What type of photo sensor/light combo best for detecting a falling drop of water?

I am trying to setup a rig for water drop photography. This is where you take a picture as a water drop is impacting against another object or body of water. The trick is to detect the drop as it falls and then use a tuned time delay.

I want to detect when a drop of water interrupts a light source as it falls. The most important factor is that it be repeatable. I need it to respond to the interruption consistently. The sensor also needs to be fairly quick as the drop only interrupts the light very briefly.

I have seen photo resistors, photo transistors, and even photo diodes. Which of these or other parts would be best for my use?

What sort of light is best? Should I use a laser or IR or something else? Water is clear in normal light which could be problematic, is it clear in all spectrums?

How do I deal with ambient light? I have heard that by pulsing the light source at a known frequency it can be differentiated from surrounding light.

I am basing the project on an ATMEGA328 but am happy to add other parts to work with different sensors. I am also interested if anyone has a solution other than light based sensors.

• Approximately how fast do you estimate the drop will be falling? How much ahead of the impact (or photographic event) does the camera need? To make it relatively easy, I would expect the system to need plenty of time between detecting the drop, and triggering the camera, if it is a still camera. – gbulmer Oct 23 '15 at 7:17
• If you look at waters light absorption spectra en.wikipedia.org/wiki/Electromagnetic_absorption_by_water#/… You will see that water absorbs more at infra red frequencies than at visible frequencies so you will probably do best with a an IR LED and photo diode pair. – Icy Oct 23 '15 at 7:29
• @gbulmer the water drop will be travelling fairly slow in terms of MCU and camera speeds. Even if it was going critical velocity for a drop of water it would not exceed 7.5m/s. I expect about a meter of buffer between the sensor and the target so speed of response is less important than repeatability of response. – HighInBC Oct 23 '15 at 15:27

Initially, I'd try a narrow angle visible light LED and phototransistor, and set up the phototransistor in 'line of sight', maybe inside a tube to reduce the chances of refraction or reflection from the droplet effect the sensor. Something cheap and simple like an old pen body

The rationale is the visible light LED would make it easy to set up. Infrared (IR) can be awkward because it's invisible. If it works well enough, then it's okay. If not, move to IR. Assuming you make a stable mount for a visible LED, say a 5mm hole, an IR emitter should fit in the same place.

You might find that reflection from the water droplet, either above or below the trigger point, needs to be dealt with.

A phototransistor would be good because the change in output, with a resistor in series, may be enough to directly trigger a digital pin. The AVR/Arduino ADC is quite slow, about 10ksps, so try to avoid it if you want high precision.

A phototransistor can typically respond in under 20µseconds, so it should be fast enough if the drip has fallen only a few metres. This is significantly faster than the AVR/Arduino ADC.

Most AVR/Arduino's include a comparator, so it may be used to detect the phototransistors change, and ensure that is adequate to detect using a digital pin.

Edit: One of the comparators inputs would come from the connection between phototransistor and its resistor, the other comparator input sets the 'trigger' voltage. That can be adjusted using a potentiometer as a voltage divider, so the threshold could be tweaked manually to get a good result.

Flashing the emitter is to avoid ambient noise. The sensor is 'sensed' when the emitter is on and off. However, this is most useful when using the sensor as an analogue device. Ideally the sensor wil be working as a digital device, and with some light shielding, will be somewhat insensitive to other light sources.

If visible light proves too noisy, use an IR emitter and sensor, with visible light blocking.

The parts should be under \$2, so having two attempts shouldn't be to costly.

• Using the analog comparator you mentioned should allow to use either of photo resistor/transistor/diode as input. On the second pin the voltage reference and thus the triggering behaviour could be adjusted with a potentiometer. – Grebu Oct 23 '15 at 14:25
• Good point. I wasn't clear enough, so I'll update. – gbulmer Oct 23 '15 at 14:35
• Good idea about using the phototransistor to trigger a digital interrupt. It allows me to avoid polling which is the enemy of precision. I am going to give this a bit more time but will accept this answer if I don't get a better one. – HighInBC Oct 23 '15 at 15:30

This is an absorption spectrum of water:

Source (Kebes at English Wikipedia [ CC BY-SA 3.0 or GFDL], via Wikimedia Commons)

It shows the absorption coefficient $\mu$ for the formula of the intensity $I=I_0\cdot e^{-\mu\cdot x}$, where $x$ is the "thickness" of the layer of water.

Common lasers and LEDs are in the wavelength range of 350nm to 850nm, and according to the spectrum, 850nm shows the higher absorption coefficient of about 10/m. Assuming a huge water drop of 1cm, the formula gives 0.9, which means that 90% of light are transmitted, and only 10% are absorbed. For

So, for common light sources, absorption is too low.

• Absorption is pretty much irrelevant. A drop of water will act as a very short-focal-length lens, and refraction will cause the beam to be interrupted. – WhatRoughBeast Oct 23 '15 at 14:48
• @WhatRoughBeast: I thought this is the outcome of my answer... One question was if water is clear for all wavelengths, and I just wrote how to find this out. – sweber Oct 23 '15 at 18:50

I'd build a photointerruptor based on an IRLED pointing at a silicon photodiode across a gap through which the water droplet would fall.

The diode would be operated in reverse-bias mode and would be illuminated all the lime by the IRLED so that when the water droplet fell through the beam it'd momentarily shut off the IR incident on the photodiode, creating a pulse which could be used to start your timer.

You deal with ambient light by using a daylight-filtered photodiode and an IRLED with a narrow beamwidth and an output wavelength matched to the photodiode's. You could also enclose the thing in a housing (a box with 2 holes in it for the water droplet to fall through) designed to eliminate extraneous light from getting to the photodiode.

I don't think the clarity of the water will matter, since the droplets's geometry will disturb the beam's interaction with the photodiode, which will give you the output you're looking for.

Visible or IR won't make much difference, and you've got answers that go into the electronics better than I could, but don't forget the optics.

Essentially everything has to line up very well. If you're using an LED you'll probably want to have a small hole in some foil in front of of it so that the droplet covers the hole as it falls past. A matching hole on the receiver wouldn't be bad idea. A laser won't need a pinhole in front of it but you'll probably have a harder time tuning your detection circuit -- a hole in front of the detector may help with this. For an LED you probably want everything quite close together, but for a laser some distance between the drop and the detector is a good idea.

I'd try to use a laser a few mm to a few cm from where the drop falls, and have a phototransistor a few 10s of cm beyond that. If the laser is just off centre on the drop that will increase your contrast, as will a tube around the photodiode. You may be able to use an ND filter in front of the diode, which will attenuate the ambient light and the laser equally, but the laser will be so much brighter to start with (when there's no drop) that your contrast will go up.

I did some drop photography about 20 years ago using a different approach: the drop was charged to around 2.5kV using a photomultiplier power supply and the detected using the voltage pulse it induced in a coil just out of shot. It worked perfectly, which it needed to as I was using 35mm where mistakes get expensive.

• Wow, I never thought of charging the drop of water and detecting it with a coil. Clever. I can't imagine trying to do this with film, it is hard enough with instant feedback and free pictures. – HighInBC Oct 23 '15 at 16:04
• Getting the timing offset was interesting but simple mechanics combined with analysing the coil voltage pulse on a scope meant I could actually get a really good timing estimate, then a simple RC delay (IIRC) manually set up did the actual triggering. It was a physics undergrad project so a sophisticated trigger circuit would have been wasted effort. I wonder if I've still got my writeup. – Chris H Oct 23 '15 at 18:08
• I did try an LED-photodiode pair but couldn't get it to work (not enough contrast). I assume this is for artistic purposes rather than scientific. Because adding a little methylene blue to your water would make it very good at blocking red lasers. – Chris H Oct 23 '15 at 18:21

Ancient thread, But I just came upon it today. Sweber post on light spectrum solve it for me.

So, for common light sources, absorption is too low. Instead, use the reflection and deflection of light in water. @gbulmer has already written anything I would about this, so I'll stop here.

IR LED/IR phototransitor won't detect a drop of water when these two are in front of each other. BUT placing these two inside a tube, at 45º angle from each other, and it works by detecting the reflection of IR light in the passing droplet.

Pure genius...