3

I'm interested in building my own laser doppler vibrometer using laser diodes. The reason I mention laser diodes is because (1) I want it to be cheap and (2) I want to incorporate it into robotics projects. However, I'm unsure if it's possible to build something that satisfies my requirements for reasonably cheap (say, the components of such an LDV would cost < 500 USD). Specifically, I am trying to measure vibrations that have a frequency of 1 MHz – I am not trying to build a general-purpose measurement device, as most commercial LDV devices are. I was thinking of using self-mixing laser interferometry with a laser diode, but the research I've read seems to indicate that such self-mixing laser diode systems don't get anywhere close to 1 MHz. Is it possible to build such a reasonably cheap device? And, if so, how? I would greatly appreciate any references to research papers.

EDIT: I also read this Reddit thread from 8 years ago, where the OP seems to be trying to build their own LDV.

EDIT2: The distance the LDV would have to operate is very short (say, 1 meter or less).

The Pointer
  • 1,241
  • 1
  • 11
  • 24
  • 1
    What lasers do you have at hand, in terms is line width? You say "reasonably cheap", but that sadly doesn't tell me whether you mean 50€ or 50000€ with that term... "Is it possible for me" is also sadly something we can't answer, as we don't know your equipment nor skill set! – mmmm May 27 '21 at 08:18
  • @mmmm Let's say < 350€. I'm not so much concerned about skill required, as that can be learned. – The Pointer May 27 '21 at 08:20
  • 1
    Edit your question, please. Also, this really seems unreasonably cheap, assuming you asking things this way implies you don't have a working interferometry setup lying around already. What *do* you have lying around already that helps with the task? I.e. of the Wikipedia diagram, which components are you missing? – mmmm May 27 '21 at 08:21
  • @mmmm I have edited my post. I want to build a version of this from scratch that can be incorporated into robotics projects (which is why I wanted to use laser diodes). – The Pointer May 27 '21 at 08:25
  • @mmmm I assume the required optical elements are relatively expensive? In sum it would probably be a four digit number but could be doable below five digits? Ah, this self mixing device looks interesting.. but is that doable with a laser diode?? – Ben May 27 '21 at 08:31
  • @Ben Are you sure about that? I'm not trying to build a commercial general-purpose LDV – just one that satisfies my specific requirements. – The Pointer May 27 '21 at 08:48
  • @ThePointer Sure about what exactly? The vibrometer will probably rather expensive, the interferometer might be promising. But as mentioned, I'm sceptic whether a laser diode is sufficient for that. Can't imagine. – Ben May 27 '21 at 10:32
  • @Ben Well interferometers and LDVs are the same thing, right? – The Pointer May 27 '21 at 10:33
  • @ThePointer I was refering to the linked articles. The first one is probably much more expensive. I did some optics projects but can hardly remember the involved costs. I think a "simple" lense was already some hundreds.. – Ben May 27 '21 at 11:50
  • Applied Science looking at vibrometry using laser diodes: https://www.youtube.com/watch?v=MUdro-6u2Zg – user1850479 May 27 '21 at 14:06
  • Do you actually need 1 MHz? That is too fast to be easy to implement by moving a mirror and too slow for a lot of common modulators, so you would need two phase locked modulators (e.g. 60 and 61 MHz). – user1850479 May 27 '21 at 14:14
  • Sorry, I thought you meant modulate at that frequency since 1 MHz is uncommonly fast for vibration. In that case you can use a ~40 MHz AOM (cheap) and at least a 41 MHz photodiode (also cheap). The self mixing idea is interesting, but as you say it's unclear how well it would work since you are depending on what the laser diode can do. – user1850479 May 27 '21 at 15:02
  • @user1850479 so how would this work? Say I have an object that has vibrations at a frequency of 1 MHz on my lab bench. How does this setup work? – The Pointer May 27 '21 at 15:06
  • It would work the same as any other laser doppler system. You would build the system in the first figure of your wikipedia link. – user1850479 May 27 '21 at 15:17

2 Answers2

1

Maybe consider the principle that atomic force microscopes use.

They reflect the beam (very weak diode laser) off a vibrating object (in that case: the AFM cantilever) onto a quadrant photodetector.

After that you can condition the 1 MHz analog signal using standard electronics and ADC.

You don't necessarily need a quadrant photodiode, a differential one would also work. Or even a single one to some extent. The main advantage is cost, as you don't need an optical table and precision mounts. You also don't need to preserve phase of light as you do for interferometry. Therefore, you also don't need coherent light. So a well collimated LED will do.

tobalt
  • 18,646
  • 16
  • 73
  • Thanks for the answer. What makes you think that the way it is done for atomic force microscopy is relatively cheap and will work for 1 MHz? (I don't know anything about AFM.) – The Pointer May 27 '21 at 11:04
  • I suppose it will be cheaper because you don't need ultra-precise alignment as you will for interferometry. I don't know exact cantilever frequencies of AFM from the top of my head but i recall it is already in higher kHz range (mainly due to cantilever mech. resonance). Photodiodes and electronics work at much higher frequencies, so I see no reason why this won't work at 1 MHz. – tobalt May 27 '21 at 11:07
  • 1
    From what I've read, it seems that AFM works by using a photodiode to detect changes in light reflected off a reflective cantilever scanned over the sample in question. So, in other words, it requires a reflective surface, right? The surfaces that I will be working with will not necessarily be reflective. And one of the reasons LDV was so attractive was because it is non-contact, but this is lost in AFM due to the required contact (or close contact) of the cantilever. – The Pointer May 27 '21 at 11:19
  • Yes the AFM cantilever is polished on its back to be well reflective. To adopt this to dull object a mirror sticker would need to be attached. For interferometry to work well, your object also should be flat to within fractions of the wavelength over the spot size of the laser. But maybe LDV is different in this regard. I am not familiar with it – tobalt May 27 '21 at 11:25
  • @tobalt An AFM for ~500 bucks? – Ben May 27 '21 at 11:55
  • @Ben no a LED, collimator, quadrant photodetector, opamp, ADC for 500 €. No need for all the damped stage, precision steppers, software. But anyway, as the OP cannot attach a reflective sticker on the DUT, it won't work. – tobalt May 27 '21 at 12:14
  • @ThePointer Reflection can give you more power, but may be very difficult to align on a robot, whereas scattering is less power per angle, but emits at a range of angles and so much easier to align your detector. Probably you would be interfering scattered light rather than reflected. – user1850479 May 27 '21 at 14:02
  • @user1850479 so what are you suggesting? – The Pointer May 27 '21 at 14:11
  • @ThePointer That you don't need a reflective surface. – user1850479 May 27 '21 at 14:17
  • @user1850479 hmm interesting. But, naively, it seems like that would be even more difficult to use for measuring acoustic vibrations, no? How would this work? – The Pointer May 27 '21 at 14:23
  • with a scattering surface a tiny movement of the object will not produce a marked change in reflected angular power distribution, so no signal. At least not in the AFM kind of detection scheme. – tobalt May 27 '21 at 14:48
  • @tobalt I once built an interferometer that measured the velocity of red blood cells in living people, and we definitely didn't polish their blood to wavelength flatness. An interferometer doesn't know if light coming in is a plane wave because it was reflected or because you have a narrow aperture and a lens. – user1850479 May 27 '21 at 14:56
  • @ThePointer Sounds like you should edit the question to describe what you're trying to do. – user1850479 May 27 '21 at 15:23
1

You are trying to build a 1 MHz center frequency vibrometer of unknown bandwidth with a range of 1 meter to an unknown target that will be mounted on a robot. You have a budget for materials of $500. I'm going to ignore the last part in the comments about using it to measure ultrasound in people since there isn't enough information to give a useful answer.

This seems entirely possible if you don't have to pay for your time. You can buy or make all of the components for a lot less than $500, except possibly for the computer that is going to process this data. You can buy a photodiode for a few dollars and make a transimpedance amplifier for a few dollars more. You can buy surplus modulators for tens of dollars and build your own RF drivers. At the coherence length you need (~2 m), laser diodes are a few dollars and drivers can be made or bought for relatively little. Since you don't care about the exact splitting ratio, you can use ordinary plate glass (tens of cents) for the beam splitters rather than precisely calibrated dilectric or silver beam splitters. You can machine or 3D print the optomechanics. You can build the A/D converter for tens of dollars and you could implement the mixing digitally in python which is free. That is probably less than one hundred dollars, so you have plenty left over to buy power supplies, optical filters, aluminum mirrors and lenses if you need them.

So yes possible, but you will probably spend orders of magnitude more on your time than you save by not buying commercial products.

user1850479
  • 14,842
  • 1
  • 21
  • 43
  • Thanks for the answer. So would I be looking for a 40 MHz AOM and a 41 MHz photodiode? I'm not sure where to get the 41 MHz photodiode from, since 41 MHz seems to be a very specific number for it to be manufactured at. – The Pointer May 27 '21 at 20:24
  • Those numbers are just examples, but ~40 MHz is a relatively low speed so many typical photodiodes would work. If you find a different modulator, change those numbers. – user1850479 May 27 '21 at 20:35
  • So, for my application, the absolute numbers don't matter, just the relative numbers? Also, are there any cheaper options for modulators than acousto-optic modulators? – The Pointer May 27 '21 at 20:44
  • @ThePointer I suggested 40 MHz because it would be cheap. You can use other numbers if you prefer or find a good deal on hardware. – user1850479 May 27 '21 at 20:56
  • Hm, for your Doppler detection to not get buried in laser phase noise, your laser's line width would need to be narrower than 1 MHz, which sounds like distributed feedback lasers. How do you make that with a 500$ budget? – mmmm May 27 '21 at 23:21
  • 1
    @mmmm This is an interferometer, so the line width of the laser is irrelevant so long as the coherence length is longer then the optical path length mismatch between the arms. Reason is any phase on one arm will be subtracted from the other within one coherence length. – user1850479 May 27 '21 at 23:31
  • Ah good point! In that case, wonder whether a simple SFP fiber optic transceiver (which are really just dumb laser diodes and IM-DD) would work as relatively easy to interface source and detector – mmmm May 27 '21 at 23:35
  • By the way, small typo in your last paragraph: then vs than – mmmm May 27 '21 at 23:38
  • Hmm, why was this downvoted? Hi @mmmm , do you have any additional advice to add ? What is this "SFP fiber optic transceiver" you mentioned? – The Pointer May 28 '21 at 01:42
  • @thePointer the cheap optical networking transceiver modules, used for 1 Gbit/s optical Ethernet. Can be plugged into something like https://osmocom.org/projects/misc-hardware/wiki/Sfp-breakout . – mmmm May 28 '21 at 07:04
  • @mmmm Ahh, thanks for clarifying. Do you have any ideas for how this would work? I've been doing some reading on the device in question, but it isn't clear to me how it would be used for this purpose. – The Pointer May 29 '21 at 14:56