How does laser meters work?

[jmich]

Hi,

There are several brands of laser meters, like the ones from Bosch and Fluke, and thye have some quite impressive specs with accuracies down to 1mm. I was wondering how they achieve this.
I know that some of the cheaper ones are actually just ultrasonic with a laser pointer, but I'm talking about the 'true' laser devices.

If they use time of flight, light travels 1mm in around 0.3 picosecond and it has to go both back and forth, so let's say the accuracy is 1.5mm which takes 1 ps total for the round trip. To measure that they would need a timer running at 1000 GHz which I find very unlikely. Try to find a crystal at that speed and what about counters running reliably at that speed?

Using phase shift seems just as unlikely since any laser wavelength is so short that it would shift phase millions of times even during a short measurement distance. For instance a laser at 1000 nm would shift phase 1 million times over a one meter distance and these devices can obviously measure a lot more than that.

Maybe I'm totally missing something, but I just can't figure out how they work, so I hope some of you have a good idea 

[IanB]

I'm not sure how they work exactly, but I imagine they use some kind of interference effect. If you split the beam and send one half to the target and back while keeping the other half local then they will be out of phase when you recombine them and you can detect this. Supposing you didn't directly want to use the light frequency itself, you could modulate the beam with a lower frequency signal and use this in the detector, then you wouldn't have to split the beam optically. By comparing the local reference oscillator with the recovered signal from the returning beam you could deduce the time delay.

[grumpydoc]

Most use triangulation, not timing, AIUI

[mikeselectricstuff]

They definitely use time-of-flight. Not sure of the exact details, but they typically use a very large number of samples (many thousands), so various sampling techniques are possible to reduce noise and improve resolution  If you look at specs on industrial configurable TOF distance sensors, they tend to have a selectable tradeoff between reading rate and accuracy. 
There are also analogue techniques to measure very small differences in pulse delay that avoid the need for ridiculous clock speeds. For example imagine you have a capacitor, start charging it when you send a  pulse and stop when you receive it. This gives an analogue voltage proportional to the pulse delay.   

[ve7xen]

I think they probably do use time-of-flight and some cleverness.

A one-shot interpolating counter can fairly easily resolve 100ps or so (see the PICTIC project) with cheap components. The basic technique is to combine a traditional counter with mid-cycle interpolation. Basically, between cycles a constant current charges an integrator, which is gated by the input signal just like the 'main' counter. NIST has a good paper describing the technique - from 1994 - http://tf.nist.gov/general/pdf/870.pdf

Combine that with integrating a large number of measurements and I think you can get down to the necessary resolution relatively 'easily' (for definitions of 'easy' that include lots of R&D, but cheap components).

Let's find out! Time for a teardown .

[pickle9000]

Good question.

[smashedProton]

How do they receive the reflection of the beam?  Light isn't as easy as microwave to bounce off of stuff from what I know.  Pretty interesting.

[grumpydoc]

I've been googling around. Not exactly definitive, I know, but some useful information.

There seem to be at least two types - the ones marketed to golfers, hunters etc which measure out to 500m and seem typically to have 1m accuracy.

Time of flight for those would require a 1.6us interval measured to 3ns precision which seems very achievable.

Then there are the DIY type meters such as the Bosch/Fluke ones mentioned in the original question. These measure out to 40-50m and claim roughly 1mm accuracy.

Time of flight for these would require a 160ns interval measured to 3ps resolution which seems an entertaining problem to solve!

There are quite a few references to trig based finders including homebrew ones such as this example, these look as though they would easily give the 1mm accuracy.

So it looks as though different techniques are used depending on the range/accuracy required.

[G7PSK]

I have tried lase measures and do not rate them for more than an indication of length or size, they are ok for carpet salesmen to make an estimate (a lot of carpet is wasted in fitting) and for real estate sales men, but for engineering you need the old steel ruler every time as depending on the reflective surface the measurement can be out by far more than the manufactures spec's.
The first time I saw this was with a fire escape which when it came to fitting it the landing was a meter to one side and 2 metres low, any one rushing out of the door would have gone straight through the local hospitals A&E (emergency room for US) roof. Under ideal conditions they tend to be good for only + - 1 CM at best so although I have one for estimates I always measure up for a job with a steel rule, steel rule's OK.

[kripton2035]

I use a bosch every often and it is really 1mm accuracy.
so if you need more than that accuracy yes you have to use a steel rule !
but for wood working it's really ok with them.
I measure distances in rooms, to make furnitures come right into a place and with 1mm accuraccy it's really ok
to measure 3m with a steel ruler is not very easy.

here a tutorial how laser range finder works :
http://www.eng.buffalo.edu/ubr/ff03laser.php
and also another one here :
http://sites.google.com/site/todddanko/home/webcam_laser_ranger

[mikeselectricstuff]

Then there are the DIY type meters such as the Bosch/Fluke ones mentioned in the original question. These measure out to 40-50m and claim roughly 1mm accuracy.

The Bosch definitely uses TOF, but is doing some odd stuff. Had a probe around one - uses a variable high frequency modulation - maybe using phase shift at various frequencies to determine time.  There is also a mechanical thing that can direct some of the TX light back to the receiver internally - I presume this is for a zero reference.   Uses a custom chip for the tricky stuff, I think from a company that does a lot of RF chips.  I imagine the front end is not unlike an RF receiver.
I was wondering if it might be feasible to do it by making an  an oscillator using 1GHz RF front end chips, with the light path in the feedback loop, to get a frequency to distance function.

[amyk]

It's relatively easy to generate signal delays in the ps range; most of the time that's not something wanted in circuits and called skew, caused by small differences in length. You can have two oscillators tuned to frequencies whose periods are separated by e.g. 1ps, and connect them to separate counters. Start both counters when the light exits, and stop them when the light returns. The difference between the two is the desired measurement.

Of course calibration is essential, but the absolute precision required isn't that high; the period of 1GHz and 999MHz differ by ~1ps already, and a frequency error of 20ppm, easily achievable, correponds to ~20fs (0.02ps) error in the period or ~6µm of distance error.

[jmich]

Hi guys, thanks for all the replies with lots of good suggestions.

I think ve7xen's link (http://tf.nist.gov/general/pdf/870.pdf) which suggest Time Interval Interpolation as the method sounds as the most realistic bid.
Edit: forgot to add a link to a similar PDF http://ztc.wel.wat.edu.pl/met4_1_004.pdf which describes the method in more details and is more recent (2003).

I know that sub ps delays can be achieved with delay lines, but the problem is that it must be combined with very long measurements, so combining a coarse MHz counter with TII sounds like it would do the job. It's still not something for the average DIY project, but should be doable for the companies that produce those devices.

I don't believe in triangulation at >100m (330ft) like some have suggested. It's just too little deviation on such a small sensor and would require a pretty big resolution.

The post from mikeselectricstuff sounds interesting too - there might be some "odd stuff" going on to further sharpen the accuracy.

One thing that also puzzles me regardless of the timing method is, as smashedProton wrote, how they manage to capture the reflection on those distances while still keeping the device eye safe?
It would definitely need to be pulsed, but still it's incredible to capture a reflection of a eye safe laser at 20m, 100m and even 500m as some of the golfing/hunting devices offer.

I can only say one thing:

TEARDOWN PLEASE 

[grumpydoc]

I don't believe in triangulation at >100m (330ft) like some have suggested.

I think I'm the only one to mention triangulation but I haven't suggested it is used for the >100m rangefinders - in fact I think I specifically said that thought that the timing requirements for time-of-flight for 1m accuracy at 500m didn't sound too bad at all. Actually I forgot to multiply things by two for flight and return so it's even more achievable - a 300MHz clock should give you about 50cm accuracy.

you can have two oscillators tuned to frequencies whose periods are separated by e.g. 1ps

That sounds like an interesting technique - how do the numbers stack up?

A target at 10m (20m flight path) is 66.67ns delay so we'll count 66.67 full cycles of a 1GHz clock and 66.63 cycles of a 999Mhz clock - move the target to 10.01m and it's 66.73ns so 66.73 cycles vs 66.66 - as integers this all looks like 66 so I'm not clear how this helps.

Hmmm - possible light going on - if you make one fixed and one variable can you get any information by trying the counts with slightly different frequencies until the counts do differ by 1?

[jahonen]

I'd guess that those use similar techniques than in a FMCW radar:

http://en.wikipedia.org/wiki/Continuous-wave_radar

Regards,
Janne

[NiHaoMike]

A cheap IR ruler I took apart used a high pulse power IR LED or laser diode (metal can package, but little heatsinking) and a small linear array of photodiodes. It had a range of about 25 feet and could measure down to 2 feet with 1 inch resolution. I'm not exactly sure how the measurement circuit works since most of it was in a custom ASIC, but the analog circuits were nothing special (just some generic opamps and passives), so it most likely used parallax.

[jmich]

I don't believe in triangulation at >100m (330ft) like some have suggested.

I think I'm the only one to mention triangulation ...

Triangulation is used in these links too:

here a tutorial how laser range finder works :
http://www.eng.buffalo.edu/ubr/ff03laser.php
and also another one here :
http://sites.google.com/site/todddanko/home/webcam_laser_ranger

you can have two oscillators tuned to frequencies whose periods are separated by e.g. 1ps

Might have been possible IF such oscillators did exist. Guess it would have to be custom made.
One of the problems with GHz oscillators in general is the price - another is all the noise they make 
And what about GHz rated counters?

One of the advantages of the Time Interval Interpolation is that the oscillators and counters don't have to be very fast.

I'd guess that those use similar techniques than in a FMCW radar:

http://en.wikipedia.org/wiki/Continuous-wave_radar

I didn't read it all, but generally radar is used either to measure speed using Doppler effect or to measure distance using frequency modulation which is not possible with lasers.
Please correct me if I've missed something in the article.

[4to20Milliamps]

Lidar

https://en.wikipedia.org/wiki/LIDAR

[awallin]

High resolution measurements are interferometric. There's a description of the metrology lasers in Sam's laser FAQ:
http://www.repairfaq.org/sam/laserlia.htm#liaint2f
and an image:
http://www.repairfaq.org/sam/inter2f.gif

If I understand correctly these techniques rely on the HeNe laser gain-curve being constant (i.e. does not move/drift in frequency/wavelength), and stabilize two laser-modes wrt.  the gain-curve. That way the (mean) wavelength, and thus the measurement result, is known with >7 digits precision.
HeNe lasers are old technology, the require a dangerous/bulky HV supply, and the efficiency is lousy (the tube gets hot!)

It would be an interesting exercise to build a similar system based on a diode laser (cheap, small, efficient). One would need to stabilize the laser to an atomic transition or use some other tricks to know the wavelength with >7 digits. The optics and electronics are pretty straightforward once the wavelength-stability problem is solved.
This would be useful for calibrating e.g. CNC-machines, 3D printers, etc. that have moving stages.

cheers,
Anders W

[Gall]

The idea is to measure the phase shift between the modulated outgoing and incoming signal. A small phase shift results in large change in amplitude. This may be done entirely on optical side and thus works fast enough. It is theoretically possible to achieve 10nm precision if unmodulated light is used.

Optics may be used for signal processing on its own and it works much faster than Si electronics

[CarlG]

There are two basic principles used for electronic distance meters (EDM): phase shift and time-of-flight (TOF) as described in this Trimble White paper. High resolution measurements can be done with TOF as well as with phase shift/interferometric methods.

The phase shift principle is described e.g. in this HP Journal article.

The TOF principle is pretty well described in Ch 2 and 3 of this Pasi Palojärvi paper.

However, I think the main question was regarding the time measurement principle, which was answered in links in previous posts (here and here).

As you can see, GHz oscillators are not needed to acheive mm resolution.

[jmich]

Thanks CarlG, those links are really useful.
I think that both methods are likely candidates for the commercial devices. Looks like Trimble offers both.

But it still puzzles me how they manage to detect a reflected class 1 laser at those distances... In the Trimble paper they are talking about distances over 800m and even up to 5000m when using reflecting prisms.

[NiHaoMike]

Thanks CarlG, those links are really useful.
I think that both methods are likely candidates for the commercial devices. Looks like Trimble offers both.

But it still puzzles me how they manage to detect a reflected class 1 laser at those distances... In the Trimble paper they are talking about distances over 800m and even up to 5000m when using reflecting prisms.

They pulse the laser (very low average power) and the receive chain is AC coupled. Ambient light is ignored and by using multiple pulses, the noise is averaged out.

[CarlG]

Thanks CarlG, those links are really useful.
I think that both methods are likely candidates for the commercial devices. Looks like Trimble offers both.

But it still puzzles me how they manage to detect a reflected class 1 laser at those distances... In the Trimble paper they are talking about distances over 800m and even up to 5000m when using reflecting prisms.

They pulse the laser (very low average power) and the receive chain is AC coupled. Ambient light is ignored and by using multiple pulses, the noise is averaged out.

Somewhat right. Yes, the pulse peak power is of course much higher than the Class 1 limit of 1mW (average). But, the receive chain does not have to be AC-coupled; the incoming light to the receiving device (typically a PIN diode or APD (avalanche photo diode) must be dimmed so that (ideally) only the reflected pulse is sensed by the receiving device. Averaging is done after detection, so each pulse must still have sufficient energy to be detected (of course). Max distance is (mainly) determined by emitted peak power, beam divergence, target refllectivity, aperture area, and the receiver sensitivity. And of course, the stated distance is under ideal conditions