hello - new forum

[snoopy27]

The Pico universal stepper control board is in my mind the best solution for controlling steppers with Linux being more adaptive than window based products. It would be interesting to see what it could do with a Mach software driver because of the support Mach has and its much easier instaalation and interface.

[jmelson]

Originally Posted by snoopy27 The Pico universal stepper control board is in my mind the best solution for controlling steppers with Linux being more adaptive than window based products. It would be interesting to see what it could do with a Mach software driver because of the support Mach has and its much easier instaalation and interface.

Interesting, and I'd like to see it, but I don't know enough about ring zero Windows drivers or have the time to do it myself. The driver for EMC2 is, of course, public, and could be used as a starting point. The environment Art Fenerty created for the Mach driver does not allow floating point arithmetic, so some changes would be needed. Also, a method for loading the tuning data for the PID servo loops would be needed for the servo versions.

Jon

[m37]

Jon
as art is "semi retired" but still tinkering with mach
why don't you see if he will work on a driver for
your board?
mike

[jmelson]

Originally Posted by m37 Jon as art is "semi retired" but still tinkering with mach why don't you see if he will work on a driver for your board? mike

I talked to art at (I think) the 2007 CNC Workshop at Roland Freistad's shop in Galesburg IL.
One of the big problems is that the ring zero environment that Art's driver runs in does not allow floating point to be used. I think that is a total hard requirement, and you couldn't finesse it in any manner. Windows device drivers should have no need of floating point.

The existing driver in EMC uses some floating point, but that could be changed. The problem is we would need to provide a PID servo control loop ALSO in this driver, and floating point computation would be a definite help in that. But, PID and general servo control can be done in fixed point, it was done that way for years in the "old days". Another concern is how to accomplish tuning of the servo loop. That requires a lot of connections between the driver and the user space environment. In EMC2, we have a real-time graphing program, modeled after a digital oscilloscope, called HalScope. I generally look at instantaneous encoder velocity and instantaneous following error at the servo sampling rate on this while making test moves. I also have a program that allows you to view and manipulate all the servo tuning parameters. So, you tewak a parameter, jog and watch the scope trace to see how you are doing.

I have no idea how much effort would be involved in getting this to run in Windows and with Mach.
Art certainly didn't seem to be chomping at the bit to implement this.

Jon