Tracking Ping-Pong Ball in 20 feet of Glass Tube

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
Building a very tall PID control ping-pong levitation tube.
Closed loop, with blower speed control.

Tube- 20 feet of vertical pyrex tube, made in joined sections, 42 mm ID
Ball - Standard Ping Pong ball

I need to figure out a way to get solid position feedback with the following specs:

10 readings / second or more
Resolution- 9 bits or better
Accuracy - 2 inches- not terribly critical.
Power- not critical, AC is available

The ball will be illuminated by a powerful (300 mw /color) RGB laser engine, with PWM brightness control, this might cause problems with laser-based ranging systems?

It is preferred to accomplish ranging AND illumination from the bottom of the tube.

I was hoping to use a simple ultrasonic ranger, but after researching the physics a bit, it seems that
short sound wavelengths don't like long tubes.

Thinking I might be able to use low frequency sound pings, the same way that the ultrasound units work, but
at a lower frequency (4khz) that will propagate in a 42 mm tube? The thinking here is that I could use an inexpensive tweeter and a tiny capacitor microphone to accomplish this cheaply.

To get good return echos, I might need to increase the acoustic impedance of the ball by coating the inside with some resin.

Any ideas?
 

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
Camera and image recognition?
The finished project requires that these tubes are modular, self-contained, 64 units total.
It's a high profile art project.

The vision idea would work, but it violates the design / installation requirements.
 

#12

Joined Nov 30, 2010
18,224
Just noodling...
speed of sound is 1126 f/s
time for 40 ft = 35.524 milliseconds.
Number of measurements at 10/s means 100 milliseconds allowed per measurement.
If the air stream was interrupted with a spinning disk with holes in it, such that it allows 10 pulses per second, you could measure the time of the pressure bounce in about a third of the time allowed. That keeps you out of the realm of aliases.

The altitude of the ball can be modulated by the amount of pressure behind the disk.
 

wayneh

Joined Sep 9, 2010
17,496
Won't the position of the ball effect the resonance of the tube? Like an organ pipe or a slide whistle? Just thinking out loud.
 

GopherT

Joined Nov 23, 2012
8,009
I am guessing these will develop a natural resonance tone depending on the height of the ball. Try a microphone in a quiet room to find your response curve. Then use software filters to find the tones at the final installation

Edit: Damn, interrupted and my post was delayed! Nice work, WayneH
 

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
I am guessing these will develop a natural resonance tone depending on the height of the ball. Try a microphone in a quiet room to find your response curve. Then use software filters to find the tones at the final installation
Hmm? how to excite the tube to continuous resonance? Or pump in some kind of impulse in and see what the resulting resonance frequency is?
Different approach than the sonic time-of-flight idea, using a ping or burst.
 

#12

Joined Nov 30, 2010
18,224
Look at the way an organ pipe is built. Air is injected at the bottom and a slot causes the air movement to produce a continuous sound. You already intend to inject air. Discover how the shape of the slot works and you have the 20 foot long frequency generator whipped!

You can make the injection port and slot as a separate module to avoid needing the skills to work with glass to get your whistle ringing.
 

ErnieM

Joined Apr 24, 2011
8,377
Do you need to continiously measure the height of the ball? I would expect corelation between airflow in the bottom to height of the ball.

Do a set of measurements in the prototype once and use the resulting data to build a look-up table.
 

wayneh

Joined Sep 9, 2010
17,496
It doesn't work that way. The ball is almost always at terminal velocity regardless of its position, so it could be anywhere in the pipe at that precise flow. You need to vary the flow (with feedback) in order to position the ball by briefly accelerating and decelerating it.

There are flow meters that use, instead of a tube, a cone. In that case, the position of the weight depends directly on the upflow rate.
 

John P

Joined Oct 14, 2008
2,025
How about a TV camera looking up the tube, so the apparent size of the ball would be larger or smaller according to its distance? I can imagine that scheme working tolerably well if the ball is low down, but at a range of 20 feet, you wouldn't get much accuracy. It might work better if the ball has dots or patches on its surface which would be recognizable by a vision system.
 

GopherT

Joined Nov 23, 2012
8,009
Ok, here is an out of the tube idea...

Add a pressure sensor to the bottom of the tube near the blower. The sensor cannot measure the height because, as pointed out above, the ball is at terminal velocity all the time (relative to the flowing air).

However, to find the height, you need to cut the motor for some fraction of a second, the ball will fall the same distance (~1") no matter where it is in the tube. The key is, the percent pressure change caused by the falling ball will be different depending where it is in the tube. And the rate of recovery of the original pressure will be different based on the different heights of the ball. Mostly an idea to get people thinking.

This idea will only work with a fast and sensitive pressure sensor, and the response will be greater with greater ball weight and smaller gap between tube wall and ball. Pressure sensors are available that can measure the atmospheric pressure change as you walk up each step of a building. I think this idea is possible, it just may become a year long research project.

It's the best I have - you (or your artist) are asking a lot.
 

wayneh

Joined Sep 9, 2010
17,496
Turn up the air. Tether the balls with fine monofilament. Reel them in and out with a stepper motor, like a printer head.

I think the right filament would be very hard to see inside a glass tube.
 

Bernard

Joined Aug 7, 2008
5,784
To view wayneh's filament suggestion see Bernard's pictures or photos as in avatar, filament suspending planes is .37 mm mono filament.
 

Thread Starter

Sensacell

Joined Jun 19, 2012
3,432
Thank you all for the interesting ideas.

The monofilament idea is really creative and clever, I will see if my client will bite on this one, but if he does not, I still need another solution.
The only problem I see is the laser beam illuminating the ball would need to be off-axis from the monofilament, or perhaps run from the opposite tube end, via a pulley system? Can not have the beam glinting off the monofilament line.

Some general comments to address some of the questions asked:

1) The ball will be very busy, constantly moving up and down, with static dwell periods at random locations. The idea is a choreographed motion ballet of sorts.
The ranging system needs to provide rapid and repeatable feedback so the control system can have a prayer of performing the task.
System reliability is also very important - 64 units must work - for a long time.

2) Ultrasonic waves of wavelength shorter than the diameter of the tube tend to refract and dissipate in a long tube, rendering this technique useless for all but short tubes. I confirmed this via a discussion with sales reps from a variety of ultrasonic sensor companies, they all said that this technology would not work for my application. I am however investigating the idea of using lower audio frequencies (4 khz) that are supposed to propagate through tubes as "planar wave fronts" - like the way old ships had long tubes between the engine room and bridge for voice communication. I want to try using a tweeter and a tiny condenser mic to do the ranging. This idea is the front runner, pending some testing / RD.

3) I don't see a pressure sensing concept working, the ball is in dynamic flight at all times, the problem being the large clearance around the ball, which is necessitated by the tolerances of the tube diameter. The dead volume under the ball might undergo some pressure change, but it would be vanishingly small and greatly influenced by the momentum of the 20 foot long slug of moving air in the tube, which has a non-negligible mass.
 
Top