[nworroll]

Working on my Early 80's YCM30 Mill. It had a flaky Bandit III Allen Bradley control unit.

I removed the Bandit crap out of it. I finally got my servo figured out and working. A +/- 10V signal on the inputs makes each axis move one way or the other. My initial control will be a manual control with a Readout of the quadrature encoders on a PC. The manual control will be sets of adjustable voltage regulators for +/- to control each axis. I built a ISA expansion card using a LSI chip LS7566. I made software that successfully talks to the board, just need to actually hook up the encoders and expand the software to get the info from the chip and process it to a real measurement.

So that's the easy part.

Phase 2 is to create a second ISA card to take commands for the computer and drive the axis'. The current idea is to have a Digital to Analog converter (DAC) chip. This would have a +10v and -10v going into it. The chip would output 0 volts with a 127 (in decimal) put on its data lines, a full +10v with 255, and a -10v at 0, and a linear percentage in between. So this would set the required feed rate. Then I am thinking of a timer circuit to switch the DAC output on for a certain amount of time. So if I wanted to move the xaxis 1 foot at a feet rate of 1in per second, I would set the DAC to an output voltage that equals 1inch per second and set the timer for 12 seconds. The software would take G codes and turn it into movements like that. With some math combining different feedrates on a shared timer should make diagonal lines. I figure for arcs/circle, I would have the software break the circle up into really small lines and send that to the controller.

Phase 3 will probably entail making the ISA expansion card 'smart'. Using a micro controller to directly accept G-codes and do some thinking on it's own.

So the questions -
1) With the servo hooked up and no inputs, the axises are harder to turn by hand but I can still move them by hand. While I'm moving them a red light does turn on and then off when I stop moving it. I believe it's the light labeled "set". I was under the impression that with the servo properly powered up the motors would be caged and would resist any manual movement. The servo does have a tach signal being feed from each axis. Is that how it should work? Or is it the job of the control software/hardware to sense movement from the encoders and resist that movement?

2) How does the TYPICAL controller hardware/software use the encoder readings? For instance, the machine wants to move x axis +10 inches. Does a typical controller send it the signal to move it 10 inches, and A) verify after movement with the encoder reading to insure it went where is should, B) verify as it's moving with the encoder to verify it's moving where it should, and stop where it should, C) don't verify with encoders and assume it moved 10 inches. Encoders are only for the Display.

3)Arcs and circles - Does a typical software/hardware break up the arc into little straight lines or do they use some math to actually vary the feed rates over time to produce curves?

Thanks!

Obviously it would be more 'practical' to just replace the servos/drives with step/direction units and MACH 3 software. But this is for personal use, and I can't invest much money in this. And this is a fun learning project for myself.
Nick

[Al_The_Man]

So that's the easy part.


2) How does the TYPICAL controller hardware/software use the encoder readings? For instance, the machine wants to move x axis +10 inches. Does a typical controller send it the signal to move it 10 inches, and A) verify after movement with the encoder reading to insure it went where is should, B) verify as it's moving with the encoder to verify it's moving where it should, and stop where it should, C) don't verify with encoders and assume it moved 10 inches. Encoders are only for the Display.

It depends on what type of drives you have as to how it holds or cages at rest, older tach feedback drives were typically velocity controlled , now tach is not generally used as modern drives are used in the current or torque (trans-conductance amplifier) mode.
Also without encoder or under the control of the CNC controller there will be no feedback control, this is known as the PID loop and when a command is issued the controller attempts to keep the following error as close to zero as possible, whether moving or at rest.
For more information on the type of feedback controller you are trying to build and details on the PID loop tuning, see the instructional video's on the Galil Motion site.
Al.

[nworroll]

It is an old tach feedback drive

This is a thread from a while ago when I was trying to get it to work, you gave me some info then. - http://www.cnczone.com/forums/servo_...ld_bandit.html

[samco]

You should look at emc2. It already does what you want and has inexpensive hardware that would interface with your drives/encoders.

LinuxCNC.org - Home

With mesa you could get a 4 axis interface package for <$190 (epp printer port interface +/-10v interface dautghter board) + 24 i/o
For less than $290 you could get a pci 4 axis solution + 48 i/o.

(and many other combos of host/daughter boards)

The best of all is emc2 is free and open source.

sam

[nworroll]

That looks interesting and affordable, but again this is a educational experience for me, so I'd rather build the hardware and software myself.