resonance of a circuit that includes an optical path to measure distance.

Thread Starter

adouglas88

Joined Jul 4, 2021
7
Hey all, I am thinking about a device for optical distance measurement, and wondering why it hasn't been done already/ if it could work. Basically it works like this:

You have an emitter, perhaps a laser diode.

A reflector. Not a retroreflector, they have their uses but in this case I am thinking a ball bearing. This way the reflection occurs from exactly a known point on the surface of a sphere, which has many advantages.

a receiver, a photodiode of the type used in telecom, for instance, so capable of fast response.

An oscillator, whose output is to the laser diode, and whose input is the photodiode.

The goal here is to make a circuit which oscillates at a frequency that is a precisely known function of the optical path.


Thus, the changes in the distance from a reference point on the emitter/reciever thing, perhaps the center of the emitter, and the center of the ball bearing can be calculated easily, from the frequency at which the circuit oscillates This is interesting because while time periods are hard to measure, frequencies are easy and cheap to measure to very high accuracies. If the frequency can be converted with a precisely known function to a length, and vice versa, that opens very important doors for measurement from one point to another. The application I have in mind is CNC machinery.

I know it wouldn't be trivial to make this work, filtering, perhaps frequency filtering, might be in order. A phase locked loop that locks to the resonant frequency could be part of it, you can see. My main concern would be signal to noise ratio, there is not much light returned by the reflector. I can think of some means of improving that, such as a shade and a filter for the laser light.

I don't see why it would not work though? Precise point to point distance measurement would have major implications for the cnc machinery industry. Currently, optical distance measurement is very expensive and labor intensive because it is not point to point, and is complicated and expensive in other ways.
 

Papabravo

Joined Feb 24, 2006
21,158
The thing that is not clear to me is the component that would determine the length of the optical path and change the frequency of the oscillator.
 

KeithWalker

Joined Jul 10, 2017
3,063
It would only work over a very short range because most of radiation from the source would be reflected away from the receiver by the curvature of the ball bearing. That is why they usually use a retroreflector.
The concept would require rather complex circuitry to make it work reliably. Would it offer any advantages over existing distance measuring methods? Timing reflected pulses has been a very successful and reliable way of measuring distances for over two miles with very good precision since it was introduced by Hewlett Packard in 1976.
 
Last edited:

Beau Schwabe

Joined Nov 7, 2019
155
I've used optics in a "quenching" effect to determine small distance as well as optical clarity by passing the light through a medium where the light received controls the brightness of the light being transmitted. Less light received, the brighter the transmitter ; more light received, the dimmer the transmitter.

A derivative of this is how TOF measuring sensors work, BUT in the TOF situation, the light is pulsed. You are talking about a continuous stream of square wave pulses due to the nature of the time ON vs. the time OFF you suggest. The analogy would be like the feedback squeal you hear from a microphone too close to the driven speaker, but only with light instead.

Once you get past the propagation delay and reaction times of the discrete components in your circuit it 'could' work providing you have a fast enough processor. You will have different propagation paths for received HIGH signal vs a received LOW signal which is why pulsing in a conventional TOF may be used. It only has to deal with one propagation path.

Honestly I would look into fiber optic cards used for computers. Many years ago I used to work for National Semiconductor and we were converting many of our cards from copper to fiber. One thing that was really cool to get my head around was that we were transmitting data so fast through several thousand meters of fiber optic cable that the data we were sending was completely in transit within the cable before it reached the receiver card. IOW, the transmitter card had stopped sending, and the receiver card was still waiting for the data which was still propagating within the cable.

Note: light travels 1 foot in just under 1 nano second
 

MisterBill2

Joined Jan 23, 2018
18,167
The challenge will be in the numbers. Time of flight will demand a rather precise measurement of a very short time interval. The big problem is that the response time of both emitters and detectors is mostly much greater than the time of flight of light over a short distance.
So while it is an interesting concept it will be a tricky thing to make function. Using it as an oscillator is a valid concept, but the TS does need to work out the numbers to verify that it can actually be produced in a reasonable way.
 

Thread Starter

adouglas88

Joined Jul 4, 2021
7
Hey all, yes, thanks for the thoughts, it is interesting to be part of a community that can actually discuss such things.

Hm. I'm not terribly interested in a patent because I think it is probably a basic enough idea that it would not be patentable. Some of the details of how to make it work might be.

Yeah, it's those numbers.

The advantage is that we can easily measure a frequency to within a single PPM or so, with only a few bucks of components. A timed pulse is much harder to measure. A PPM means a micron per meter, which is
 

Thread Starter

adouglas88

Joined Jul 4, 2021
7
whoops, pressed enter and it submitted before I was done :

...quite a high degree accuracy.

I agree, the retroreflectors greatly increase the amount of light received, and that is the main reason they are used. However I think this may still be doable. In the retroreflector situation, I believe they are trying to use interferometry. This imposes one wavefront of light over another, and the amplitude of the resulting fringes must be detected quantitatively. Quantitative detection of such brightness levels is quite imprecise and requires a relatively large amount of light.

If you think about it, the signal to noise ratio that can be teased out by modern electronics is quite high. Consider the bomb guidance systems that use low powered lasers to "paint" a target. The bomb steers towards a spot of laser light that is only a few milliwatts, reflected by a poor reflector.

A radio signal has an extremely low signal to noise ratio, and yet with filtering it can be plucked out well from the background.

Factors that could be used to increase signal to noise ratio: the light is monochromatic, so a colored filter could be used.
A shade around the reciever could be used, to reduce input of surrouding ambient light, since you know approximately where the reflector is. A lens could be used to collect more light, far more. This would be a bit expensive and has some downsides.

I really like the idea that this could be couple dollar scale cheap. It could be used to calibrate 3d printers and other machines. I know there isn't much point in such a device for a printer because they are not that accurate anyway, however you get the idea. It helps open the door to highly cheap yet highly accurate devices, somehow, some day.

If the math and electronics and other details could be figured out, it would allow you to build a cnc machine that has these devices integrated into it. The solidity of the mechanical aspect of the machine is still needed, but it doesn't have to be straight and square and so on, because you can compensate in software for a lot of such lack of perfection, even as the machine wears. To be fair, you could do something similar today, just build a solid but inaccurate machine, calibrate it using a CMM machine or similar, and then just pair that data with the machine. This is actually done for some machines to some degree, but only part way and only on high end machines. I envision highly accurate machines cheaply available to all....
.
 

Thread Starter

adouglas88

Joined Jul 4, 2021
7
Another idea: suppose you know approximately what frequency the system should be oscillating at, because you start with a known optical path, say 30 cm. You could use a highly specific notch filter to filter out anything that is more than slightly outside that frequency. The frequency is then measured precisely. As the frequency changes, that is as the reflector moves, the filter could be modified so that the resonant frequency of the system remains the strongest signal. So, a moving filter that uses the information regarding the center of the resonant frequency of the system. This makes it so the reflector can only be moved so fast before the "lock" is lost, however that may be plenty fast enough for practical purposes.
 

nsaspook

Joined Aug 27, 2009
13,079
whoops, pressed enter and it submitted before I was done :

...quite a high degree accuracy.

If the math and electronics and other details could be figured out, it would allow you to build a cnc machine that has these devices integrated into it. The solidity of the mechanical aspect of the machine is still needed, but it doesn't have to be straight and square and so on, because you can compensate in software for a lot of such lack of perfection, even as the machine wears. To be fair, you could do something similar today, just build a solid but inaccurate machine, calibrate it using a CMM machine or similar, and then just pair that data with the machine. This is actually done for some machines to some degree, but only part way and only on high end machines. I envision highly accurate machines cheaply available to all....
.
A neat idea but IMO you will find the time-distance transfer function will be on the same level of complexity as TOF systems that use other indirect methods to measure time clock ticks between events.

For instance the CTMU in a microcontroller can have sub-nanosecond resolution of using a timed (discharge, start,stop,sample) constant current source and a fixed capacitance to provide a linear transfer function of voltage to time. The ADC that measures the sample voltage sets the possible resolution of the system.

Here you see a linear charge slope (time/amplitude) from the CTMU in action that varies as the value of capacitance is adjusted while time is held constant. If capacitance is held constant and the start/stop charging period is adjusted we get a voltage value on that slope that can be used to calculate a time value for TOF calculations.

I've used a system like this to measure FOT from a transmitter to receiver over light-links. The system is calibrated using several standard length fibers from low, middle, high range of the expected measurements.

https://forum.allaboutcircuits.com/threads/tdr-ctmu.179176/post-1631398
 
Last edited:

Beau Schwabe

Joined Nov 7, 2019
155
You will also need to use special mirrors if you are going to reflect the laser beam.... a standard mirror will create a double image from the glass itself and the reflective backing. You need a polished flat metal mirror.

Several years ago my wife worked at a church and we were great friends with the Pastor. He was a science geek and allowed me to use the facility for various electronic experiments after hours. One was just a proof of concept speed of light measurement. Where I "folded" the light beam from a HeNe tube laser about 10 times using hard drive platters for my mirrors using some of the long straight hallways of the church with a total distance reaching about 600 feet. The processor I was using was running at about 80Mhz (12.5ns per clock). The SAME processor controlled the Laser and waited for the receiver detection. Counting Clock cycles, the expected number was somewhere in the neighborhood of 48 ... 600 / 12.5 = 48 ... but the number I got was closer to 15 because of overhead ... ( time for the laser to come ON, and propagation delay from the detector ) So roughly 400 ns was spent in propagation delays. I expect most of the delay was from the laser.
 

Thread Starter

adouglas88

Joined Jul 4, 2021
7
A neat idea but IMO you will find the time-distance transfer function will be on the same level of complexity as TOF systems that use other indirect methods to measure time clock ticks between events.

For instance the CTMU in a microcontroller can have sub-nanosecond resolution of using a timed (discharge, start,stop,sample) constant current source and a fixed capacitance to provide a linear transfer function of voltage to time. The ADC that measures the sample voltage sets the possible resolution of the system.

Here you see a linear charge slope (time/amplitude) from the CTMU in action that varies as the value of capacitance is adjusted while time is held constant. If capacitance is held constant and the start/stop charging period is adjusted we get a voltage value on that slope that can be used to calculate a time value for TOF calculations.

I've used a system like this to measure FOT from a transmitter to receiver over light-links. The system is calibrated using several standard length fibers from low, middle, high range of the expected measurements.

https://forum.allaboutcircuits.com/threads/tdr-ctmu.179176/post-1631398
hm, I don't quite understand, except that yes the electronics might get fairly complicated for my idea. However, they are similar in many ways to a radio, which is very cheap. It's the precise measurement in changes of frequency.

Sounds like you are using a method to make a very high resolution timer with high accuracy. However even if you get 1 ns accuracy, overall, factoring in resolution, that's still 300,000,000 meters divided by a billion, 0.3 meters = 30 centimeters.accuracy. I guess some TOF systems are indeed capable of even better still as they do advertise better resolutions and accuracy even than that. However, I need *micron* level accuracy.

That would require a timer with impractically high resolution, yeah? However, if the frequency can be measured, it can be measured to within a PPM no problem. And thus, changes in the optical path of only a few microns could be reliably measured, hopefully.

In a way it is like time of flight, except that you take like a million sample, then calculate the average time of them all. A million samples greatly changes the equation.
 

Thread Starter

adouglas88

Joined Jul 4, 2021
7
You will also need to use special mirrors if you are going to reflect the laser beam.... a standard mirror will create a double image from the glass itself and the reflective backing. You need a polished flat metal mirror.

Several years ago my wife worked at a church and we were great friends with the Pastor. He was a science geek and allowed me to use the facility for various electronic experiments after hours. One was just a proof of concept speed of light measurement. Where I "folded" the light beam from a HeNe tube laser about 10 times using hard drive platters for my mirrors using some of the long straight hallways of the church with a total distance reaching about 600 feet. The processor I was using was running at about 80Mhz (12.5ns per clock). The SAME processor controlled the Laser and waited for the receiver detection. Counting Clock cycles, the expected number was somewhere in the neighborhood of 48 ... 600 / 12.5 = 48 ... but the number I got was closer to 15 because of overhead ... ( time for the laser to come ON, and propagation delay from the detector ) So roughly 400 ns was spent in propagation delays. I expect most of the delay was from the laser.
I was suggesting the use of a metal surface, no glass involved. A spherical metal ball, polished. I guess it should be stainless steel or the layer of oil on the surface might slightly complicate things and/or oxidization.
 

nsaspook

Joined Aug 27, 2009
13,079
That would require a timer with impractically high resolution, yeah? However, if the frequency can be measured, it can be measured to within a PPM no problem. And thus, changes in the optical path of only a few microns could be reliably measured, hopefully.
There is no timer in the CTMU TOF. The resolution limits are the switching ON/OFF speed of the constant current to charge capacitor switch and the resolution of the ADC system.

What's the expected frequency range for the optical path oscillator? What's the process to measured within the required PPM for a few microns resolution no problem if the result frequency is a several GHz oscillator.
 

MisterBill2

Joined Jan 23, 2018
18,167
nsa states the challenge clearly. The variation in time of flight will be the variation in cycle period of the oscillation, which defines the frequency of oscillation. This means that the measurement will need the same resolution and accuracy as a TOF system.
"Kaman" used a similar scheme for a precise displacement measuring system many years back and achieved a useful product. I suggest reading about that system.
 
Top