3

I want to detect the rotation speed of an impeller (fan wheel). Imagine a fan without a motor submerged in a liquid. This device is used for visual indication of liquid flow. I want to detect the rotation form the outside without modification of the device, therefore the only choice is optical.

The fan wheel is behind around 1mm of plexiglass, is black color, and is submerged in red liquid which is transparent. Also its rather higher speed, but not as fast as a regular fan (maybe up to 100rpm).

Is there any SMD sensor that could be used for this project? I was thinking of a color sensor, to detect the black fan blades, but i'm unsure if they are fast enough.

sgt_johnny
  • 809
  • 4
  • 11
  • You don't need a color sensor because black is not a color. I would use a white led and a phototransistor instead. – Sim Son Jan 30 '21 at 22:33
  • 1
    Not colour sensor, IR LEDs with phototransistors with modulation for noise immunity, or just get IR sensor modules. But test because that submerged part might throw a wrench in things. Most common plastics are transparent to IR which is good for the plexisglass...but what is your impeller made of? – DKNguyen Jan 30 '21 at 22:34
  • 3
    The usual way for fluid velocity meters is with a magnet. Put one in each tip you could then detect it many inches away. – Kevin White Jan 30 '21 at 22:34
  • If you really won't consider adding a magnet or reflector, with a laser (who knows what color? pointed at the actual blades of the fan at the right angle (especially if you can angle it to shoot through gaps in the blade, you could sense the dot from another angle and count blades. – K H Feb 01 '21 at 04:04
  • Oh red fluid. Probably has to be a red laser. – K H Feb 01 '21 at 04:34

4 Answers4

2

Omron and others make reflective sensors, an IR emitter and photo-transistor fixed in position in a plastic housing. They are angled toward each other to set up a detection range. Some have digital outputs, which might work against you. One with an analog output, or just the phototransistor pins, could be processed to extract the sinewave signal of a passing fan blade. That signal drives a comparator that gives you a square wave signal with one pulse per fan blade (not per revolution).

If there is too much interference from the environment, the next step up in complexity is to drive the emitter with a high frequency square wave, something like 100 kHz. The received signal then is bandpass filtered to reject ambient junk, and peak detected to extract blade information.

AnalogKid
  • 16,865
  • 1
  • 13
  • 25
1

at 100rpm (x number of blades), nearly all electronic sensors will be fast enough. Provided you can arrange consistent lighting (and it can be set to a known frequency, well above the one you're detecting, to make it more robust), it sounds like you're looking for a photodiode or phototransistor, with probably lens and a pinhole aperture. There are convenient integrated industrial solutions for this . . .

Also do consider magnetic as mentioned in the comments.

Pete W
  • 1,207
  • 1
  • 5
  • 12
1

I know that the turbocharger impeller speed is measured with the securing nut being magnetised - rpm way above what you expect in your water based device.

Perhaps you should consider magnetic cf optical - no issues with light conditions and sufficient working range.

If you purely want the flow, there are ultrasonic sensors that clamp to the outside of the pipe and will detect the flow - if installed very well they can be accurate but they can also be troublesome. You do need a sufficient length of straight pipe for the flow profile to develop though.

Solar Mike
  • 6,319
  • 1
  • 12
  • 28
  • I thought of magnetics too, but this would require tampering the device (to install the magnet), also i'm unsure how the magnet would interfere with the rotation since the rotation is only generated by fluid – sgt_johnny Jan 31 '21 at 16:46
  • @sgt_johnny if you can't interfere with the device, how will you fit reflectors on the blade or blades? – Solar Mike Jan 31 '21 at 18:00
  • @sgt_johnny see the update. – Solar Mike Feb 01 '21 at 07:41
1

If you use a very narrow beam LED (a few degrees) you’ll get a high light intensity at the receptor, which you can also put in a tube so it has a small viewing angle. If you high-pass filter the output it should ignore most changes in ambient light. If you want more robustness, modulate the LED at (say) 100kHz and notch filter at the receptor. A trick I once did was to transmit characters at a high baud rate over the optical link and then see whether anything arrived at the other end - about 6 lines of code on an ATtiny microcontroller.

Frog
  • 6,686
  • 1
  • 8
  • 12