On Sat, 23 May 2020 22:16:36 +0100, Peter Fairbrother
Back in the 80s I built my first CNC positioning table. It was a
simple X Y table that I bolted on to my Bridgeport table. I used
stepper motors on it and snap action Micro switches for homing. I used
the Micro brand but I think any high quality snap action switch would
just as well. They repeated reliably within .0005" total. They were so
reliable that I stopped worrying about checking position all the time
and just made parts. The biggest problem I had was the steppers
themselves because they would lose position if I made them move too
fast. Now I just use servos. In any case, try a snap action switch,
I'm sure you will be pleased.
On 23/05/2020 23:55, email@example.com wrote:
I was wondering about the switches used to locate objects to be machined
- do you know anything about them?
0.0005 is good, but I'd like a micron or less if possible. This is a
small envelope mill (converted/beefed-up BCA) used to make small parts.
On Sun, 24 May 2020 18:50:12 +0100, Peter Fairbrother
It sounds like you need some sort of digital encoder. You can buy
linear encoders with .0002" resolution, which is half a micron, for
pretty cheap on ebay. Then use the encoder as the switch. The encoders
are pretty standard and are easy ti interface to. For info try
I use their products. They sell chips for pretty cheap that interface
with digital encoders and these chips can be used for getting a signal
that is very easy to use. For example, I have an encoder that is
driven by a shaft. I use the signal from that encoder, conditioned by
a chip, to provide the signals for a servo amp that in turn drives a
If you are looking for a mechanical switch that has 1 micron or
less resolution then I can't help.
If you use linear encoders then you can also use the encoder for a
position display. You could even hack a typical digital readout to get
your position signal using one of the chips from US Digital. And I'm
sure these chips are available from other vendors closer to your neck
of the woods. I use US Digital because they are close to me and
shipping is fast and cheap.
On Tue, 26 May 2020 00:41:23 +0100, Peter Fairbrother
Digital encoders often have a "Z" line that is used for homing. All of
my rotary encoders do and at least one of my linear scales does. If
you were to look at the lines on a linear scale you would see a whole
bunch of lines in a row. Every so often will either be longer lines or
separate lines above the row. These lines are used for a homing or
marker pulse. So what you do is use a mechanical switch that is close
to the marker line and when the switch is triggered you move slowly to
the marker or Z line. The read head will have 3 LED and photodetector
pairs. These are A, B and Z. The A and B pairs are staggered, so
either A will lead B or B will lead A, depending on the direction of
travel. This allows the scale to be read in quadtrature, so you get a
resolution that is 4 times the native resolution of the scale. The
read head will have A, B and Z wires along with ground and V+, usually
5 volts. So if a read head that comes with the linear scale has 5
wires it is a pretty good bet the scale has the Z lines on it.
I wonder how accurate they actually are. I have a Sylvac digital DTI
which indicates to 1 micron but the calibration chart shows the actual
deviation along the 25mm travel and that varies from 0 micron up to 2
micron and with a quoted sensor error of 2 micron it says maximum error
maybe 4 micron (sensor + scale error).
Well, there is accuracy and there is resolution. But mostly it doesn't
matter in machine tools as the error is over a large distance and you
just care about the resolution over a small amount of travel. So a
part is made, measured, and then the machine cuts are adjusted to get
the desired part dimensions. Over time the machine warms up, cutters
get dull, tool pressure goes up, and the machine cuts are adjusted
again. Even when it comes to inspection indicators are rarely used
over their full travel to make absolute measurements but are instead
used as comparators. Even excellent screw micrometers can have very
high resolution but a relatively high error over the full travel
distance. Same thing with machine tool ballscrews. To get better
accuracy with machine tool positioning the errors in the leadscrews
are mapped using standards, gage blocks for example, and the errors
are programmed into the machine tool parameters. Then the machine tool
takes these errors into account when positioning. Sometimes it is just
the backlash of the leadscrews that is measured and sometimes it is
the actual positioning error at several places or even over the
On 26/05/2020 23:51, firstname.lastname@example.org wrote:
I understand those items and the calibration chart gives the +- error
along the range and is mainly 0 or +-1 micron, it's only 2 micron at 1
point, so if I want to do some really accurate stuff I can chose my
range and have workshop quality (grade 2? UK made) slip gauges to check
if I want that level of accuracy. Temperature can be critical even in
less accurate items as the heat resulting from machining can throw out
the readings by some 0.001", so if important I go away and leave things
to stabilise before taking final cuts. One of my neighbours has a home
made CNC router with a 3' x 3' bed and he quoted some accuracy and I had
to correct him as the machine is in a shed and not temperature
controlled so a mainly aluminium machine will move but he realises this
and for what he does it isn't important, others he knows don't and
assume the machine is as accurate as the ball screws. Ball screws can
vary, a company I used to work for made force and torque testing kit and
the ball screws were made in Japan IIRC but later moved to a new plant
in Europe and they found the original offerings from that new plant
weren't to the same standard of accuracy even though they were supposed
Yes I do, for example the probes produced by Renishaw.
You do not rely on the movement of a switch toggle because
that is unpredictable as to when it will flip over because
temperature effects affect the springiness of the spring
and there can be variation in the distance you have to move
the carriage to make the switch flip over.
So, that is what you DON'T do.
What you have are two contacts, possibly brass, that are
in contact with each other, one fixed, and one movable.
The movable contact is held under very light pressure just
enough to maintain contact.
What you arrange is for your carriage to come along and just
lift the movable contact OFF the fixed contact, and that point
of disconnect, no matter how diminutive the movement, is your
This is something that you can make for yourself.
Just to summarise, you are looking for the break in electrical
contact to indicate that your carriage has arrived at your
fiducial point, and the point of that breakage is reproducible
to a gnat's cock of distance.
Actually Peter, a chap I know, not a hundred miles away from
you in The People's Republic Of Trowbridge, used this very
scheme for the homing positions when he developed his own
multi-colour 3D printer with an 18" cube building capability,
although he used 4 (NSEW) fixed contacts and one movable
contact with the ball end being the actual contact so a
slight movement of the probe would result in at least one of
the contacts breaking. That does mean 4 sets of electrickery
because you can't just wire them all together because just
one contact breaking would be masked by the continuing
contact of the other three.
With these sorts of detection, the movement required at
the sensor is just the distance of one molecule to achieve
Remember this? :-) ...
Two little atoms
Walking home from school
Bumped into each other
And made a molecule.
On Sun, 24 May 2020 18:50:12 +0100, Peter Fairbrother
I was way off in my dimensions, as David pointed out. I don't know
what I was thinking. Anyway, I can measure accurately to 1/2 micron
with my 20 millionths of an inch per division indicators. I have had
to make parts that were round within 30 millionths of an inch so
that's why I have the inspection equipment to check this kind of
stuff. And temperature REALLY changes dimensions at that resolution.
Even touching a part with your fingers can make the part measure out
of round because of uneven heating. Is your machine really accurate
enough to want sub micron resoultion? In any case a digital encoder
would still probably be the cheapest way for you to get sub micron
positioning. The way CNC machines determine position using digital
encoders is to use a switch for coarse position feedback and then the
machine moves slowly until it sees the index on the encoder. You will
need to do the same if you don't want to constantly overshoot your
position marker. Just how small are the parts you are making? And how
accurate do they need to be made?
On 25/05/2020 18:19, email@example.com wrote:
Us metric folks use "hundredth's" almost exclusively as the measurement
division while hand machining - that is 1/100th of a millimeter or 10
microns. This division is approximately equivalent to the imperial "few
We are taught to estimate the dial reading to a tenth of a division, ie
one micron, though we would only do that on the most accurate parts.
My analogue micrometers read to a hundredth and can be estimated to a
micron or two, my digital mikes read to a micron. My (small, A4 size)
granite surface table is accurate to 1/4 micron. Wanting a part accuracy
of a few microns, requiring a basic mill reference accurate to +/- 1
micron or better is not really a big deal.
(I don't do submicron work - as yet anyway)
For this mill a large part would be 80mm long and 50mm diameter. 10mm to
30mm part dimensions would be more typical.
Mill is a converted BCA with hella-expensive 2mm pitch ball screws and
chunky overpowered 200 step motors using 16 microsteps/step to give
0.625 micron resolution, and a 30,000 rpm 600W ER8 fluid damped spindle
to reduce cutting forces.
XY repeatability (without zeroing) is within 2 microns using LinuxCNC
motion control software. With present zeroing (standard microswitches)
repeatability gets considerably worse.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.