Re: servo update freq.

[Jon Elson <elson-at-pico-systems.com>]

Dave Engvall wrote:

> list;
>  
> In the discussion of the new control board there was a passing comment 
> about the load to emc caused by the servo update frequency.
>  
> Anyone care to stick their neck out on a semi-quantative or quantative 
> relationship between axis velocity and the update rate necessary to 
> track.
>  

There is none.  The real relationship is system bandwidth vs. servo 
sampling rate.
On low powered machine tools, like a Bridgeport with 1/4 Hp axis drives, 
the
large signal bandwidth is a couple of Hz, and the small signal bandwidth 
might be
100 Hz, tops.  1 KHz is totally adequate for this application.

Now, on really high performance machines, with 25 - 50 Hp (peak) axis 
drives,
the ones designed to make sharp corners at 1000 IPM, the bandwidths are much
higher, and so the servo sampling rate needs to be higher, too.  Optical 
stages
can have high bandwidth because they are so light, so they, too, need higher
sampling rate.

> I get the impression from Jon's posts that even a P200 might stay with 
> three axes at 2K and maybe 10K update rates.
>  

You probably need a bit more than that.  But, my measurements are that a 
333 MHz Pentium II
can do 3 axes in 50 uS (with my parallel port interface, ISA should be a 
little faster, and PCI
should be much faster).  That allows 10 KHz updates for 50% CPU load.

> To further complicate this post: is there an easy way to quantitate 
> frequency response for a given servo/machine combination?
>  

Yes, but you need to know peak linear force developed by the servo 
drive, and the mass of the
table.  Other things like electrical and mechanical time constants of 
the motors, spring rate
and natural frequency of the leadscrews, damping coefficients set into 
the servo amps, etc.
all enter the equation.

> I get the feeling most machines are pretty slow....at least if one 
> thinks about it in audio terms.

I have a dynamic signal analyzer, and have run servo response graphs of 
my servo amps.
I was amazed at how high the electrical bandwidth was, I could hear the 
motor whining
up to several KHz!  But, due to motor inertia alone, I know none of that 
response was
actually getting to the machine table.  It was kind of interesting to 
see and feel the machine
following the moves commanded by the analyzer up to 30 or 40 Hz, though.

In audio terms, this is like a BIG sub-woofer, certainly no response 
over 100 Hz.

Jon