CTDELAY to frame time

Mark and I mapped frametime as a function of ctdelay.  The scope was connected to the reset pulse.  We were running an ND pattern - during the N frame, there were no reset pulses and during the D there were lots of resets.  We could measure the time on the scope where there were resets pretty accurately.

We had the following data:

CTDELAY          Frame Time
0                           137 msec
5                           273
10                         411
15                         549
25                         827
33                         956
50                         1509
70                         2055

Using CTDELAYMAP.JOU, I did a linear fit to these data.  The 33 ctdelay point was the only one below the fit and the y-intercept for the fit was 130.7 msec, not the 137 msec we got in the data.  Removing the 33 ctdelay point improved things.  I remeasured the 33 ctdelay point and to

33                        1036 msec

Putting this into my fit, I got a nice linear function where:

Frametime = 27.40 msec * ctdelay + 137.1 msec

Solving for ctdelay given a frametime, the relation is:

CTDELAY = ( Frametime [msec] - 137.1 ) / 27.40