Re: SERVO

> "Arne Chr. Jorgensen" <instel@...> wrote:
>
> Hi everyone,
>
> Some comments:
>
> Bertho Boman: As far as I understand what you wrote, then what
> you try to do is just to make an external servo controller. There
> are many of them around, - you just have to add the bridge. These
> get the velocity from the encoder as you say. There is also asic DSP
> chips with a lot of functions included. Anyway - all the software
> you need is difficult and a lot of hard work. So - why make another
> one ?
> As far as I know, - you will get some trouble to "sync" several
> axis, - and this has to be feed from your PC ( if you don't put in
> G-code interpreter into it ). The way I understand it - you are
> back to where EMC started out, and have tried to solve.

---------> snip
===================================================
Not at all Arne. If you look at for example Jon Elson's servo amp, or any other "real" servo amp, you will see that it takes
and analog signal as input and compares it with the velocity feed back from the analog tach. It also has current feedback loops
but lets ignore them for a minute.

What I am saying is that the PC's analog control voltage +/- 10VDC is coming from a D-A in the PC which the EMC or other
software writes two bytes to define the output voltage. Basically, we could move the D-A from the PC onto the servo amp board
and everything would work just as before. The only difference would be that we have a digital interface between the PC and the
servo amp instead of an analog.

Next step is to digitally read the encoder pulses and in logic calculate the speed. That information can be feed to a D-A and
we have now created an all electronic tach signal that can be feed into any servo amp. We just replaced the electro-mechanical
tach with an electronic equivalent one. The hardware and software exist to day, nothing new.

Our system and performance is still identical as before. Note, the encoder signals are always going to the PC and are being
monitored by the motions control software.

After doing what I suggested as an mental exercise we have a servo amp that has digital velocity input that will be converted to
an analog signal and a digital encoder signal that will be converted to an analog velocity feed back signal. These analog
signals will be amplified and filtered and then converted to a digital signal again (PWM)

My final point is that we do not need to do the digital conversions. Do the math directly on the digital inputs and digitally
create the PWM.

Same performance, and it is still fully controlled by the motion software. We just got rid of the hardware tach and a bunch of
analog circuitry, plus we can now tune the servo amp in software without even opening the case.

Bertho