Re: [CAD_CAM_EDM_DRO] PID primer, was Dedicated mb for PID

Doug Harrison wrote:

> I just looked at the pdf for the LM628 on National's site. While intended
> as a datasheet it is nonetheless the best primer on PID loops I've read to
> date.
>
> Now I see what the sampling interval for D and the integration limit are
> for - they keep the filter from blowing up. Still, they are simple to
> implement. 10KHz should be easy in any compiled language.
>
> Go here if you want a primer on PID.
> http://www.national.com/ds/LM/LM628.pdf

I'm not sure the LM628 and LM629 are very useful. I looked into them quite
deeply at one time. I also talked to the applications engineers at National
about keeping multiple axes synchronized. My worry was about circular
interpolation, which is not really supported in the chip. The scheme is that
you leave the chips in run mode, and update the endpoint and velocity
registers while it is running. Since everything is really done in fixed point,
there will be roundoff errors, and the axes will drift out of sync (perhaps
VERY slowly). I suppose the main CPU can read real-time positions and
add compensating factors to realign the separate axes, but that sort of
defeats the whole idea of using these chips. As they are QUITE expensive,
($50 - 80 depending on speed and package) it just doesn't make sense to use them
unless you really get a large benefit out of them.

Jon