Update on laser distance meter project (master thesis) and question

Yannick · 2005-04-01T21:05:25+00:00

Maybe some remember i was developing a laser range finder for mymaster thesis. Now 2 months from the end i can say it is a succes. Ican measure distances up to 10meters from white objects. Theresolution is good the first few meters (1-2cm) and afterwards worseand worse (10-15cm for 10 meter). The problems lies on several things, i still use a dds frequency generator with 125Mhz clock , for thehigher frequenties (20-30Mhz) the sinus is far from perfect. Thesecond problem is ofcourse the photodiode preamplifier wich is stilldesigned around the opa657 in the most simple configuration. Thethirth bottle neck is that i use a phase detector ic from analogdevices (AD8302) wich does clipping on the signal before multiplying.Soo signals lower then noise cant be detected, this will be solved dueusing a DSP processor (ad's blackfin) and with signal averaging i canmeasure signals lower then noise (it will only take more time).There is still something strange with the photodiode preamplifier , itried many things including the R-C-R trick described by winfieldhill, although simulation in microcap gave me stunning results andhuge bandwidth improvement, in practice it didnt work, i the best casei had exact the same results and mostly worse. Then i thought ofsomething simpler, just using a resistor tree network to reduceresistor values and still have the same total feedback resistor (68Kin my case). This is described in photodiode amplifiers from geraldgreame and i did it exactly like he pointed out. i use for R1 = 22K ,R2=270 and R3 to mass = 100 soo i will have a total feedbackresistance of Rfeq=R1*(1+R2/R3) = 81.4K.In simulation this worked wonderfull,more gain, more bandwidth thenwith my singe 68K feedback resistance. But in practice the result wasworse. What can this be? everything is made in smd components with aopen ground plane to reduce capacitance to ground and everything isshielded in a RFI case.I want to thank Colin,Winfield hill and al the others for their helpin the...

C

colin 21 years ago

ah ok, yes theres more i need to explain, its quite clever ... obviously the two signals of 20 amd 20.4 cant be in phase however the signal you get from the mixer/detector is 400khz and this is what is now your signal so you need a reference signal of 400khz to compare it to. you get this from mixing the two original signals of 20 and 20.4 mhz in a standard mixer stage, as you increase the frequency you use say 30 and 30.4 mhz the result is still

400khz in both cases.

the clever part is that if you do the maths or just think about it you see that the phase change in the reflected high frequency signal is actualy directly contained in the 400khz signal. ie the actual delay in time is multiplied by the diference in frequency.

a mixer is a just a multiplier, the photomultiplier gain is determined by the bias voltage and so the output from the APD is the light signal multiplied by the bias voltage. its handy that it has a high slope at point of use so you dont need to superimpose a high ac voltage.

yes you definatly need to keep the bias voltage wel over 140 volt or the capacitance change is phenominal. and yes as you can easily change your ac voltage within your DDS it would be better, its dificult to design a analogue variabl gain stage that has a totaly constant phase, this has hindered me quite a bit, i found this solution worked reasonably well. i have found a few volts change in bias voltage cuases a very wide change in gain with minimal change in delay, dont forget the apd capacitance is very smal anyway, and at 400khz a slight change wil have minimal efect.

however to get an extremly wide agc range you may find your ac bias voltage needs to be a few uv youl find this cuases considerably problems. changing both at the same time is of course an option wich i would probaly try to use, you can also atempt to make erros cancel eachother out.

another option is to use an optical shuter such as a lcd

you need to adjust the bias voltage to some extent to acount for changes in temperature. and to set the optimum gain.

the 400khz signal you get should be close enough to a sinewave to give good acuracy, narowband filter wil of course take away any distortion. you can do the maths for mixer stages with non ideal multiplication charectersitcs, u just asume the input signals are distorted and fed into an ideal mixer, then u multiply al the distortion products too, but i doubt you wil find it cuases any problem.

yes drift is potential problem you have to assess the point at wich its aceptable. you can get ceramic IF filters used in old radios about 440khz although i found it surprisngly hard to get hold of one from any handy stockists.

i c now that you are thinking of using a digitiser and doing away with the detector altogether. i realy dont like the detector you are using its not designed as a detector at all. i didnt realy look at it before im surprised you geting such good results with it actualy. im wondering how big your lense is ??

of course at 400khz your digitiser is a whole lot easier.

good i always fancy writing a book but doubt itl ever hapen, its good to be able to at least write about it here. oh and please feel free to mention my name if you want (Colin P. Rowe)

i sugest you look up superhet radio operation if you havnt covered it already.

it al depends how easy it would be to add a second DDS to your project (and a simple mixer), hardly any other significant changes are required. this may be a whole lot easier than ading a digitiser anyway, as your curent phase measuering wil be adequate with a much cleaner signal.

can you get dual DDS ?

yes it is surprisingly chalenging isnt it, but thats what makes things so exciting. my latest project is a bit of a spin off trying to measure smal changes in speed of light to investigate the validity of the Silvertooth experiment wich suposedly contradicted the Michelson-Morely experiment. its probably more chalenging as i hope to make it simple enough for it to be easily reproduced so the results cvan be easily demonstrated. if it actualy works i make no prediction about the outcome of the experiment tho as this would be rather contreversial anyway (im realy not sure, wich is why i want to try it).

Colin.

Vote

Y

Yannick 21 years ago

Yes its a very exciting project, did you do it on your own or in cooperation with a company? What kind of resolution did you get and maximum distance ?

Indeed i use a 3mw CW laserdiode (655nm) and yes the photodiode preamplifier is the greatest limitation in my current prototype as is the phase detector. I will however change in a later state to a DSP solution like u did. But first i am going to try optoelectronic mixing like described by Colin.

Yannick

Vote

Y

Yannick 21 years ago

Yes ofcourse.

ok i got that part.

yes but this will always polarise the light in one way , soo you get inevitble 50% loss.But yes if you use it at the laser diode you could use a higher power (6mw)laser soo after the lcd it's still 3mw, and do the gain control with the laser power.

Yes i know, but this will be done in the final stages , are you already doing this?

its 80mm :) , you mean my phase detector the ad8302? what's soo bad about it , only the clipping is not good i thaught. Maybe a normal gilbert cell was far better and then just using the cos-1 function in my microcontroller to get the phase difference. I actually was thinking, if no clipping occurs you don't have to use a DSP to detect signals out of noise because if you multiply the received signal with the reference signal(with the mixer) and then doing ad conversion of the result wereafter you integrate the result in the microcontroller. Then this would be equal to fouriertransformation and this is actually a correlation soo if you do it long enough the signal has to rise out from the noise. I was always thinking it had to do this with a fase ADC and a DSP processor but actually you can do the correlation analog at first sight.

Ofcourse i will do, i will mention it at the references for the final chapter and in my thank word.

hmm never saw any, will look it up.

Wow, you are also interested in the whole thing about speed of light and relativity i see. I didnt hear about this silvertooth experiment however i am a bit sceptic because what reference clock will u use as if the speed of light would change (wich would be indeed very controversial , einstein would not be happy if this should be the case:) ) then the atoms in the reference clock will do the same and the clock will tick accordingly matching the difference out.

I always wondered why the speed of light is independed of the speed of the reference point. Once you take this as true the rest follows like time goes slower when you move due acceleration.It's very very interesting these things... I am curious how you will try to do this...

Yannick

Vote

C

colin 21 years ago

good point and idea, although this will be noticable as the laser is useful for tergeting. there are other devices wich dont polarise the light, some1 sugested here some time ago but i forget what it was now, but it wld be rather bulky.

yes i find that the agc alone sets the bias voltage, with no signal the bias voltage rises until the dark noise reaches the agc level, this is a reasonable method another, way would be to interupt the light path mechanicaly, or in your case you could digitaly adjust both the transmited signal and the reference signal amplitude.

thats a lot biger then mine. i stll used the 15mm.

yes the cliping is the bad part, if you use 2 gilbert cells and adjust one signal by 90' to each of them you get two signals out wich you can compare and get full 360' measurment otherwise the signal strength from 1 cell dominates the output voltage. (hence the cliping to make it simpler). using

2 gilbert cells in this way and analyizing the filtered then digitised reults in you mcu would get rid of most of the noise. or as someone sugested in this thread, just using one cell, rotate the phase of your reference signal through 360' and note the null.

hmm maybe shld start another thread, but i dont actualy measure the speed of light absolutly i just measure any tiny diference between two oposite directions. i intend to use 2 matched clocks initialy synchronized (atomic ones would be nice but im gona try make do with state of the art crystal ones) 1 meter apart then receive the signal in the middle. then il turn the aparatus thru 180' and note any phase diference. the clocks wil always be moving at the same velocity relative to eachother so should not experience any diference between them.

for simplicity leds will indicate positive and negative diference and so you wil be able to see the orientaion that has maximum diference as they light up as it all rotates.

of course the efect of moving even at 100,000kmh has a small efect on the speed of light, with 2 1ghz clock 1 meter apart i need to resolve a few uv from the phase detector and of course the clocks have to behave like perfect clocks for the period of rotation. hopefuly any jitter wil average out over the the period of rotation. i wil probably put the clocks in a phase locked loop but with a response many times slower than that of rotation. i might have to move the clocks further apart but would be hard to rotate them, of course the earth is rotating so i could put them several thousand miles apart but that would cuase problems, not least is the clocks would not stay synchronised for 24 hrs.

the Sagnac efect makes use of the diference in speed of light in a circular fibre to detect rotation movement and is used in aviation guidance, but its harder to do it linearly. the explanatin Silvertooth gave why his experimant gave a result but previous ones didnt is beyond me to see why it aplies to one but not the other. I think the important thing to do is not to try measure the speed averaged over the forward and reverse directions as this any change wil be imeasurably small, maybe this is the diference.

Another way to syncrhronise the two clocks would be a very high speed rotating shaft with encoders on the ends. but the problems here are far more mechanicaly orienteted. I have heard about a rusian scientist used a mothd involving a long rotating shaft with slited discs to determine the speed of light and he published complete with unexplained variations in the speed of light, but this remained hiden for almost 200 years.

again i like to point out i make no prediction about the outcome although i suspect my design wil need some improvement before it is sensitive enough.

Colin =^.^=

Vote

L

LRCR 21 years ago

In both cases. On my own for my Master Thesis and for my former employer, Hughes Danbury Optical Sysytems, CT.

My Thesis used a coarsed resolution Laser rangefinder:

Resolution: 3.44ft (due to the DSP sampling rate) Max distance: About 44ft., (Laser Power and optical telescpoe Limited) Laser : CW Diode at 826nm (100mW) Collimated (Melles Giot) telescope: Refractor (Gold plated mirrored telescope would have been better, but can't afford it). Mixer: made by minicircuit for the heterdyne part of my design local oscillator: 1.64MHz Output from the mixer is 30kHz APD detector used

From my work:

Resolution: 0.5ft Max. Distance: 1,000 ft. Laser: 1.5Watt CW Diode laser (Very Expensive) Collimated (Melles Giot) Telescope: Gold plated mirrors Cassegrain telescope Also an heterdyne Mixer design. Local oscillator much higher (can't remember now) than the Thesis version in order to achieve the resolution it has. APD detector used

The great thing about this project is that you learn about different areas in engineering(Analog and Digital Electronics, Optics and Optical Geometry, Power, Laser, Detector, filters, Rf, DSP, HV for the APD, etc.). Also, cuting and drilling metal parts (thanks to what tools can find in Sears and Al metals from my local HW store) was fun.

This was a fun project and I believe it helped in my engineering carrer both directly and indirectly..

Vote

Y

Yannick 21 years ago

hmm thats interesting, i didnt know it existed.

that's what i inted to do but my dds doesnt allow me to adjust the output amplitude digital. I am going to use the ad603 variable gain amplifier after my dds.

yes but you are using the optoelectronical mixing soo you dont need such a big signal at the input like me. what were your results without the mixing?

that's a good idea if i understand it correctly u would use 2 gilbert cells because the detected signal is lower in amplitude as the reference signal and soo to get the normalised 0.5cos(dphase) output you compare the output of two gilbert cells wich are in quadrature.

i am goig to read it carefull but not much time right now, it sounds interesting...

Yannick

Vote