On Sunday 10 March 2019 15:45:06 John Dammeyer wrote:

> > -----Original Message-----
> > From: Nicklas Karlsson [mailto:nicklas.karlsso...@gmail.com]
> > On Sun, 10 Mar 2019 11:57:53 -0500
> >
> > Jon Elson <el...@pico-systems.com> wrote:
> > > On 03/10/2019 05:18 AM, Nicklas Karlsson wrote:
> > > >> Hey Chris,
> > > >>
> > > >> AFAIK, the Pi has better HDMI support compared to the Beagle
> > > >> but then
> >
> > it was designed more for multimedia.  The Beagle PRUs have the
> > advantage that they have access to some of the Beagle Processor RAM.
> >  That gives them a bit of an advantage over the PI/32 bit hybrid
> > using SPI.
> >
> > > The PRUs are 200 MHz 32-bit RISC processors.  There are 3
> > > shared RAM data sections between the PRUs and the ARM.  The
> > > PRUs have a limited number of direct IO pads that bypass all
> > > the ARM high-level I/O fabric, and so can be read/written at
> > > the 5 ns rate of the PRU, which is a big plus for various
> > > special bit-fiddling I/O tasks.
> > >
> > > Charles Steinkuehler wrote a general driver for Machinekit
> > > that uses the PRU for step generation, PWM and encoder input.
> >
> > But why use a PRU then this kind of hardware is usually part of the
> > micro controller and very cheap?
>
> Jon,
> You make a very good point.  Why bother, with the new 32 bit
> controllers that do so much even bother with the antiquated idea that
> the LinuxCNC PC should close the loop on positioning.  Running
> encoders from a servo motor back into the PC to some sort of
> controller board that creates +/- 10V out is now not really a good
> option.
>
> Whether it's a PC with a plug in card running LinuxCNC or a BeagleBone
> Black running MachineKit there's not much point. Or is there?
>
> For a lathe, spindle tracking in order to move the carriage and if
> also tapering at the same time to track and move the cross slide does
> require something tightly coupled.
>
> So a Pi with an external STM32 may well send step/dir signals or SPI
> based messages to the driver but tracking that spindle encoder for
> threading is a bit more intensive.
>
Which is what I'm doing on the Sheldon, with an alu  bracket, curved to 
match the size of the bull gear on the spindle. Biggest problem was my 
math, getting the pair of ATS-667 hall effect sensors properly spaced to 
get a decent quadrature timed set of pulses out of it.  For index I 
reversed the third one, and I have a piece of a 8-32 screw gooped to the 
side of the gear so the end of the screw matches a tooth. Its only a 60 
tooth gear, but thats 240 edges per revolution, and works really well. I 
do not have a pid in the spindle control, its a vfd. In making that 
10-24 thread on the back end of a shorter stylus for the touch probe, I 
forgot to put a g4p1 in front of the g76 to give the spindle some accel 
time to get to 100 rpms, so when the g76 starts the first time, I hear 
the z accelerating with the spindle, but on thinking about it, the z is 
just following orders, which is as it should be. I didn't fix it.

Approaching the correct size, I wound up with a fit I smoothed somewhat 
with the locknut, and which fits the plastic spider a heck of a lot 
better that the shlopppy darned near stripped factory fit. But I picked 
up enough noise on an unshielded testing cable that my first test run 2 
days ago blew some of the 74HCT245's in the C1G rev 4, my fav BoB, and 
Arturo does not supply schematics, so its shotgun all 4 of those, which 
I have, but theres also 3 or 4 74HCT04's which I might have to order. I 
can see the probe working on the cards leds, but its not getting to the 
pc.

I have got to see about getting this 6040's framing all grounded together 
in spite of quite a few layers of epoxy paint on everything.  Its going 
to be fun stringing 1/4" braid thru all the cable chains just to get a 
ground on the spindle motor, picking up the rest of the gantry on the 
way by.  Then find and break all the ground loops that will no doubt 
cause. Fun? Not!

My point is that on a lathe, with spindle power to spare, for threading, 
all you need is for z to take orders, and that pretty straight forward, 
no real pid magic needed. All you need is a fine grained enough spindle 
encoder, and 1.5 degrees of resolution seems to be more than enough when 
you have a relatively high motion smoothing mass in the spindle and 
chuck.

On the G0704, with the spindle in low gear, I have 360/14250
(nominally)=0.0252631578947 degrees per edge, and I can run plumb 
ridiculous amounts of Pgain in that spindle PID. Speed control is STIFF.
Rigid tapping, up to the 17 amp current limit set in the servo amp, just 
works. And thats as close as I'll ever get to getting a 2200 lb 
Percheron sized horsepower out of that PMDC motor.

> John Dammeyer
> http://www.autoartisans.com
>
> > > Jon
> > >
> > >
> > > _______________________________________________
> > > Emc-users mailing list
> > > Emc-users@lists.sourceforge.net
> > > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
> > --
> > Nicklas Karlsson <nicklas.karlsso...@gmail.com>
> >
> >
> > _______________________________________________
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
>
> _______________________________________________
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>



_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to