The University of New South Wales 
School of Geomatic Engineering 
GMAT8001   Thesis

Temperature Induced Changes in the Lines-of-Sight of Two Topcon DL-101C Digital Levels

by Gary Donovan 
Supervisor: A/Prof.  J. M. Rüeger


Two Topcon DL-101C precision digital levels (Serial Numbers NJ0100 and NJ0106) were evaluated to determine the changes in the line-of-sight as a result of temperature acclimatisation.

This was achieved with a modification of the method used by Sheil in 1997. The line-of-sight changes of two instruments were evaluated simultaneously, with one being the control instrument and the other acclimatising to ambient (laboratory) temperature after spending eight hours in a temperature test cabinet at either +45°C or -10°C. Figure 1 shows the experimental setup, with the two digital levels mounted on the same heavy-duty tripod using a special mounting bracket.

Figure 1 - Setup of Test Equipment 
Photo of front of test setup

The difference between their respective staff readings was assumed to be solely due to the temperature acclimatisation. The temperature induced change in the staff readings over two sighting distances was used to separate the line-of-sight deviation into a vertical shift (Dd) and an angular component (Dg). These are closely dependent on the temperature induced change in the height and collimation errors, respectively.  The shift and tilt terms are illustrated below in Figure 2.

Figure 2 -The Two Components of the Line-of-Sight Deviation 
Diagram showing how the line-of-sight deivation is divided into shift and tilt terms by repeated tests at two different distances

There were 10 separate tests involving both digital levels (Serial numbers NJ0100 and NJ0106) at both sighting distances (2.0 m and 9.8 m), over both temperature change cycles (cooldown from 45°C and warmup from -5°C). The experimental data for each test was compressed into one value summarising the total change in line-of-sight of the instrument. Using the data from several tests, the shift and tilt terms (Dd and Dg) for each instrument were determined, along with their 95% CI's (Confidence Intervals). These results are summarised in Table 3:

Table 3 - Summary of Results 
Table summarising results

Note that the cooling cycle for instrument NJ0106 was determined three times, but the three results are partly correlated.

The shape of the graphs obtained agree well with other researchers. The calibration of the line-of-sight deviations cannot be directly compared with previous results, owing to the substantially different techniques used. However, they are similar in magnitude and sign to testing by other researchers (eg. Schauerte (1997)).

The expansion of the digital level's case is probably a significant component of Dd in the table above. However, even extreme refraction effects are minimal when compared with the angular deviation of the line-of-sight.

Several unexplained measurement anomalies were encountered during the tests.  Figure 4 below displays some of these.   Each "Test DL Height" data point represents one measurement in the first 30 minutes of the test and the mean of two in the remaining 3.5 hours.  The precision of a data point is +/- 0.044 mm in this test with a sighting distance of 10.9 m.   The scatter is worst between 30 minutes and 150 minutes.  The graph shows a 0.06 mm downward tooth between 32 minutes and 51 minutes and another (0.05 mm) between 118 minutes and 150 minutes.  A (permanent) drop of 0.11 mm occurs between 149 minutes and 162 minutes.  The figure also shows that two consecutive measurements (being a mean of two measurements) can differ by as much as 0.27 mm.  The reasons for these anomalies are not known.  The anomalies should be investigated further to establish their inpact of the use of the DL-101C for precise digital levelling.

Figure 4 - Graph of Some Measurement Irregularities 
Graph of Test #5, Test DL, Height difference measurements

The testing procedure was significantly aided by the writing of DLay, a small program to control the DL-101C by serial cable. Both the DL-101C remote control interface and DLay are very primitive in nature, and there is great scope for improvement in both.

Further Information...

For more information, contact:

back arrowBack to Surveying & Spatial Information Systems Home Page   «»   revised 16 January 2000
Original: Gary Donovan   «»   Last revision: Gary Donovan
© 1999 School of Geomatic Engineering, UNSW, Australia.