the sensors, representing most of the cable length, show only a small gradual cooling of about 0.1°C per day, as the site heads into the winter season. When the temperature drift is removed from the data from the middle of the longest conduits, the residuals are very small, Figure 7. In fact, the quantization effect of the A-D converter, at 0.012°C, hides any obvious terms. While many other records have been taken since the middle of December, there remains no evidence of a diurnal temperature fluctuation in the middle of the long runs at these sensitivity levels.

 

Figures 8 and 9 show a two week example of the temperatures taken in February after the conduits have begun to warm from their mid winter values. The character around the hand holes remains the same, while the longer runs exhibit nearly flat lines. RMS diurnal variations over three days in the middle of the run are shown on the insert in 9. Note that the vault currently has a very large diurnal signature, having a black top surface still exposed to the sunlight. Presumably the vault signature will drop a great deal once it is shielded by the control building modules. The vault performance should be monitored closely if long lengths of equalizing fiber will be needed to operate the array.

 

Discussion

 

Implied Performance - If we associate a length of virtual Sumitomo cable with the data from each sensor, it is possible to infer the phase fluctuations we can expect from the vault to the antennas. The distances associated with the sensors that lie within the hand holes are obvious. The sensors near the hand holes are not so straightforward. We do know that sensors even a couple feet from the hand holes, such as 6, 8, 13, 21 show almost none of the characteristics of the hand hole and are much more nearly like the middle of the long runs (see figure 8). To be very conservative, 3m of fiber has been associated with each of these "nearby" sensors. The rest of the fiber between the hand holes has been divided among the evenly spaced sensors in that path. The associated lengths are given in Table 1. The total, 745 meters, is the length of required fiber from the vault to foundation 20.

 

 

 

 

Table 1: Lengths of Fiber Associated With Each Sensor (meters)

1

2

3

4

5

6

7

8

2.1

2.1

9.1

24.4

61.0

3.0

3.4

3.0

9

10

11

12

13

14

15

16

24.4

24.4

24.4

24.4

24.7

3.0

1.5

21.0

17

18

19

20

21

22

23

24

3.0

3.0

82

82

3.0

3.0

1.8

60

25

26

27

28

29

30

31

32

60

60

60

60

3.0

4.6

-

3.0

 

 

 

 

The phase variation in the Sumitomo fiber was predicted by Moran, based on data taken in the SMA IF laboratory. The formula

 

 

Delta(meters) = L (meters) x 8 x 10-9 x (T(oC)-5.5)2

 

 

fit the data quite well near the inflection point. This formula was used to combine the lengths in Table 1 with the temperature data associated with Figures 8 and 9 to obtain a prediction for the phase variations that would have been seen had the Sumitomo cable been in place. These variations are plotted for this two-week span in Figure 10. While we can clearly see the overall rise in temperature and effect of the sensors in the hand holes, the rate of change and magnitude of the variations are unlikely to degrade the SMA data.

 

Figure 10 also shows the diurnal variations associated with each sensor over the same period. Note that the total is most greatly affected by sensor 23, located in hand hole 6. Sensors only 1-2' on either side of the box show much lower contributions in spite of 3 times the associated length. Next highest contributors are the vault and the two foundations.

 

Insulating the Hand Holes - It was suggested that we identify the impact of insulating hand hole 6 on future data. A roll of fiberglass insulation was cut into lengths and stacked over the fiber cable in the hand hole. Total thickness of insulation was such that the space appeared about 1/2 full. This simple exercise reduced the thermal cycling on that section of cable by about an order of magnitude, from ~7oC p-p to ~0.4oC p-p. The result was about a 10% or a 1 micron reduction in the predicted peak to peak phase variations from the vault to pad 20. Figure 11 shows the predicted phase variation for a two-week period in April after the insulation was installed in hand hole 6. The insert shows that the associated diurnal phase contributions for each sensor are now more nearly equal. (Channels 22 and 23 have been swapped since February and sensor 22 is now in hand hole 6.) Note that the diurnal signature gets larger as the ducts get farther and farther away from the 5.5oC ideal temperature.

 

With Electrical Cables - In as much as the bulk of the conduits appear to be very temperature stable, it is tempting to ask whether or not we need the Sumitomo temperature compensated cable at all, but could instead use electrical cable, at least for the inner rings. After all, many of the fluctuations seen in any single cable will be similar for all paths and cancel each other during the observations. The portions of the cable in each of the foundations should act nearly the same.

 

Ultra stable electrical cables, such as those referred to by the brand name HELIAXÒ , distributed by Andrew, have electrical length changes at these temperatures of about 4 x 10-6/ oC (see FSJ1 and LDF2). A simulation was done which compared the predicted phase at a foundation #6 antenna vs the phase at a foundation #10 antenna. We have direct thermal measurements as far as foundation #10 and can predict the thermal character in foundations 1 and 2 by using the cable temperature of pad 20. The thermal data from mid December (Figure 5) was combined with the temperature coefficient to produce the simulation shown in Figure 12. In short, while it might be possible to operate with electric cables on the shortest paths, there would obviously be a huge loss in the phase stability.

 

Conclusion

 

The good thermal stability of the conduits, coupled with the characteristics of the Sumitomo cable, permit the distribution of signals to each foundation well within the requirements for the interferometer.