Ultrasound for data transmission

[Gallymimu]

Anyone have any experience using ultrasound for data transmission?

In our use case we are transmitting through steel (about 2 inches) into water (anywhere from 6" to 36") back into steel and into the receiver.

Basically we need to get data into a large steel chamber filled with water (using ultrasound).  We are using ultrasound as opposed to other methods for a few reasons which I can't get too deeply into because of IP issues.

We've been looking at using some transducers from American Piezo (APC).  It seems that lower frequencies (10s of KHz) should be better due to lower attenuation in water but I've had a few simulation guys suggest that we should be looking at a few MHz.  

Our biggest challenge so far has been with acoustic matching of the interface materials as well as piezo electrical resonance changing markedly when hard bonded to the steel surfaces.

If interested we've been looking at Item 81 here: http://www.americanpiezo.com/index.php/?option=com_content&view=article&id=128
The piezo is a 25mm(circular) x 2mm(thick) hard piezo with feedback electrodes.  Approximate resonances at 80KHz(diameter resonance) and 1MHz(thickness resonance).

I'm interested in bouncing ideas and thoughts around with anyone who has worked in this space before.

Thanks!

Other info added by request:
Data rate is very very low, bytes per minute would be perfectly fine.

[FriskyFerretReloaded2]

Do you think mentioning some silly little detail like the data rate requirement might be a good idea?

>...which I can't get too deeply into because of IP issues.
That means you ain't working on no university research project but a commercial, for-profit project.

>I'm interested in bouncing ideas and thoughts around with anyone who has worked in this space before.
Bet you are. You're interested in picking our brains for your great gains $$$$$$.

One of your recent posts. Seems contradictory to the starter post in this thread.

[LabVIEWguru]

I helped to do this, but I can't give you too many details. You are correct about the acoustic matching will make or break the project. We used two different types of materials - thin teflon sheets in one instance, and in another they used a "closed cell" material, it was almost like foam but was a plastic material. it felt like the material that was packed with expensive chocolates. They used an electric arc between two wires *just touching* the surface. It disrupted the cells on one surface but left the cells intact on the other surface. The closed cell surface was glued to the transducer and the open cell (or disrupted cell) side was the interface to the air/moisture environment. The foam cell material was in the 20-30 Khz range, the teflon material was used in the mhz range.

The transducers were constructed using ceramic material discs silver coated. They had a polarity. They were about 2 1/2 inches in diameter and had a 1/2 inch hole through the center. A bolt went through the center and was torqued into the transducer face. there were brass shims at the top and bottom of the stack to connect the excitation signal. You had to short the electrodes through a resistor because if you torqued them without dissipating them through a resistor, compressing the piezo elements would literally knock you on your butt.

The completed mechanics were put in a finished housing and a rubbery potting compound was poured in to fill the transducer. The potting compound had to be mixed and immediately pumped down to a vacuum to remove air from the compound before pouring it into the housing. Air in the compound changed the characteristics of the transducer. 

The physics of it was really interesting. My job was to write code for the system and then add code to a completed system to test the transducers.

If your tank is flat, you can probably use that closed-cell foam interface @ 20 - 30 khz. If the tank is round, obviously the face of your transducer will need to be round. The stuff in the mhz range was short range, giving a lot of detail because the pulse width was so short.

I'll see if I can find the name of the foam we used for an interface, APC might even know what it is. Was that the guy that he and his daughter made the transducers and the ceramics?

I'll see what else I can find if I get a chance. It was really fascinating stuff.

Oh! I almost forgot: One really, really big secret was to "ping" the transducer, wait "X" cycles and then place DC across the transducer - this stopped it from ringing. You could then listen for something short range or listen for a specific event farther out. But, that is a secret, so don't tell anyone. You can look up "synthetic aperture" and "binning filter." I was really proud of the "binning filter" until I found out it had been used for years.

[Gallymimu]

<Irrelevant Banter Section>
Frisky, dude, chill, seriously take yer meds, take yer nap or whatever    

Yeah it is commercial.  If that bother you then don't help, (though I'd rather have your input).  I'm happy to help people working on commercial stuff.  Makes no difference to me but if that's your shtick more power to ya.

The only think you stated that has any topic relevance is the query to data rate.  It's very very low, few bytes per minute would be fine, so any multipath, high Q ringing, settling, etc. wouldn't be issues.  And you are right, if a question comes with a fully fleshed out and over specified it does make it easier.  As you know, design parameters which are front of mind on a daily basis are easily left out when communicating to others.  You got me, my bad, should have included it, and probably other parameters.

I also greatly resent your insinuation that I am trying to steal your valuable knowledge to make a ton of money.  Shame on you.  I've done my fair share of helping and advising people on this forum and in real life.  Having picked though my posts looking for "contradiction" you should be well aware of that.

Thanks though Frisky for reminding me what it's like being on the receiving end of a DICK post.  Good reminder to stay polite and helpful even when frustrated with people.

And yeah, I'm happy to share our results and what ends up working but not the end application as I already mentioned.

Don't worry Frisky, I'll still help you when you have a question or don't know something.  No one knows everything.  If you know about this stuff, then we'd love your input, otherwise step off.  

Note: your attitude has been logged!

Now, let's hug it out, cause we're all friends! I love you dude, really I do.

<End Irrelevant Banter Section>


Labviewguru,

Thanks for the thoughts.  I am perplexed by the closed cell foam, as, I would expect that to have an acoustic impedance almost exactly equal to air.  Was it in fact used as a matching layer or was it used as a backing layer behind the transducer?  I've heard of closed cell foam being used as a backing layer before potting the transducer to avoid impacting the resonance on the back side.

The idea of hitting the transducer with DC to stop resonance is pretty clever.  I'm not sure we'll need that since our data rate is low and we can wait for things to ring out.

I dunno if APC was a guy and his daughter.  They are a pretty small company of about 30 people.  I've been working their their VP Engineering.  He's really pretty friendly and always eager to help.  Unfortunately most of his application knowledge is experimental.

One of the things that I've been wondering is how the transducer performance differs when hard coupled (like epoxy) or loosely coupled (like a thin oil or water interface).   We've seen the resonance go all to hell as I mentioned when hard coupled but the electrical resonance looks good and predictable when loosely coupled.  Since we don't have a good test setup I haven't been able to determine if the energy transmission is much different between hard coupled and loosely coupled.

[solutions]

4 inches of steel???

With what little info you've provided, it sounds like the project has some really stupid constraints to me.

Weld a bar between the two walls and the water disappears from the equation, for instance

[LabVIEWguru]

The foam was closed cell on the transducer side and "disrupted cell" on the output side - it was definitely an interface or matching layer.

I remembered something later - when I was a teenager one of the guys I ran with had a transducer or resonator that bolted to the car frame - it was probably 6 inches in diameter and 2 inches thick. You connected the other end to an amplifier and it resonated the car frame.. I always assumed it was electromagnetic. Bernstein - Applebee (anyone remember them) had a similar product in their catalog that you attached to your wall and it "drove" the wall. I was always going to get one to play with but never did.  This was about 1970 - 1972. Maybe your local stereo shop has something off-the-shelf that would solve the problem.

I like Solutions' idea - I think that steel is going to "ring" for a long time. Unless you are under a *lot* of pressure (and I think you are) you could isolate a bar or rod of another material from the tank into the water.

Something else I thought of - I went to the Chem Show in the mid 90's. There was a company there (sorry - I don't remember who) that had a transducer that was full of some silicone-based oil that was going to be the greatest new thing, solve all the typical problems and so forth. I don't recall ever seeing it after that, but maybe it is something you could look for. The silicone oil was the interface between the transducer itself and whatever it was bolted to.

Underwater Acoustics: Analysis, Design and Performance of Sonar
Wiley | 2010-08-10 | ISBN: 0470688750 | 366 pages | PDF | 10.26 Mb

This was The Bible on the subject. I have a PDF & I'll post it (hopefully) later today.


Jean-Paul Marage, Yvon Mori , "Sonars and Underwater Acoustics"
http://uploaded.net/file/xrkdntcx

[Gallymimu]

Thanks for the book suggestion.  These are the primary references we've been using:

Science and Technology of Ultrasonics, Raj, Rajendran, and Palanichamy, Alpha Science 2007
as well as the guide from APC Piezoelectric Ceramics: Principles and Applications, APC International, 2011
Sorry I only have hard copies of the above.
Also have this: Foundations of Biomedical Ultrasound, R. Cobbold, Oxford 2007.  But I'd not recommend it as the ultrasound basics are poorly described and the book (as you'd expect) only covers imaging.  The other two above are great books the Rajendran one having a lot of practical information in it.

Nice idea on the bar of steel.  There is actually a water tube that penetrates into the tank and we were having pretty good luck bouncing the ultrasound down the tube.  There is pretty good internal reflection (>15 degrees from normal) between steel and water.  I can't actually put a substantial amount of anything inside.  Space it pretty limited.  Though this will be used on "new" systems it is also desired to use it as a retrofit.  I'd very much rather use optical with a window in the vessel if we have to make modest changes to it.  Also to clarify, we're not transmitting "thru" but rather "inside" of the vessel.

There was a period of time in the 90s when the "bass shaker's" were popular.  I did consider the idea of just a infrasound vibration or even something like an impulse response hammer ping on the chamber.  We never ended up trying either.

I'll look for the oil.  In our experience (which is limited) we didn't see a lot of difference in behavior with the very thin interface materials, (water, grease gels) as long as they were on the order of many fractions of a wavelength thick.

Did the silicon oil act as a thin interface or was it more substantial in volume/thickness?

Posted on: November 24, 2013, 03:33:28 15:33 - Automerged
BTW, we've found the VNWA-3 low frequency Vector Network Analyzer to be super useful with the piezo characterization, for anyone else who is interested in this stuff.

http://sdr-kits.net/VNWA3_Description.html

[Parmin]

For bytes / minutes transfer rate you can use simple modulated pings.
in steel you have to wait until all the echoes gone before you start another ping.

[Gallymimu]

For bytes / minutes transfer rate you can use simple modulated pings.
in steel you have to wait until all the echoes gone before you start another ping.

Thanks Parmin, that's exactly what we were planning.

Bigger issue is impedance matching, tuning the driving circuitry to the transducer resonance (as the resonance shifts when the transducer is coupled to the chamber) and of course minimizing reflection and attenuation such that we can actually see something on the other transducer.

[Parmin]

I think this has been done before.
Research WW2 submarine sonar arrangements, or sonar geo location apparatus.

Further, I also think you could get away with morse type coding with pauses as your data bits and pings as your stops.
This would give you a bit higher bandwidth.

Unless you furnish us with more info, that is about all we could comment.

Good luck

[Gallymimu]

I think this has been done before.
Research WW2 submarine sonar arrangements, or sonar geo location apparatus.

Further, I also think you could get away with morse type coding with pauses as your data bits and pings as your stops.
This would give you a bit higher bandwidth.

Unless you furnish us with more info, that is about all we could comment.

Good luck

Thanks Parmin,

If you think of anything specific in terms of details needed I can share just about anything related to the dimensions, materials, methods attempted, assumptions made etc.

It's a pretty niche area but about 80% of ultrasound mirrors RF. 

A lot of the sonar and war era stuff differs since the material thicknesses at the hull of the ship are controlled for acoustic impedance matching while in our case we are stuck with some arbitrary thickness of steel which unfortunately is on the order of fractions of a wavelength at the frequencies we are considering).

One thing that would help with the resonance matching issues are use of a self resonant amplifier (utilizing the capacitance of the piezo as part of the tank).  Anyone done that before?  I'd imagine that would compensate for a pretty broad center frequency drift but maybe at the cost of settling time.  I'm not an analog amp or oscillator guy so I'm a little weak here.

[Faros]

@Gallymimu
Sorry for my late reply, I was little busy. you are welcome to ask and I will be pleased to answer whatever I have answers to. I don't mind if you are making money out of provided information, on the contrary it makes me more keen to help. 
I have some experience in using ultrasonic transducers in various applications [medical physiotherapy  @40Khz - 50Watts and 1Mhz – 10watts ,ultrasonic dental scalers @ 40Khz,  ultrasonic range finders @40Khz and 120Khz,  GPR (ground penetrating radars) @ high-low frequencies (depth dependent)]
The ultrasonic transducer has an equivalent circuit, the weighted element in this circuit is the capacitance, the key to ultrasound transducer operation is to drive it into resonance otherwise you will not gain enough power out of it, since such transducer are born to thermally divert from resonance you must use a self resonating analog circuit that will track the change in resonance frequency and readjust accordingly. It is not as hard as it sounds, those circuit are classical oscillators and easy to use provided you follow their rules.
Please define the frequency and power that you need and I will try to share what I have here.

Regards,

[Gallymimu]

Thanks Faros,

I'd be very interested in your suggestions.  So far we've been targeting 1" diameter piezos with a free air resonance near 80KHz.  I think 10W should be sufficient.  We might also try 1MHz again but I think 80KHz would be better for attenuation and desired beam divergence but I'm certainly no expert.