Pinhole lens and a photodiode

[htrx93]

I was wondering if anyone had any recommendations on whether using a pinhole lens with a photodiode is feasible for picking up a modulated LED a few meters away. My background is mostly electrical, so when it comes to optics I get a bit lost. The general idea I had was to use laser cut holes on a thin piece of metal over a bunch of photodiodes. Having 100 surface mount photodiodes is feasible for me to make... but 100 induvidual lens assemblies is not, so I wanted to explore other options.
Thank you.

[Toxible]

Hi htrx93.
If i understand, you want receive a led light with a photodiode at a distance of few meter.
We need to know if you want a narrow angle for the receiving or you are looking for an "isotropic" receiver.
why use a pinole lens? is there any particular reason?

[FTL]

Again, I'm not sure of exactly what you want to accomplish, but using a pinhole to achieve a narrow field of view will only work if you have lots of signal strength or lots of gain.

The light projected by a lens or a pinhole is defined by its f-number. That is the distance to the image divided by the diameter of the aperture. With a pinhole the f-number can be very large. For instance, a 0.5mm pinhole (which is pretty big for a pinhole), projecting an image 50mm away, will have an f-number of 50/.5 = 100.

A lens has an f-number defined by its focal Lent when focused at infinity. So a 10mm lens with a 50mm focal length will have an f-number of 50/10 = 5. This will project 20 times the light on the subject. You would need 20x more input light, or an additional 20x gain to achieve the same results with my example lens and pinhole.

[htrx93]

Yes, I want the angle the photodiode receives a signal to be pretty narrow and still get a good signal. Both the receiver and sender would be fixed in place. In general I'm looking to build a "light-gate" with a large enough resolution to image the outline of objects that pass through it. An object in the gate will block paths from receivers to senders. So limiting the angle the photodiode can see would allow me to place the photodiodes closer together.

The reason I thought of pinhole lenses is because I would rather have a single piece of metal with 100 laser cut pinholes, instead of 100 individual glass or plastic lenses for each photodiode.

Thanks for the crash course on the F-numbers... it seems I will be needing some very hefty amplification.

[Gallymimu]

Not an optics expert but this might not work very well unless you have the senders well columnated.  You should be able to solve this with simple ray tracing though.  Consider the dispersion angle of most LEDs.  I don't know that you will find less than 15 or 30 degrees.  That coupled with the distance it needs to travel (you should share that information so we can help you better) you can figure out if the light from an adjacent diode will hit each sensor.

you might want to read a little about pinholes if you haven't already:
http://en.wikipedia.org/wiki/Pinhole_camera
http://www.huecandela.com/hue-x/pin-pdf/Prober-%20Wellman.pdf

Have you considered a fresnel lens array?  That might do a better job and there are companies that make them.  I don't know how readily available they are though.

[pickit2]

why not use a digital camera module, with image software.
there is a lot of such projects to be found, with google.

[solutions]

http://www.youtube.com/watch?v=VTmEtzseuaA

If you can't control the light sources, as in my Youtube hint, above, just put the equivalent of drinking straws over each diode.

I'm with pickit2 in using a CCD camera, though.

[Toxible]

I'm with solutions and pickit2 for camera, but in this case you need a camera, and a pc. I don't know if htrx93 want build a such complex system.
You can use a IR led for transmitting the signal and a.... IR Led (yes a led) to receive. I've done this for one project were was required a narrow beam. For example HIRL5010 is an 850nm emitter with 6° of aperture angle (half intensity). With the proper schematic (if you want i can post a solution) you can receive the signal correctly and do your work

[solutions]

Don't ask "if you want"...post if you have something that works. This thread may prove useful to someone a year down the road

[htrx93]

Fresnel lenses are a very interesting idea I hadn't considered. My concern with the CCD camera would be the framerate, but maybe it's cheaper to get a CCD that can handle a higher framerate than to build some sort of lens array. As toxible mentioned I am a little concerned about adding the complexity of a PC.

That sounds very interesting Toxible! You've just blown my mind here. So correct me if I'm wrong, but reverse biasing will charge the junction and then if you time the discharge it should be possible to estimate the light. So I could use probably use a digital IO pin and wait until it crosses into digital low. It looks like the HIRL5010 has a really low junction capacitance, so it could be possible to do this very quickly. I would be very interested in your solution. In a real product this means the receiver and transmitter could be almost identical to reduce manufacturing costs as well...

[optikon]

Fresnel lenses are a very interesting idea I hadn't considered. My concern with the CCD camera would be the framerate, but maybe it's cheaper to get a CCD that can handle a higher framerate than to build some sort of lens array. As toxible mentioned I am a little concerned about adding the complexity of a PC.

That sounds very interesting Toxible! You've just blown my mind here. So correct me if I'm wrong, but reverse biasing will charge the junction and then if you time the discharge it should be possible to estimate the light. So I could use probably use a digital IO pin and wait until it crosses into digital low. It looks like the HIRL5010 has a really low junction capacitance, so it could be possible to do this very quickly. I would be very interested in your solution. In a real product this means the receiver and transmitter could be almost identical to reduce manufacturing costs as well...

umm, if you use an IR LED as the reciever, yes it can work but it is a TERRIBLE receiver so you will need a higher sensitivity amplifier to get your signal to a usable level... there goes your cost savings..

[solutions]

I reread the original posting. Are you trying to build an encoder? Knowing the app could take us down a completely different road. Tell us the app and constraints/approach and I bet we could come up with a much better solution than merely guessing.

@htrx93: FWIW, you're not going to attach a photodiode to a digital I/o pin and see it wiggle. You'll be waiting until sundown to see it go low....

[htrx93]

The original application I had in mind was to put a sender and receiver on either side of a highway lane and use it to:
1. count cars
2. image the profiles of the cars and use it to guess the vehicle type
3. do all of this at high speeds (120+ km/h)

I had two thoughts on how to do this:
1. Try and use as narrow beams as possible (which is why I was asking about pinhole lenses) and create a 1D profile, which over time would become a 2d profile.
2. Use wider beams with structured light. If I can transmit a structured light pattern I can evaluate the transmittance of the path between multiple senders and receivers and build a rough 2d section of the vehicle passing through and with multiple cross sections over time build a rough 3d scan of the vehicle.

I hope that clarifies

[sphinx]

here in sweden i have seen a solution where they use a hose of some sort as a pressure sensor to calculate
the amount of axles. this might also be a way to do it.

[Ichan]

The 1D scanning by 100 photodiode is interresting, what is the distance between the photodiodes?

Things comes to my head:
1. Narrow viewing angle photodiode with dome lens
2. Thick plate of the pin holes
3. Invisible laser diode as light source
4. Wonder how critical the timing to capture an image of F1 car on it's top gear 

Hmm, i guess 2 cm distance between photodiodes is viable, 100 of it will make 2 meter of scan head height - still can not capture an image of a big truck...

-ichan

[Gallymimu]

I don't think you'll get this to work with LEDs and photodiodes unless you can columnate the light, and then you will have alignment problems.  I'd vote for cheap CMOS camera.

Are you planning to modulate the LED output to reduce stray light interference?  Are you planning on any narrow band (optical) filtering?  You can pretty easily saturate your sensors if the sensitivity needs to be high with daylight.

[optikon]

The original application I had in mind was to put a sender and receiver on either side of a highway lane and use it to:
1. count cars
2. image the profiles of the cars and use it to guess the vehicle type
3. do all of this at high speeds (120+ km/h)

I had two thoughts on how to do this:
1. Try and use as narrow beams as possible (which is why I was asking about pinhole lenses) and create a 1D profile, which over time would become a 2d profile.
2. Use wider beams with structured light. If I can transmit a structured light pattern I can evaluate the transmittance of the path between multiple senders and receivers and build a rough 2d section of the vehicle passing through and with multiple cross sections over time build a rough 3d scan of the vehicle.

I hope that clarifies

LED's & Pin hole photodiodes, across a highway and you expect to build a profile of the car even while its dirty & raining, in hopes to build 2D or 3D images to determine make/model. How many years and millions of $$$ do you have to spend on this again?


Serious advice: Rethink the whole concept. Buy the elaborate imaging system and use your expertise in optical image recognition and algorithm design to build the software around determining make & model of car.

[solutions]

Or you could do the world a favor and not ENABLE the bastards that take away our freedoms by using the technology we design and build, thinking it is "cool" or that it makes money.

Freedom has a price. Don't cheapen it or sell everyone out for short term personal gain.

Think about the world you leave behind in your flash of existence and try to leave a better one than the one you found.

Yeah, you can build it (or you think you can...yours won't work), but SHOULD you?

This is meant to provoke EVERYONE, no matter where you are, to think twice about enabling the robbers, crooks, and oppressors, even if it's money or opportunity lost for you personally.

FWIW, this works VERY well:

[htrx93]

Thanks for the input, this was never intended as a method to recognize specific vehicles, just broadly categorize them for doing things like evaluating tolls and predicting road wear. Yes I definitely agree the pervasiveness of camera systems and surveillance these days is frightening.
It's amazing what you can do with a garden hose and some pressure switches
I currently have very little $$$ and would only play with this on a small hobby scale. When I do things on my own though I like to try and focus it in a direction I'm interested in, and also which will improve my engineering skills in some way.

[sn00p]

Can I just suggest that you don't waste your time on this, find another interesting project that will work!

I work in this area (vehicle detection/classification) and this is not going to work!

[Toxible]

Hi all! Sorry for the long time but I was very busy.
@htrx93: yeah what you've wrote is correct.
I've used an IR emitter in reverse mode and I've amplified the signal whith an OP-Amp (that implement a band pass filter)
A simple bjt at the end of the circuit implements another amplifier.
At this moment I'm in my classroom. I'll post the schematic when I come back home.
I'm worried about the resolution. If you have a transmitter and a receiver very distant, an angle of 6° may be not enought narrow...

[Tanuki]

Piezoelectric coaxial cable is commonly used in traffic counting applications.
There are application sheets and other data on this website.
In highway use applications, generally an additional sheathing is used to limit wear on the sensor cable.
Depending on the term of deployment, either a surface mount or an embedded sheathing systems may be used.

http://www.meas-spec.com/piezo-film-sensors/piezo-cable.aspx