The reports of 05 April 2003 and 10 April 2003 describe the observed impact on pointing of what appears to be an incorrect application of the F2 term of the track model. The TCS error reported in the latter leads to a table describing the errors incurred. It is important that the sign convention for this table be clear. The case of F₂ is presented here.

• The tabulated version of the track model lists values of (F₁,F₂,F₃) for certain values of azimuth.
• At a position given by (azimuth,elevation) = (az,el) the value of F₂, viz. f_2,az, should cause the azimuth of the telescope to be adjusted by +f_2,az*sin(el) arcseconds on the sky in addition to the corrections applied using the terms of the 7-parameter telescope model.
• Hopefully this term makes a correction that counteracts the irregularities in the track at (az,el) such that a POINTING (or FIVEPOINT) executed at this position shows the telescope to be pointing exactly where it ought.
• The erroneous TCS, described here, used an adjustment of +f_2,az*sin(el)*cos(el).
• Thus the telescope position was in error by
  delta(az) = +f_2,az*sin(el)*(cos(el) - 1)
• This value is the one tabulated.
• Note that a subsequent POINTING/fivepoint would suggest that the antenna would need moving by
        - f_2,az*sin(el)*(cos(el) - 1)
• . . . and this would be the number recorded in the pointing logs as daz, and plotted in the various plots throughout this website.
• So, for instance, a SCUBA jiggle-map taken using the erroneous TCS code, and ostensibly centred at (az,el), was really centred at (az + delta(az), el), and azimuths extracted from the map should be corrected by -delta(az) to yield 'true' azimuths.

I hope that's clear.