From harwood@hgea.org Thu Oct 2 09:01:07 2003 Date: Wed, 01 Oct 2003 15:12:25 -1000 From: Jim or Pat Harwood To: denault@jeans.ifa.hawaii.edu Cc: Tokunaga@ifa.hawaii.edu, techgroup@jeans.ifa.hawaii.edu Subject: Requirements meeting, item on Software Tony, When you brought up at the requirements meeting the following bullet item under "7. Software", and my written comment about it, I too was kind of puzzled at first, since my comment did seem like overkill: [Statement in document:] Calculate Target (RA, Dec) to mount (HA, Dec) coordinates at minimum rate of 10 Hz. [My comment:] 10 Hz may well be insufficient. I would say 20 Hz minimum. You will probably wind up doing this important calculation at 100 Hz. I was probably thinking of the selection of the control cycle frequency when I wrote that. There is no need, as you said at the meeting, to update the target to mount corrections at such high frequencies in order to precisely position the telescope. On the other hand, with today's computing power, there is no need to compute the target to mount corrections at a slower rate than the control cycle. Currently, with the LSI-11, I do the intensive mathematical computations to correct for mount flexure, atmospheric refraction, and nutation/aberration/mean to apparent/proper motion at slower frequencies, from every 10 sec to every second depending on the item, rather than load down the computer with those calculations at the 10 Hz control cycle. It is quite a bit more complicated to do the target to mount calculations at a different update rate than the control cycle. You have to compute what the magnitude is of the correction projected out over the calculation period (10 sec for example), compute a rate from that figure, and then output that correction rate to the tracking table to be applied over the next calculation period. When the next calculation update time rolls around, you then correct any accumulated position error over the interval (due to the rate approximation) by computing the position correction for the current telescope's position, then repeat. This is done for each of the types of position correction. Most of this is unnecessary if you simply compute the target to mount position corrections each control cycle. No correction rate needs to be derived. The computed position corrections are loaded into the tracking tables along with other inputs each 20 Hz control cycle. With today's computing power, there is no reason not to simplify by updating the mount corrections at the control cycle frequency, even though it is "overkill" from a positioning point of view. -Jim ==================================== Jim and Pat Harwood, at home 1928 McKinley St. Honolulu, HI 96822 harwood@hgea.org ====================================