Sound Based Feedback System

Rube, is that you? Come out from behind that AP mask. What is you expect to hear - the exhaust air or click from a solenoid? Either way, the only way to really monitor the conditionposition of the cylinder is with a sensor on the cylinder or whatever the cylinder actuates. Maybe just a limit switch. What happens if the solenoid actuates but something, anything, prevents the cylinder from actuating?

What about something like this....or based on this method?

http://www.ifm.com/ifmuk/web/pinfo010_040_040.htm

Try searching for, "OpAmps or multistage transistor circuits".

This does not sound like a practical approach at all. Just because you have a bunch of parts, this doesn't mean that you should use them.

A speaker is an inefficient transducer for converting electrical energy into acoustic energy. Trying to use a speaker as a microphone will be even less efficient. If you look at the signal from your speaker listening to your pneumatics with an oscilloscope you will find that you are looking at very small signal levels. Without knowing anything about the speaker or sound pressure level, you may not even see anything with just the oscilloscope front end gain. If you can read the signal then you will also find that your microphone will also be picking up other sounds that your adaptive organic processor (brain) ignores. So there is a high probability of false responses. A better sensor of an optical or mechanical limit switch is what I would use.

Now that I've explained why using a speaker seems to me to be a bad idea, the circuit configuration you should consider is a comparator with hysteresis. Some input filtering may help prevent false triggers and stretch pulse to long enough periods so your PLC can capture a pulse. A one shot circuit between comparator and PLC can guarantee that every pulse gets captured, even the wrong ones.

I'm not looking to here the air flow, I want to know when it bangs into the end of its stroke. It is a BANG, not a whoosh. And it will work, because it is intended to ignore the background noise. Consequently, I don't give a rats ass about the frequency response. It is not your usual audio application.

Think in terms of how to create a pulse, as in a peak detector, when it hears the bang. If it is outside your comfort zone, I will understand.

What's outside of my comfort zone is anonymous "guests" looking for free engineering of a bad idea. I find it hysterical the idea that your idea will work because you intend to ignore the background noise.

I will give you another BIG hint. You must care about frequency response. A pulse contains many frequencies. Your noise floor will be related to frequency bandwidth. Your "bang" acoustic signature will be a burst of multiple frequencies, not a pulse. Your data acquisition conditioning circuitry is what you hope to convert to a pulse that is compatible with your PLC.

IMHO you should explore two paths and compare costs.

First, locate a storage scope and measure the signal level coming from your speaker during the "bang". Notice the waveform and amplitude. The feasibility of your approach relies on this information. Then pursue either an envelope detector or comparator detector design to convert this complex waveform into a pulse suitable for your PLC. Don't forget to test your circuit with anticipated background noise. With the large number of devices to be monitored you should also consider the circuit being fabricated on a circuit board to be placed inside some box for safe mounting.

Second, identify a limit switch that can be triggered by the movement of the pneumatics. Wire the contacts directly to the PLC.

Well, I've put my foot in the water, so here goes.

The ubiquitous AP (AP seems to be so busy with so many projects) has, " thousands of pneumatic devices" and, "expect I will need about 300 peak detectors".

So, we have 300 detectors sampling 1,000 devices, with unknown frequency/amplitude and spacing/distance from said 16 ohm flat speakers/detectors and the AP believes he will get meaningful data to determine the function of said 1,000 pneumatic cylinders that will go "bang", instead of "whoosh".

I am obviously unschooled in the nuances of acoustics and how one samples 1,000 random bangs with 300 sensors, that are re purposed 16 ohm flat speakers and determines which cylinder is not functional.

There goes the cynic in me again.

Cheers for that! I was pondering about how to relay my concerns about this without being bashed by AP2 for not knowing what I am talking about!

I leave this to you now, but I am sure you can take a few hits!

I fully agree that cylinder activation controlled by sound will be the same as a fart detector with a fan. It might work but you can never be sure it really did!

http://www.canfieldconnector.com/

Reed switches in bulk they will sell at a very reasonable price.

Also you could use a push button switch switch at the end of each stroke.

I don't know what your trying to build. But to use sound as a signaling medium sounds like a disaster. How you going to isolate all these speakers from the other devices that make sound. Let a lone outside noises. Here we are real close to train tracks. i came imagine every time the train came by and blows it horn at the crossing the machine going hay wire.

Using a speaker as a microphone is used in PA systems on boats, ships and yachts.

The quality and efficiency for audio, however depends on the type of speaker. You can get a good pulse signal out of the speaker once you let your piston push the cone inwards the speaker magnet.

A 16 Ohms speaker can generate up to 20 Volts (a good one) this way and deliver even a power output that is in general used to move it.

Some disadvantages:

The cone must be centered into the magnet when pushed, otherwise the coil might be damaged by scraping the shellack off the wire.

The cone, spider and suspension have a certain built in life time. Rubber, if any can degenerate. Paper can get wet and become floppy etc.. etc..

You can use these speakers with success, even to drive a speaker at the receiving part to hear the plops. But the longevity will depend on your concept of construction and materials.

Some points.

Using a speaker as a voice microphone is nothing at all like using a speaker as an acoustic sensor, when the distance between the source and the sensor is unknown. One speaks directly into a microphone. From inches away. Remember sound decays with distance, rapidly.

As the AP 2 said, distance is what kills sound. High frequency sound dies sooner than low frequency sound. Elephants can communicate over miles with a single low frequency grunt.

Microphones use very small diaphragms, with minuscule mass. Not cones, with surrounds, and voice coils that weigh tons compared to microphone components.

I know a few things about speaker design and assembly and sound sampling and acoustic energy propagation.

Cones and rubbing aren't the issue here.

Speakers make quite good microphones but the frequency response is not so good for music. The low output impedance will produce low signal voltage but with all that noise around you should still get several mV during the impact.

I suggest to the OP that you bolt the LS to the static part of the frame that gets hit at the end of the travel and consider it as an accelerometer - forget the sound per se. Band-pass filter the output to suit the spectrum of the impact and then pass the signal to a rectifier and trigger with hysteresis and a suitable time constant.

A general purpose opamp with moderately low noise - even a good old 741 - will do as a preamp if you want to roll your own electronics but you may be able to find a chip designed as an audio amplifier with a 'voicegate' which will provide almost all the circuitry you need. Over to you!

Hi Lyn,

As I read it, the OP is only interested in detecting the end position, or what he refers as the detectable point and if this is right the speaker can be used for it.

As far as talk back systems on board, they use the speaker also as microphone or used to do it until 15 years back. (this is amplified) They are used to communicate with the crewmen on boats e.g. on the bow deck between the pilot or captain and the crew. They work switched bi- directional.(and in bad weather too)

When you connect two speakers, not speaker enclosures with filters etc.. but bare speakers together, and you apply sudden pressure (deep movement and short time), the other speaker will try to follow the movement and produce the plop you make by pushing on the first one. There is some loss due to efficiency in the speaker coils and cone suspension.

When we did P.A, the only microphone that survived the "grosse caisse" (Base drum) was a modified speaker with high impedance - 200- 600 ohms. (or a very expensive AKG)

A microphone is also a speaker when electro-dynamic and is build the according the same concept, indeed with less mass in the membrane and suspension. They are made more sensitive just to catch weaker signals to set the generator (linear) the system in motion

Our company in Belgium did a lot of research for JBL, Electrovoice, and Altec Lansing in the seventies, and had the sole repair facility for these makes and others like Sennheiser. (those days plenty membranes were interchangeable for microphone as well as headphones) We used to modify speakers for wedge monitors that were super anti Larsen - extra short coil that moved out of the magnetic field when overpowered)

Some LVDT transducers work with the same principle. A circular coil moving in the air gap between 2 magnetic poles.

A speaker can be used as a vibrator in subsonic situations. When you connect one to a function generator you will see the movements of the cone while not hearing them. If you do this while talking, you can even generate a unnatural sound of your voice with the doppler effect and air displacement. You actually modulate your speech with the cone displacements.

If the OP finds a reliable way to move that cone and keep it sane, he has a signal to work with.

When I was 10 years old I made a connection between our neighbors with 2 speakers of 3" (250 ohms) and it worked very well. No batteries, no amplifiers involved, just load speaking in it and carefully listening.

Just for the people who might be interested in it.

I have no quarrel with anything you say here. Cops in our town all use a duplex microphones to communicate.

Now, I digress. We used Sennheiser headphones when we began developing our active noise canceling headsets in 1990. Superb quality.

And I also visited the JBL factory in Northridge CA. many times when we partnered with them to build some HUGE speakers we mounted on an M-60 tank to silence the engine noise.

We got to play with the tank in the desert at Edwards Air Force Base for a few days.

Enough of that.

Bye.

This pneumatic switch is available from Clippard. This particular one is set for 65PSI switching. Other pressures are available. If you determine how much pressure is required to achieve full travel, and use a higher pressure switch, it will only trigger when more pressure is detected than is required for full travel,, thus insuring the stroke is completed.

Hi bob,

The opposite of what you suggest, sense the decaying pressure is the norm in pneumatics as a non contact way to detect a cylinders position.

This method is used because the pressure of the air entering a cylinder goes up to system pressure virtually immediately so is not a reliable sensing method.

By setting the switch to about 5 psi and connecting it normally closed, by sensing the exhausting air between the cylinder and the speed control valve, the cylinder will have reached the end of stroke when there is low pressure on the vented side.

As previously mentioned, it is no guarantee that the cylinder has not stuck anywhere in it's stroke but it is still used often in circuits (mostly using pneumatic valves as the sensor)

Best regards,

John

The first thing is to determine if you can "see" the bang.

Start with an oscilloscope trace and see if you can observe the pulse "bang" from the background noise. You may need to include a circuit that opens a window x msec from the solenoid firing to observe the acceptable strike.

If the signal overall is too low, use a simple 741 OA with a differential input with various gains, say 10 or 100, and see if you can now see the "bang".

If you can, and it is a simple peek you see, then a voltage comparator (or opamp connected as such) that determines the peek exceeded your threshold level should do. If that is too quick for your PLC, then have that trigger a pulse from a 555 timer.

Lots of trial and error here.

Stepping outside the box can be rewarding - that is the spirit of invention.

Remember much experimenting is done before viable products are created.

It may work, it may not - but have fun. You are not investing more than a couple weekends of your time.

Send us a copy of your 'scope trace including the solenoid firing signal and the sound feedback trace with the "bang".

Here are the economics of the situation. I have nearly 5000 pneumatic devices I want to listen to for unwanted delays. Some mechanical engineer here believes there is economy in running those air cylinders at about 50 to 60 psi. Consequently, some don't respond so well at that low pressure. And these devices typically are used well beyond their rated life. So, I need to be able to pick out the "non-performers".

Since my speaker type sensor will be in direct contact with the rack upon which groups of 10 devices are attached, I can use a 1 to 10 ratio of sensors to devices. I would never be able to justify 5000/10,000 position sensors and there are not enough left over inputs for that many position sensors. Secondly, the air cylinders do not have magnets so that idea is dead in the water.

Having the speaker attached to the rack is tantamount to using a stethoscope. Using the circuit that I clearly am going to have to build will have to "squelch" out the background noise. By being in direct contact with the individual racks, I expect the pneumatic cylinders to produce on the order of 40 to 50 dB more of a signal than the background.

The current design now looks like a peak detector Op Amp circuit in series with a comparator to act as a "squelch" circuit for background noise. I may have to offer some of these for sale after I prove the design.

AP#2 has nothing to contribute but noise.

It's interesting (and not untypical) that you have completely changed the way the speakers are being used.

This proves to me that you have no idea what you are doing, nor how to explain it or how to make it work.

"Sound Based Feedback System" "The idea is to convert the sound into a pulse"

Furthermore, you obviously don't know how a stethoscope works. A stethoscope is indeed an acoustic device.

So, I'd be much more inclined to take advice from, and even speak to AP 2 than I would you.

<unsubscribe>

As bigg has suggested, use limit switches. That's what they're for.

I've come back to this thread to, once again, for the absolute ban on anonymous posters.

This AP#1 has proven to be childish and rude, by voting any responses he doesn't like as OT.

BAN ANONYMOUS POSTERS.

You are one to talk! How about banning you! Seems to me that you are AP#2!

I somewhat agree with Lynn about anonymous posters. It does seem like there is more junk posting from APs and not so much from those that identify themselves.

That being said, regarding your problem - it sounds (ha!) to me like what you are trying to do is use the speaker as an accelerometer to pick up the vibration caused by cylinders hitting the ends of stroke. I don't know much about the circuit or electronics, having always had EEs (ETS?) around to do that bit, but maybe the change in terminology will spark more useful discussion.

One accelerometer fixed to the rack of 10 cylinders should give a predictable pulse frequency and amplitude per cycle or time period. Establish a tolerance window around those values and when it deviates out of tolerance you have a fault in one of the 10 devices. Its the same concept as some bearing failure devices used in BIG motors and generators.

Good luck, and get a name.

Send pictures when you get this to work.

I'm curious to see how one can accurately sense one pulse in a rack of 10 cylinders.

If, indeed, you don't change your story yet again.

The OP probably wants to write a music opus for cylinders and PLC's. Guess it will be in B mineur.

Do yourself a favor and...

As the cylinders fail, replace them with cylinders with magnets on the pistons and install reed switches as you go. You could even wire the switches in parallel to save PLC inputs.

If they are BANGING as loud as he has indicated, he should consider some cushions also. Of course, if he ever gets his acoustics idea to work, and then they eliminate the BANG, he's back to the drawing board. I can't visualize a machine with groups of 10 cylinders in an enclosure. What do they actuate? Have I missed something?

I also do not understand if one or up to ten cylinders should respond to one event and if sufficient time for resonant acoustic decay exists between events. As I continue to harp on this thread, a careful analysis of what these speakers actually pick up will determine if this is a foolish or frugal approach to a problem.

They are (if I understand the latest version of AP#1's story) mechanically coupled to some type of "rack" that has 10 cylinders mounted, somehow, to it.

These cylinders would bang, not whoosh, when they reach the end of their travel and AP#1 would then divine which ones were deficient, or, in his words " be able to pick out the "non-performers".

So, acoustics have, if I understand the latest version of AP#1's story, no place in the discussion.

If you detect a note of cynicism, you should.

You may also wonder why I would waste my time on this ridiculous folly.

That could be the masochist in me.

This, too will be voted OT by AP#1 soon.

The delusional AP#1 says he has," 5,000 pneumatic devices I want to listen to". #22

Once again, I believe we may be chasing pixies.

But, what do I know? I can't even comprehend his "feedback system".

This will be voted OT by AP#1 soon.

Patience my fellow designers! I will provide oscilloscope pictures when I get my hands on the appropriate amplifier. Unfortunately, I have no budget for testing ideas and frequently I have to promise results before the managers will spend a dime. Consequently, I am doing all of the development out of pocket.

We also have pneumatic devices with magnetic reed switches on the appropriate type of air cylinders on another machine. About every two or three months one of these slips out of position on a machine that has only six sensors and the machine stops working. I consider those type of sensors to be highly unreliable in the long run.

At the moment, I am being distracted with meeting my annual goals and "subjectives" against that irritating whip known as a timeline. So, if you will excuse me for a little while, I will re-start this thread with the screen shot of drive pulse verses the sound pulse. And I will consider revealing my identity.

There are several reasons why people don't wish to reveal their identity. One is due to the sensitive nature of the question because a complete disclosure could help competitors. Another is because a certain topic falls into the category of "outside the usual description of work". In this case, I am attempting to turn an analog signal into a digital signal. There are certainly lots of devices that do just that, albeit this one falls outside the norm. And I too could take the stance that "it can't be done" but I am far too stubborn to do that now since I have been so completely challenged.

Much of the work I do is technically digital. And frankly I resent the negative comments and pokes at the so called "level of experience". Paraphrased from my grandmothers best advice, "If you don't have anything constructive to offer, keep your mouth shut! You just might learn something!"

This is not the first time I've ever been around nay-sayers. I'm quite used to exceeding all of their expectations. Still, I expect nothing, especially acknowledgement. But maybe that is why I have patents and trade secrets credited to my name while the nay-sayers generally do not. And it is not my intention to put those people in their place, no matter how much they deserve it. My patients and trade secrets speak for themselves. Creativity is frequently unappreciated by those who lack the imagination to follow the thought. But fear not, I will not disappoint. I will refresh all of those rusty brain cells in analog analysis and provide the results within the next 60 days. I am committed to prove that this works despite the rude comments. In the long term, I expect to find another way but I really can't resist the temptation to prove the concept.

I have gigagobs of experience with reed switches and cylinders equipped with magnetic pistons used on high speed packaging equipment in the food industry.

As in your application, the reed switches were used to confirm motion of the device activated by the pneumatic cylinder.

If your reed switches slip out of place from time to time, may I suggest to you that they were not, IMHO, installed properly in the first place.

Bimba makes beautiful Hall Effect reed switches (with no moving parts inside them to wear out. They come with 304 grade stainless mounting straps sized for the proper cylinder diameter. When installed properly... a drop of silicone under the reed switch about half the size of a pea will ensure that, once cured, the reed switch will never move out of alignment because of vibration in the machine environment.

While I recognize and understand your determination with respect to this concept. You really cannot get much more of an elegant solution to this issue then magnetic equipped pneumatic cylinders with "properly" installed reed switches.

As to your identify... no one cares. The issues around here with anonymous posters is not that they are anonymous, it is that frequently the things they bring to the forum are of little to no value.

As to the "bean counters" in your organization that hoard dollars and are loath to pass them out... they do so to protect the enterprise from waisting precious financial resources on half baked ill conceived ideas.

Thank you for the recommendation for the better reed switches. I will look into replacements for the cheep versions we were stuck with from an outside machine builder. As to applying it to my nearly 5000 air cylinders, and adding input cards to read them, the cost alone would kill the idea before the ink dried.

As to the "bean counters" and their proof of concept before any work is done, I would expect that as a normal and expected requirement. As to their dedication to saving precious financial resources, I have to laugh. They frequently waste money, especially patting themselves on their back while they take credit for the work performed by their minions. Nothing new there.

You said...

"As to their dedication to saving precious financial resources, I have to laugh. They frequently waste money, especially patting themselves on their back while they take credit for the work performed by their minions. Nothing new there."

LOL... I pretty much agree.

If you decide to use limit switches, you can wire the 10 switches in each bank in series, giving you a total of 500 data lines. You can use each to light an LED. Arrange the 500 LEDs in a rectangular array of whatever dimensions you want. It would provide a visual indication that all limit switches were closing.

Just a suggestion...

So why do you want to post as an A P? Do you know Lyn's real name and profession, or "Old Salt"? Do you think you will be less anonomous if you simply take a street name?

The reason I like a name is because I can be fairly certain that the person behind the name is always the same person, and I can interact with a lot of background. I know that Old Salt, for instance, may have been a sailor.

The other reason for banning Anonomous Posters is to hold trolls (likie AP2) responsible for trolling. Its trolls who ruin forums, and they rarely want their name out there because they would get trolled right back. Or banned. Bad BAD troll.....

So best bet is to simply take a name. It will cause a lot less blather.