Opto interrupter precision

Electric pendulum clocks seem to be getting interesting again and a number of people are using opto interrupters to sense the pendulum. I'm interested in how precise these can be - people have reported results that indicate clocks using then for amplitude sensing are achieving less than a second of arc precision.

So I set up an experiment using my CNC mill as a positioning device with a Sharp GP1A57HRJ00F sensor. I changed the microstepping on the X axis driver to get a minimum increment of 0.2 microns, and used a probing routine to do 100 successive measurements of probed position, using either a thin shim "blade" or a 5mm rod in the chuck to interrupt the beam.

The results were quite precise, with a standard deviation of sensed position of ~0.15 micron provided that the sensor had ambient light kept off and supply voltage kept constant. We could debate just how accurate a measurement to this precision could be but it was very consistent.

This website also reported some measurements on an Omron device, using a manual technique, that showed repeatability to at most 1 micron, for application as limit/home switches.

My conclusion is that the devices should be very suitable for precise amplitude control.

They can be even better with some additional electronics and a slotted target. I did some work years ago to look at replacing expensive high precsion home sensors on a instrument in production.
We used a target with slot in the middle. The obscuration as the target entered the slot "armed" the circuit (We used a PIC microcontoller but a dual flip-flop and bit of logic will do) and then the narrow slot gave the actual home position. The coupler LED had constant current drive and the output was fed to a comparator with accurate reference. The sesor was shielded from ambient light. The target was thin laser cut stainless steel to reduce alighnment issues.
We were getting sub-micron repeatability. Measurement was using a Rennishaw ML10 laser interferometer. I could easily measure thermal expanssion of the 100mm long test rig if I moved the reflector from one end to the other.

The most important things are keeping ambient light off the sensor, keepng LED current constant (using just a resistor is not good enough with a 5V supply due to variation in LED Vf with temperature. A resistor is OK with a suply > about 9V) and using a comparator on the sensor output. A logic chip with schmitt trigger input is OK.

Robert G8RPI.

That's useful to know Robert, thanks. Using constant current did you look at the vcc sensitivity of the detector itself? Not sure what you mean by "arming" the circuit, was that just a characteristic of the detector or did you do something specific?

Blooming forum, Just typed a long erply and lost it.

Short version:

sensor optotransistor ran from same regulated supply as the comparator so not much variation and the signal and set point are ratiometric.

Arming is so circuit can tell the difference between nothing in the opto and the target slot. Basically the sequence is:

A/ Nothing in optocoupler, photo transistor "ON" - not home
B/ Solid part of target enters optocoupler, photo transistor "OFF" Circuit "arms"
C/ Slot in arget aligns with sensitive axis of optocoupler, photo transistor "ON" - in home position.

As a bonus step B is used to switch the speed of the axis during homing to a slower speed allowing faster more accurate homing.

You don't need the arming for a pendulum detector. If you have trouble screening the pendulum from ambient light, a long, curved target that obscures the slot throughout the swing, with slot or hole a trigger point, will help because it will block ambient light as well as the LED.

Robert G8RPI.

OK, understood, thanks.