On Sat, 7 Mar 2020, Gene Heskett wrote:
Date: Sat, 7 Mar 2020 07:06:41 -0500
From: Gene Heskett <ghesk...@shentel.net>
Reply-To: EMC developers <emc-developers@lists.sourceforge.net>
To: emc-developers@lists.sourceforge.net
Subject: Re: [Emc-developers] nother Q for Peter C.W.
On Saturday 07 March 2020 00:13:38 Gene Heskett wrote:
On Friday 06 March 2020 21:50:15 Jon Elson wrote:
On 03/06/2020 05:38 PM, Gene Heskett wrote:
I have now put 2 ATD667LSG's on this block and have failed to get
a working index pulse into linuxcnc thru the encoder index pin of
a 7i76D.
Note these sensors are open-collector, so need a pull-up
resistor.
I've mounted it branded face against a block of alu, with a green
3 pin socket to allow unplugging it quickly, and I jbwelded a
piece of a nail to the side of the drawbar cap/bolt retainer at
the top of the spindle, but I can't get a false out of it at any
orientation of the nail or a tool waved past it.
The sensor is on the end where the leads come out of the
package, but at right angles to the
leads. That's the "branded face". So, it sounds like you
have them mounted backwards.
That cross patterned face is then a target to be machined into the
mount to locate it more precisely? I'm in the house but I now seem to
recall that its the smooth side that's facing the gear in the lathe.
Duh. It takes a 16mm projector lens and damned strong light to read
the text printed on the "branded face". What the hell were they
thinking?
You're saying that the depressed cross in the 5% smaller face is not
the sensed face? That the crossed pattern should be the face glued to
the mount? In that event I also have them wired reversed polarity.
Which shouldn't hurt them as they are rated for up to 18 volts of
reversed voltage. But by now, I'll have to do some chisel work to
remove and likely wreck a 2nd one. Sigh. This is what they mean when
they say an education costs money... I only bought 6, paying about $2
more from digikey because all the $6 ebay ones were in the middle of
the virus in china and 6 weeks away.
Thats also an excellent example of a piss-poor 14 page document they
supply. That cross grooved face is much easier to identify. And mount
to the smooth face, the jbweld oozing out can also cause lead shorts
much easier.
The gear tooth needs to wipe sideways across the sensor to
be registered. So, a gear
tooth that is parallel to the leads will be sensed as it
wipes across the face of the sensor.
Or in this case, 1/2" of a 2" nail, jbwelded to the outside of the
drawbar retainer cap which will swipe across the smooth face as the
spindle rotates.
Now I am pissed at me.
Motion of the tooth will be at right angles to the leads.
Check if the sensor +5 V is drawing any current (should be
about 7 mA). Are you sure you have not
connected to the "test" output (pin 3)? They don't say what
this pin does, just tell you "tie to ground or
leave open".
Which I've always clipped off...
Jon
Thank you Jon.
Now, it looks to me that the only non-ambiguous drawing in that whole
document is the one on the lower left on page 8.
But read the text. It seems to me that the rotation reversal is going to
diddle the vertical positioning if there is any encoder counter reset
allowed while the locked condition is in effect. It doesn't give me any
confidence that the vertical positioning will be error free at all.
What I do know for sure is that if because the tap is too big to pull in
one g33.1 pass due to lack of motor power and I have close to 2hp I can
surge it to, and I write the rigid tap routine to peck the operation by
driveing the tap only 1/4 to 1/2 a turn per "Peck" until the depth has
been obtained, that the repeated passes give me an obviously loose fit
to a bolt, which I have until now blamed on a backlash take-up error.
What I haven't done is played with the backlash to see if the somewhat
sloppy thread obtained by peck tapping can be improved. But I also don't
know if a counter reset is done anyplace but at the instant the
down-stroke is started. But that text tends to tell me that since the
pulse polarity depends on the rotational direction, then the phasing
relationship is going to be diddled by quite a few counts at the bottom
of the hole when the turnaround and backout move is begun.
How can one determine if thats the case?
Its going to be very hard to see on the halscope unless a 50 u-s base
thread just to run the halscope is setup, something I don't have now.
This is, I think. probably in the 5i25 firmware and I'm not qualified to
troll thru that. I can't even imagine how that might be alleviated, even
by an automatic edge detector. Even if the polarity is reversed with the
direction of rotation, it has to be reset probably several counts ahead
of the true zero point as the nail crosses over the hall effects gap,
before it can send that edge again on the next rotation. So which edge
do we use to get the least error? Seems to me the first edge should be
ignored because that would be the reset edge to be thrown away. Given
that my encoder count per revolution is something north of a scale
setting of 14,000 when the head is in low gear, a 5 degree error because
its triggering on the first edge, can be quite a few counts difference
depending on the direction. So the second edge as its "tooth" sweeps
across the gap, would be far more accurate. Some sort of logic to
control which edge is used, controlled by the instant state of the a/b
detectors direction output.
Can you shine any light on this, Peter? Or am I years late to this
discussion?
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)
If we desire respect for the law, we must first make the law respectable.
- Louis D. Brandeis
Genes Web page <http://geneslinuxbox.net:6309/gene>
_______________________________________________
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
The encoder index is edge triggered but index is only used in one direction
(and at a constant speed) at the beginning of every peck cycle so the index
width
is unimportant.
Peter Wallace
Mesa Electronics
_______________________________________________
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
