Summary
Measurement of the 'transit error' step-size can be made using inclinometry. The initial measurement agrees with contemporaneous measures on the sky. The method provides a daytime method of monitoring this characteristic. The new measures also suggest that the corrective algorithm is currently overcorrecting the defect.

The most recent attempt to measure the 'transit' step-size by tracking an astronomical source through transit suggested that the 8" step being used was an overcorrection, but the data were of a rather poor quality. It was suggested at that time that a repeat experiment was needed : here it is :

The data clearly show an overcorrection by the algorithm by about 5". The curve fitting suggests that the (semi-)amplitude at the horizon be reduced from 4.0" to 1.3". At this level the effect is almost negligible, but since its size may vary with antenna balancing we cannot henceforth disregard it, sadly.

Allsky pointing was also performed the same night. The behaviour of the elevation residuals as a function of azimuth is shown below :

The data show a 4.7" difference in elevation residual between east and west, in agreement with that measured by the transit tracking experiment above. Adopting an average of these 2 figues of 4.8" - or 2.4" in the semi-amplitude, and therefore 2.7" at the horizon - the value of SCURVE.AMP in the telescope model was changed from 4.0" to 1.3" on 20 July - i.e the full-amplitude of the defect is now expected to be 2.6".

During daywork on July 19, an inclinometer was placed on the TMU and the telescope tipped over to about 10 degrees and the TMU positioned so that the inclinometer was horizontal. A fictitious source was tracked through transit. The elevation range was such that the inclinometer stayed within 200" of the horizontal, and therefore, within its range of utility. Adopting a scaling factor of 20mv/" as determined for all our inclinometers, the tilt of the TMU in acrseconds is plotted below as a function of azimuth (with a zero-point of 180 degrees). Click on the image for a better view :

The inclinometry-tailored version of the telescope control software omits application of the corrective algorithm. A defect is visible at azimuth 180.2. A quadratic curve was fit to the data and a portion of this fit is shown below :

. . . and the residuals of the data about the curve are shown next :

Despite the noise at each end of this segment, and the uncertainty in finding the quadratic curve that best fits the data immediately surrounding the defect - and which impacts the baseline through this segment - the data may be interpreted as revealing a 3" (full-)amplitude step, centered at 180.2, and of semi-width 0.1 degrees. The (full-)amplitude at the horizon is then 3"/cos(10) which is still essentially 3", in very good agreement with the measurements on the sky reported above.

Conclusion
The inclinometry and sky measurements agree in their estimate of the current amplitide of the 'transit-error' step-size (now ~2.6"). The inclinometer may be adopted as the routine method for monitoring of this defect.

Acknowledgement must be made that Adrian Webster (UKATC) first suggested this inclinometry technique. Thanks also to Firmin for yet another adjustment to the inclinometry software, and to Peter for help with the levelling.