On Saturday 27 January 2018 19:09:07 David Berndt wrote:
A new name, I'll say welcome in case no one else does.
I'll jump in here and be slightly off topic. I've recently been
setting up some ATS688s for spindle encoding on a big straight bevel
gear. Works great. Wish the duty cycle was a bit higher so I wouldn't
have to skew the A/B phases off 90 degrees. But the gear teeth aren't
square like the reference 60-0 target from the data sheet, so this was
pretty much as expected.
I also tried to use a ats668 to detect a drill hole on the gear shaft.
Forget it. Unless you're target really looks like a gear tooth an
ats668 is not what you want. Ordered some ats601s and 675s to try
instead for the index pulse.
I used ats-667's on the 60 tooth bull gear in my Sheldon lathe, used a
piece of 8-32 screw gooped to the side of the gear in line with a tooth
for an index generator. I gooped them into slots carved in the inside
face of a 1/2" thick alu plate carved to match the OD of the gear. So
they essentially can't move. They have worked well but no feedback to
the spindle, its a vfd drive.
Also keep the ats668 (and most others in the line I assume) mounted
fairly rigidly with short leads. I found when my mill head reversed
directions that some sort of magnetic coupling happens with the gear
teeth causing the sensor to vibrate and moved enough to hit the gear
briefly. I spent a couple hours trying to figure out if the torque
from reversal was moving that gear around, bad bearing, etc, until I
figured out it was the sensor moving. It's not super obvious when
you're staring deep into a mill-head head with a flashlight what's
going on.
Been there, done that, my sympathies.
On Sat, 27 Jan 2018 15:35:54 -0500, John Alexander Stewart
<ivatt...@gmail.com> wrote:
> Hi all;
>
> I am restarting an older CNC lathe conversion - an Emco Compact-8,
> fwiw.
>
> I have the 5i25 and 7i76, and 6 ats-667 sensors for spindle. (6,
> because if
> I only ordered 3, I'd end up breaking one, and shipping is
> expensive...)
>
> Any wiring issues I should be aware of? Field voltage is about 8v;
> well within the range of the ATS-667. I have some small relays for
> driving the VFD for the spindle, also driven off of field power.
>
> The Emco spindle has a 40 tooth gear, and the aluminium pulley has a
> set screw, so I should be ok with that - 40 teeth should be more
> than enough, right?
>
Depends on how much stability and stiffness you want at the lower speeds,
and the "bandwidth" of the control circuit. I had an opto-interrupter
based rig under the cover where I wasn't able to machine more that 64
slots in the wheel, with a long one for index. The noise and such from
such a low slot count made it impossible to run much Pgain, around 3.5
IIRC. Because the pwm-servo is FAST, it and the motor responded to all
that noise, hammering the gear teeth in the 2 speed head something
awfull.
So here about 3 months ago, I ordered a 1000 line omron encoder from
fleabay for a $21 bill, and drilled into the rear shaft of the motor to
mount an an extension to drive this encoder, but found I had a
differential encoder whose signal levels weren't TTL. That was solved by
a couple of the dollar ea. rs485 transceivers, which had zero problems
at the motors wide open speed making TTL outputs to feed a 5i25. I had
to write a hal paragraph to determine the gears ratios in the 2 speed
spindle in order to scale it. Then because it has two speeds, cut a dent
in the rim of the gear shift knob and put 2 microswitches on it to tell
lcnc which gear it was in, and took advantage of those switches to add a
small offset to the otherwise zero between gears speed, so that between
gears the motor is turning about 20 rpm. With one of Jon's Pico systems
pwm-servo amps, which is a full 4 quadrant controller, I can be turning
1000 rpms, reach up and grab the knob and turn it, and by the time its
moved 5 degrees, the motor is down to 20 revs, the knob continues to
turn until the gears are touching again, and the knob might resist 50
milliseconds while the gears mesh and slide into position, and the motor
speed is restored to the same spindle rpms in a few milliseconds once
the knob is fully homed in the other gear. So I no longer have to stop
the spindle, manually grab and turn the spindle for gear mesh. Doesn't
yet work if the spindle isn't running. I need to fix that...
I still use the opto kit on the spindle for its index pulse, but all the
two speed logic is in the hal file. Snippets available, just ask. And
all the quantization noise that made the head sound like a corn sheller
about to spit out broken bearings is gone, and the Pgain is 20.00.
Spindle speed is STIFF.
The scale went from 264 with the opto stuffs, to 7161.61 in high gear,
which scales to 14k something in low gear.
Now the size of the tap I can use for rigid tapping depends on how tight
a grip I can get on the tap so it doesn't slip and get out of time with
itself. 8mm is generally good, 10mm can be done with a peck cycle
wrapper. I just did a short 8mm in a piece of 4130 mine shafting last
week. Decent tap, it survived.
> Thanks for any advice/war stories you are willing to give.
>
> John.
Thats just one of my war stories, YMMV of course. :)
