I’m wondering if anyone has any experience with retroreflective photoelectric sensors?
I have a upcoming project where I need to measure the number of objects falling through a pipe (approx. 12"-24" diameter). The pipe is located in a harsh, salt water, environment.
My concern is how a sensor like this will perform if the reflector, or sensor, is wet. I can get IP67 rated sensors, but of course this just means they will survive being wet, not actually work as intended.
Of course, if anyone has any other solutions I’d be keen to hear them, as I’m having a real mental block with this one.
I have used ones similar to these for bag sensing in an industrial dialysis mixer. They are used to detect bags being in position for automatic opening process. The environment is similar to what you mention as citric acid wash is being blended by the machine. Afraid I can’t help on the “being wet” issue. Our enviroment is humid, but not directly being saturated at all times.
That would probably be ideal Klaus, but the range looks to be the problem with a capacitive sensor. The objects are irregular shaped so could be up to almost the full diameter away from the sensor. I appreciate the suggestion though.
Correct, we have between 3-5 years of use on various machines and humidity has not been an issue.
I agree with the range concerns on capacitive sensors. We use some of those type for no-contact water level sensing. Aside from the range in general, we have found them to be fidgety in certain conditions. For example, tune them up to read a specific level of pure water and then find them to be asserting when an acidic mix residue has coated the inside of the tank. We have had to introduce a spray rinse before checking them to add confidence to the liquid level.
What, exactly, is the definition of “wet”? Are we talking about a pipe FLOODED with salt water, as in the falling objects and sensor are all submerged? Do they just get splashed regularly? Are you trying to count the number of crabs falling into a live well, where they come in on a sluice with a bunch of water mixed in?
My guess is fully submerged and bubble-free, and humid but non-condensing are the acceptable use cases for optical. If there are droplets and splashes of unconstrained geometry to scatter light unpredictably, my guess is that’s going to be hard to sort out.
Ditto acoustic/ultrasonic, which would be my backup plan - things go haywire as soon as the sensor has splashed water on it, and also it’ll have difficulty distinguishing a cloud of droplets from a target object.
EDIT: Another thought that quickly jumps to mind, to get around optical aberrations is, rather than using a retroreflector on the opposite wall of the pipe, use a wide-angle (laser) line projector, or probably 2-3, and multiple receivers spaced around the circumference. Then if MANY sensors agree there’s a break, that’s an object to count, rather than droplets. Or at least, I bet with that setup you can look at the data and come up with some rational algorithm for discriminating the two scenarios.
No the objects will be wet and so we’re speaking about splashing rather than soaking.
Spot on! However, they are sorted by hand first and then dropped into the live well.
That’s my concern. I think I’m going to have to bite the bullet and do some testing.
I also think there might be issues using an ultrasonic due to minimum range and acoustic reverberation in the pipe since they are typically for use in free space. I’ve been caught out with ultrasonics in tanks before.
I agree that multiple sensors would allow for data filtering and also would allow for real time fault checking. I like the wide angle suggestion - I’ll explore that too.
Oh what about this? Invert the polarity of the signal. That is, instead of a breakbeam sensor, design a makebeam sensor.
Paint the entire interior of the tube black, ideally matte black, and use a diffuse illuminator like a really bright, un-lensed COB LED to light the tube up from the sensor’s vantage point. When a crab falls by, DEFINITELY on the belly but probably on the top too, you’ll see a blip UPWARD in reflected light, since it’s brighter than the black tube. Since the illuminator and sensor would both be diffuse/unfocused, they should be relatively more resilient to fouling from drops or puddles.
It also might work to use something like one of these new ST grid-array ToF sensors designed for counting people passing through a doorway. They have two zones, so you can detect an object entering the first zone, then passing to the second, then leaving, as an “entering person,” and the reverse as well. Hopefully you don’t have any crabs shooting up the tube, but the 1-2 hit probably allows you to avoid fouling from moisture, droplets, etc. Even if droplets are refracting the light and causing off-axis measurement, I bet you still see a one-zone-then-the-other change to detect a drop.
Oh and I never mentioned an idea that jumped to mind earlier - I bet, if you put a microphone in the middle of the tube, you’d be able to hear the acoustic signature of a crab flying by, probably regardless of whether the mic was wet or not (but also, that could be avoided with strategies similar to modern GoPros). If I’m right about that, you could probably use machine learning and JUST a microphone input to count. That’s also probably the most complicated strategy of all.
You could really maximise the complexity and just go straight for computer vision - which would mean setting camera back from the pipe, so you keep the camera lens dry, and get a clear, undistorted view of the spread-eagled crabs (fun finding a starting data set for that model!).
But that then made me think that the a pipe dropping into a cone (closed at the top) would allow you to use the breakbeam sensor towards the top of that with very few water droplets (where there’s a discontinuity in diameter) and still have the crabs fall down the cone to the rest of the pipe without anything to hold on to. Kind of like a collector/hopper in a rainwater guttering system…