I am making an infrared break beam to detect insects, using an IR emitter LED and detector. The insects fall into a channel and then eventually cross the beam, which travels the length of the channel.

enter image description here

The insects are quite small, so I want to make the beam as narrow as possible; otherwise an insect could block part of the beam and go undetected. The infrared light from the LED has some spread. I was using a pinhole to screen the light into a beam.

enter image description here

Through trial and error with 3D printing I found a pinhole profile that works: pinholes of diameter 1.3mm and depth of 3mm on both the emitter and detector can detect an insect with a cross section of 3mm by 1mm.

However, I want to make the beam even smaller, to detect an insect with a cross section of 0.5mm by 0.5mm, and I would like to find a more rigorous way to define a pinhole profile. I am not sure what effect the pinhole diameter and depth have on the size or spread of the beam. I have three questions:

1) How do I choose a pinhole diameter? If a laser goes into a slit, there is diffraction, but is this a problem for an extended source like an LED?

2) How do I choose a pinhole depth? A thicker depth could reduce false readings from IR light reflected from the side of the channel.

3) Do I need pinholes on both the emitter and detector? I had some challenges with IR reflections from walls when the emitter did not have a pinhole, but that might be avoided if the pinhole is made sufficiently deep.

Right now the insect trap is being 3D printed, but eventually it will be manufactured (eg injection molded). I would like to avoid lasers or lenses for size and cost restrictions. I am not interested in focusing the LED, and I understand that I will only be accessing a fraction of the LED's output light.

EDIT: The emitter (datasheet) and detector (datasheet) are both lensed to increase their angular spread, so some sort of aperture is necessary to get a narrow beam.

  • 1
    \$\begingroup\$ Diffraction applies to any light, not just laser. \$\endgroup\$ – Chupacabras Aug 21 '18 at 6:39
  • 1
    \$\begingroup\$ I had designed a multiple beam system using TDM multiplexed emitters and detectors such that blocking one path at one at one instant allows another path to maintain the AGC, so that a relative change was very sensitive. All paths were 1m apart with 5mm optos recessed behind 5mm holes and could detect moving blockage of a 0.5mm wire in the middle. Something like the answer here for a different problem. electronics.stackexchange.com/questions/329438/… never tried bugs. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 21 '18 at 7:56
  • \$\begingroup\$ I used an IRDA detector and std 10deg 5mm IR emitter with a unique 8 bit code for each emitter with sufficient intensity such that a data error in the corresponding blocked path indicates which path was blocked. digikey.ca/product-detail/en/vishay-semiconductor-opto-division/… \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 21 '18 at 8:18
  • \$\begingroup\$ Actually the lens on each is designed to reduce the beamwidth from Lambertian (165') to 40'. But emitters with <20 deg (+/-10) are better. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 21 '18 at 22:39
  • \$\begingroup\$ what is the channel width if the target is 0.5mm? and maximum light path? \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Aug 21 '18 at 22:59

The beam is the straight lines directly joining the emitting area to the receiving area.

The emitting area of an unfocused smt led is quite small, maybe 0.2mm square. So it is already the pinhole size that you want. A small led and small photodiode will give you a small beam.

You do not need to cut out the light that goes in other directions, as it simply does not lie on the beam.

Note that using an unfocused LED like this, gives you the same intensity in the (narrow) beam (mW/mm^2), that you get with a focused LED - you are not losing anything. This is counter-intuitive, but true. No lens or mirror is able to get any of the off-beam light back into the (0.2mm) beam - it's impossible.

You want to ensure that your tube is not reflective i.e. matte black, just to reduce the scattered light that reaches the detector and reduces the darkness of a blocked beam

enter image description here

If the chamber is close to the led, then the beam size is the led emission size. If it is close to the detector, then it is the detector size. You can change the beam size by choosing these.

When you have a choice, you use baffles (perpendicular to the light) not flat walls. The next choice is machining grooves in, which has a 90 degree corner angle.

You can have multiple leds with one photodiode. You can also use an image sensor, area or linear for the photodiode, and sample a big area at once. Linear sensors are quite simple and low power to read with a micro.

For lowest power consumption VCSEL lasers are also available in SMT, and achieve high beam intensity at lower DC power, as the spread is smaller (eg 8 deg vs 160 deg)

You can "remove" the lens by putting a drop of clear epoxy onto the lens, and pressing a microscope cover slip flat onto it (no air bubbles). Any flat, clear sheet will do.

You can align baffles with small holes by threading them all onto a rod or tensioned thread, from sender to receiver, then fixing them in place and removing the rod/thread. [if it is vertical then it does not droop]

When working with infrared you must test your "black" materials (ink, paint, plastic). Carbon-loaded black is black. Red+Blue dyed black is transparent to NIR. Black plastic that lets IR through is common.

Infrared leds can take very high pulse currents. Pulsing with high currents and short pulses, which have a repetition rate such that you cannot miss a bug, will likely give the best system.

| improve this answer | |
  • 2
    \$\begingroup\$ Also, modulate the beam at (say) 40kHz. That way you can apply a high pass filter at the other end for a better signal. So even if the insect only partially blocks the beam it should be easy to detect. \$\endgroup\$ – Dirk Bruere Aug 21 '18 at 8:17
  • \$\begingroup\$ @ Harry Many LEDs has lenses to broaden the beam, thus the silicon emitter area is not the beam area. \$\endgroup\$ – analogsystemsrf Aug 21 '18 at 15:29
  • \$\begingroup\$ Unfortunately both the LED and the phototransistor I am using are lensed (added datasheets in question). Would the baffles/grooves replace the flat walls in the pinhole or the channel? \$\endgroup\$ – Zach Morris Aug 21 '18 at 18:20
  • \$\begingroup\$ @ZachMorris 1) using lenseless (smt 0603) leds will be easier than constructing and aligning tiny pinholes. 2) you can remove the lenses by (a) sand and polish them off b) put a drop of clear epoxy and a cover-slip onto the lens rendering it flat \$\endgroup\$ – Henry Crun Aug 21 '18 at 19:33
  • \$\begingroup\$ @analogsystemsrf Not after my belt sander and polishing wheel have been at them they don't... \$\endgroup\$ – Henry Crun Aug 21 '18 at 19:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.