# Detecting a very fast moving object with a photoelectric sensor

I want to use a photoelectric sensor to detect a ball that is moving past the sensor at 85mph, which also equals 136kph or 3800cm/sec. The ball is 7cm in diameter. If my logic is correct that means the ball will be in front of the sensor for 7cm/(3,800cm/sec) or 0.00184 seconds or 1.84 ms. Actually, it will be in front of the sensor for a few ms longer because of the width of the sensor eye itself but I think we can safely ignore that.

I am trying to understand what sensor would be able to detect the ball. Many list something called 'response time'. If the response time of the sensor is 2ms does that mean it would not detect this ball? If it is 0.8ms would it? I am going to call an engineer at one of the sensor companies but was wondering if anyone here might be able to enlighten me about how these sensors work in terms of fast moving object and what specification to focus on.

Thanks!

• 2mS seems huge for an optical sensor. What sensor are you looking at? A photoresistor perhaps? – Trevor_G Oct 18 '17 at 1:16
• You should be using a photo-diode or photo-transistor.. example osram-os.com/Graphics/XPic8/00206695_0.pdf switching time 5nS – Trevor_G Oct 18 '17 at 1:21
• I'm just shocked no one has asked you about the environment this is in? Please provide details about the environment, lighting situation, proximity of the ball to the sensor, use of a source of specialized light (if any), and any appropriate geometry. Focusing on the sensor seems to me like putting the cart way out in front of the horse. – jonk Oct 18 '17 at 3:40
• Some other minor details are also needed such as are you using another LED on the other side to provide base signal or are you relying on reflection alone. Fun fact, you can use LED as a light sensor, but they're not as sensitive as phototransistors. Look up photoconductive and photovoltaic circuits. Having an opposing LED would likely give you best response time and if that's not possible, using emitter/sensor pair (IR most likely) and relying on reflectivity would work but you'll get order(s) of magnitude less signal unless your ball is bright white or something. – Barleyman Oct 18 '17 at 11:07
• Come to think of it, unless the ball is following a predictable trajectory, reflective setup would work better as the signal level has less variance. If the emitter is on the opposing side, there's a pretty big difference if the ball is actually blocking the emitter or offset above/below it. Photointerrupter works very well if you have something moving on a rail or conveyor belt or something. – Barleyman Oct 18 '17 at 11:15

A response time 1.4ms is a long time for an optical sensor. We can detect a short little bullet moving at 1400fps (~430m/s) fairly easily. Mechanical motion is much, much slower than the signals on fiber optic cables.

If you were to do it with discrete parts you could use most phototransistors (working into a low resistance or virtual ground to avoid Miller effect), probably not a photodarlington and definitely a photodiode, in conjunction with a suitable amplifier and comparator.

However there are quite inexpensive all-in-one 3-pin logic-output photosensor devices such as this one that will respond to 250kHz, or about 10x faster than you require.

Of course you have to meet the optical requirements for this to work reliably.

There are a lot of variables: color/reflectance of the ball, interference from the environment, distance to the ball, etc.

Without this information, there are a few things you can still consider. Your response time would certainly need to be shorter than the amount of time the ball passes by the sensor for (ideally, much faster). If the response time is longer than the amount of time the ball will be in front of the sensor, then your behavior is most likely undefined. Many phototransistors have response times on the order of nanoseconds, so you should have no problem finding one that satisfies this.

You also need to make sure you are sampling your sensor fast enough to detect it. What kind of system is this run on? An Arduino should have no problem sampling a sensor every few tens or hundreds of microseconds, but wouldn't be able to do much else except possibly basic instructions in the meantime.

• Phototransistor may very well react quickly but the problem is usually the signal amplifier after it. – Barleyman Oct 18 '17 at 11:16
• Thanks for your reply. The system will be run on a customer pcb that is being designed. I don't know if the sensor is sampled or the sensor detects and then sends the signal. I would think the latter but that's something the electronics engineers that are designing my pcb are going to determine. – electeon Oct 19 '17 at 2:37

Sorry that I did not post all of the parameters. The ball will pass by about 20mm from the sensor at most ever 2 seconds. The sensor will be mounted inside a machine that has an opening in it so a little light will be coming from the outside. The sensor I am planning to use an off the shelf sensor where I can just connect the wires to a motherboard rather than solder components together to create it. I am planning to use a diffuse reflective optical/photoelectric sensor as it does not require a reflector, just the light reflected from the ball. I am pretty sure I do not need a diffuse sensor with background suppression as the reflection from the inside case of the machine will be very low and not reflected right back.

I do not think that focusing on the sensor is putting the cart before the horse at all. Rather it is the most important part to focus on as one must first find a sensor that operates at a certain detection distance (I am choosing 30mm max), operates on the diffuse reflection method and has a response time that can detect a ball moving at a speed that puts it in front of the sensor for only 1.84ms. All else is secondary. I have sourced about a dozen sensors and they seem to all operate with response times of 0.8-5ms. None were in nanoseconds. An engineer at one of the top sensor companies, Banner, who sell 12,000 different sensors, told me that the faster the response time the better and that 1.5ms, the response time of one sensor I was looking at in their lineup, is cutting it a bit to close to 1.84ms. Also I found out that the response time is also the time the sensor needs to not be active in order to detect the next ball.

By the way, Banner does not have a single sensor whose response time is measured in nanoseconds. The fastest is 0.15ms.

I found my solution. Now comes the testing as specs are one thing. Real world testing is another.

Thanks everyone!

To get a full amplitude response you need to response at least twice as fast (<0.5T) as the event. Anything slower will significantly attenuate the signal.

We know that sensitivity of silicon is usually a fixed gain-BW product or a fixed gain/time response ratio. For Op Amps this is defined as GBW. The same is true for Photo Diodes.

But at lower light levels, the current is low and impedance is high. When multiplied with the diode capacitance, this can increase the slew rate worse than what it is capable with a low resistance feedback from a Trans-Impedance Amplifier Op Amp to convert the diode to a much lower impedance.

e.g. VEMD2000X01 is rated for 100ns at 10K load.

Now the best method is to use LASER interferometry on the reflected signal, but it may be possible with other methods depending on location restrictions.

• I think OPs goal is easily achievable with cheap off the shelf parts. No need to get exotic. – Drew Jul 14 '20 at 8:10
• Maybe. But the SNR change will be a challenge with diffuse detection of a small area reflection from a sphere. The ball is a 20 deg subtend (6/20cm) of which only a small fraction of the parallel surface reflects to the sensor which is diffused to see >160 deg. So a very low SNR. The more diffuse it is, the bigger the object appears and the slower the inverse result of the PW50 pulse width. Diffuse sensing is not the way to go. Thus you may need an accurate aperture size and slew rate to capture the event. Wonder if it worked (?) fyi @Drew – Tony Stewart EE75 Jul 14 '20 at 12:56
• – Drew Jul 14 '20 at 18:38