Re: EHP External Hardware Proposal

More thoughts:

I think Hans is going full speed but we need to talk about our goals.

A few of my suggested goals:

Keep it simple and do not reinvent the wheel. Make it possible for persons to either use what is commonly available or easily
modifiable.

To make an external multi-axis, coordinated controller board is a significant undertaking. To further complicate it by adding
the servo/stepper amplifiers on board would be a very large task to say the least. Besides the hardware there would be tons of
software required.

Doing a one channel DSP (or Micro-P) based controller like I described is a reasonable group project. That would separate the
motion control software from the servo amplifier and the tach feed back and it would also remove the need for an expensive
custom PC I-O card. Actual I-O could be either through the parallel port or even opto isolated RS-232.

The software to drive it only needs to be modified slightly: Instead of writing the the command to the PC I-O card D/A it sends
it out on the printer port or the equivalent step speed command. Instead of reading the shaft encoder counter on the I-O card
it reads it from the printer port. I think it would be relatively easy to modify CNC software to do the above.

That means multiple existing software could be used with slight modification since none of the timing or feedback loops are
changed for the software. Only the local servo constants would be modified but they are normally external anyhow.

Neither Hans or I are happy to have to fiddle with Linux but if the installation and RT kernels issues are streamlined, EMC
would be an excellent choice for the software. That would also remove the need to modify EMC to make it work well with
steppers. An other possible alternative would be if Hans or someone with the right experience can take a look at what it takes
to run EMC under a slightly modified NT to get reasonable RT performance as I mentioned in a previous post.

A small single channel board with PWM amps that can be used with either steppers or servos sounds good. Hey, stack up six of
them and we have a hexapod! It would solve several problems:

1. Pulse generator for steppers with no frequency limit or jumps
2. Amplifiers for plain steppers.
3. Encoder feedback for steppers.
4. Servo Amps with digital gain/ filter control.
5. Encoder, home and limit switches interface.
6. Electronic tach for servo feed back
7. No PC I-O card required
8. Easy opto isolation
9. Since it is a very low cost single channel board it can be used for simple motion control of just about anything.

To the EMC "crew": Is it practical what I suggest and would you be willing to support it?

Hans and Windows programmers: Have you taken a look at what is required to run EMC on NT?

Who has DSP experience?

Step up to the mic. Let your suggestions and comments be heard!

If this is pulled through, I can have the boards built on my production lines at cost (I'm not in this to make money) as long as
we can run a reasonable size batch each time. That way, software and non-electronic persons still can be part of a large group
and we can standardize our effort.

Bertho Boman
=======================================

> From: "Carlos Guillermo" <cnk@...>
>
> Hans -
>
> I've got a pretty basic question. Are we talking about offloading
> interpolation duties to external hardware? As in sending an arc-move
> command to the external micro and then letting it execute the coordinated
> motion? Or, are we talking about merely updating a count register for each
> axis' "commanded position" and letting the external micro handle the servo
> loop, PID gains and all? Or, are we just talking about blind stepper pulse
> train generation...? I think I've heard a little of each from the various
> people interested in this system architecture, and was just wondering what
> the external micro's duties would be.
>
> Carlos Guillermo