Hi Andrew,
It has been a few years since I added a spindle encoder to my lathe. I
have a web page describing the gruesome details that might give you some
ideas and some gcode to play with for cutting the disks. This encoder is
still in operation with the data being fed into a parallel port on an
older PIII 700Mhz Del box running ubuntu 10.4.
https://ve7it.cowlug.org/spindle-encoder.html
cheers
Lawrence VE7IT
On 2022-03-31 10:10, Chris Albertson wrote:
On Wed, Mar 30, 2022 at 11:18 PM andrew beck <andrewbeck0...@gmail.com>
wrote:
Hey guys will this work OK?
I think it's the same as the one from amazon
https://www.trademe.co.nz/3531064367
And I'm thinking I may need two discs. One above the other so I can use a
index sensor on the top layer. Or maybe stick a proxy sensor on it.
Looks like they have a 6mm gap between the slots in the sensor. Do you
think a 1mm wide gap gap would be ok maybe in a 2 mm thick sheet?
The link does not work for me.
Ideally the slots and the space between the slots is the same. But this
only really matters if
1) you intend to double the resolution by looking at leading and trailing
edges.
2) and when the spindle reverses direction there is a small error if the
geometry is asymmetric. Maybe the error is too small to break a tap, I
don't know.
If the plate is thinck relative to the slot width all that means is the
alignment most be good, with the IR beam exactly 90 degrees to the disk.
The Omron senors, when you look inside use a very thin disk, like brass
shim stock. This minimizes any error caused by misalignment. The very
high-end sensor use glass disks with vapor deposited aluminum to block the
light.
Don't bother with two disks, because alignment can never be 100% perfect.
Simply drill a hole for the index or make one slot deeper.
A trick used in automotive sensors is to have one missing slot. Then
software resets the counter when the expected pulse does not happen This
saves the cost of the third sensor but that only matters if you build a
million cars. IR sensors cost only about $1.
On Mon, 28 Mar 2022, 07:35 Chris Albertson, <albertson.ch...@gmail.com>
wrote:
Forgot to include a link. You can buy the bare sensor but also you can
buy
them on PCBs with connectors attached and ready to go.
amazon.com/Measuring-Optocoupler-Interrupter
<
https://www.amazon.com/Measuring-Optocoupler-Interrupter-Detection-Arduino%EF%BC%885pcs%EF%BC%89/dp/B08977QFK5/ref=sr_1_17?keywords=c+optical+sensor&qid=1648405019&sr=8-17
Best to use a thin aluminum plate to make the interrupter disk. Make
the
slots 50% duty cycle, You need three sensors for quadrature with
index.
The above sensor should directly interface to Measa or a parallel port
or
some microcontroler.
On Sun, Mar 27, 2022 at 10:02 AM Chris Albertson <
albertson.ch...@gmail.com>
wrote:
THis goes to 10K RPM? that is 167 revolutions per second. 400 slots
means 70 KHz signal.
Can you design an inductive sensor that does 70KHz but is also
sensitive
enough for near zero RPM? Optical sensors can work up to 1MHz with no
problem as they don't have any inductanve. You can buy a "C" space
sensor for a few dollars
In either case, optical of inductive, you want a 50% "slot" where there
is
as much metal left as cut away Then the "edges" of the square wave are
the
same in either direction. The 50% duty cycle in effect doubles the
sensor resolution.
Why? Think about the signal when the direction changes. You have
two
sensors in quadrature and lets say one is being blocked and one is not,
I
think you want the turnaround to take as long for each sensor.
On Sun, Mar 27, 2022 at 2:22 AM Andy Pugh <bodge...@gmail.com> wrote:
On 27 Mar 2022, at 08:09, andrew beck <andrewbeck0...@gmail.com>
wrote:
And is bigger dia better for higher resolution?
Probably not, as it will perhaps have more scope to oscillate at
10,000
rpm.
Don’t neglect the purely mechanical design of this fast-spinning part.
Have you considered optical sensors rather than inductive?
What will be counting the pulses?
LinuxCNC can lathe-thread with one sensor.
But rigid-tapping is different as it needs to accurately detect the
reversal point. So you need three channels for index and full
quadrature.
Work out what error you can accept in the reversal point detection
(as a
fraction of thread pitch) and you can get a feel for how many slots
you
need.
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
--
Chris Albertson
Redondo Beach, California
--
Chris Albertson
Redondo Beach, California
_______________________________________________
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
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
