[mmaimcal]Re: [Almasci] Antenna Acceleration

[Richard Hills]

Dear Mark,

(This is a response to the message before last.)

Whoa!  All I was trying to check was that the resolution on the transfer of 
data in the form of one position and one velocity per 48msec time interval is 
adequate for the most extreme motions that we might contemplate.  That was the 
issue raise by Morita-san.  The numerical tests of cubic spline interpolation 
simply show that it is possible to reconstruct the original waveform with good 
fidelity.

This says nothing at all about the actual performance of the servo.  That 
depends on lots of things to do with the dynamics of the telescope and the way 
the servo is implemented.  Clearly if it was the GTB on the end of the motor 
shaft we would not expect it to be able to follow these waveforms.  I am also 
sure that the performance would be pretty hopeless even with the actual ALMA 
telescopes if the servo were just a simple one driven by the error between the 
actual position and the (interpolated) demand position.  I assume that the 
implementation makes use of the fact that the required positions and 
velocities are known a little way ahead and it is therefore work out the 
appropriate torques to apply.

I therefore think that the actual performance (in terms of errors between the 
commanded and actual postions) will be a lot worse than indicated in the plots 
I sent, but I still think they will be tolerable given that we can read back 
the actual as opposed to commanded position and that this is in any case only 
the turnaround part of the trajectory, which shoudl be the worst case.

As far as the "whirly" paths I talked about some time ago are concerned, in 
fact I was trying to make those rather smooth (just using sines and cosines) 
to make it easy for the servo to follow them.

Best Richard