Advice on implementing DRO servo loop in De0-nano-soc FPGA (Hostmot2)

[god...@gmail.com]

Hello All:

First of all, I hope that I am making myself clear to everyone. (NOOB here!)

I have a mill setup that is working with LinuxCNC on the mesa 7i43 with both X and y Drives currently. My goal is to implement a DRO readout feedback loop into the FPGA DE0-Nano-Soc with machinekit. I would like to make a type of DRO feedback servo loop with the FPGA in realtime using the FPGA as the counter decoder/resolver functions. After still trying to understand machinekit I am looking for advice to see if this would be the best way to approach this problem.

I have magnetic slide digital readout with no usable output digital signals. But after probing inside the unit (Sony unit with magnetic slides) I can tap off a digital logic chip that creates a square wave in retaliation to another pin on the circuit card. Both signals would move in either direction of the magnetic slide. (basically covering which direction it is moving). The signals would be independent of the DRO itself. (for example, resetting the DRO to zero would not affect the signals into machinekit).

I guess in essence this would be a replacement of the resolver function without all the math of signal processing? If I am correct? Also, having "limited" experience with Verilog HDL, I would be creating a HDL block that would either run separately or be included into the HostMot2 function block in the mkscofpga github Project.

I understand the risks associated with a servo feed back loops such as this.

Hence, why I am seeking advice.

Your thoughts??

Thanks!

Mike Kennedy

[Chris Dickson]

Pretty untimely I know -- but I've been running a benchtop mill with closed-loop servo control + cheap chinese (igaging) slides, but took a (very) different approach which works well for me :). Haven't updated this repo for the past couple of years but for what it's worth -- see below... 

----------------------------------------------------------------------------

linuxcnc / machinekit configuration + miscellaneous setup scripts for "Ludd" - a homebrew machinekit + beaglebone-based CNC controller for a converted Sieg (aka Grizzly/Harbor Freight, etc mini-mill)

Although there are many CNC conversions to be found on the interwebs, this one is unusual in that it's set up for closed-loop/hybrid servo motor control + position feedback, using a pragmatic combination of cheap/surplus/hobbyist hardware and improvised, soldering-iron friendly electronics.

The design employs high quality, analog input / analog feedback AMC motor amplifiers (to be found new on ebay for ~ $30 each) driving a set of Parker/Parvex PM servo-motors (rescued from the Twin Cities' iconic emporium of everything liquidated, imperfect or obsolete - Ax-Man surplus)

LinuxCNC position feedback is provided via a USB/serial stream of absolute position values, obtained from a set of arduino-monitored 21-bit capacitive scales.

The scales (originally added to provide low-cost DRO) had been thrown in for free with the mill when it was obtained for $300 through Craigslist. They were adapted to serve for this project by replacing the DRO readouts with an Arduino set-up to decode the scale positions (aided by some nifty reverse engineering by Yuriy Krushelnytskiy. Viz: http://www.yuriystoys.com/2012/01/reading-gtizzly-igaging-scales-with.html )

The position data is supplemented by direct tachometer-driven/analog velocity control loops between the motors and amplifiers.

Amplifier input position control is also continuous/analog (+/-10v), synthesized from micro-transformer isolated 20khz PWM signals, using a salvaged RS232/423 driver chip coupled with a simple low-pass filter.

The position control PWM is generated within the Beaglebone PRU, using the (unaltered!) machinekit stepper control driver originally written by Charles Steinkuehler.

Code + circuits (again, for what little they are worth), at https://github.com/christophermdickson/ludd

Cheers!

Chris