Post by thread:[EE]: Piezo transducer for the evaluation of food

Hi, For a school project, I want to investigate the use of ultrasonic waves to determine the state of various items of food (eg. fruit). I have done quite extensive research, and determined that I need a piston transducer of the type used in NDT (non destructive testing) of metal components and rails. The problem is this, such units are ludicrously expensive. A ready-made unit is definitely out of my reach. However, having read the authoritative volume on NDT by the Krautkramers, I think a home made version is not out of question, in fact, it should be quite doable, assumming pinpoint accuracy and cutting edge efficiency is not required. Afterall, I plan on measuring attenuation and transit-time in apples and pears, not detecting microscopic cracks inside hunks of steel. Such piston transducers as I have described rely on the mechanical resonant frequency of the piece of PZT (or equivalent piezoelectric material). For a thin disc, this is calculated easily from a simple formula, using a frequency constant for the material and the thickness of the disc. The disc is held in place with a knife-edge ring and dampened by either a Helmholtz chamber (a smaller version of those found in studios) or a backing material (eg. rubber), this is done to even out the transducer's bandwidth (decrease Q). I've managed to acquire free from a large manufacturer as a sample a disc 2mm thick x 20mm dia. Such a disc has a resonant frequency of around 1Mhz. This is a pretty standard frequency for NDT in metals. The catch is, a 1Mhz pulse would not survive very far into such a high-water content, porous material as an apple, let alone hitting the other end and reflecting back. Apparently, 50kHz is more reasonable. Why not build a 50kHz transducer then? The second catch is, a piece of PZT having a natural resonate frequency of 50kHz would have to be 50mm thick(that's two whole inches!), with a diamter greater than that still. OK then, so what? Well, unfortunately this isn't compatible with my concept of economy :-). Furthermore the drive voltage would have to become impossible high. Commercial transducers (read $$$) get around the latter limitation by sandwiching together a stack of plates, electrically wired in parallel. This is not feasible in my case. And now comes the part where any help from you techno gurus would be tremendously appreciated. I have a few options here: 1. Use the freebie PZT plate to build a 1Mhz transducer and use it at that frequency. However, the odds are stacked against it ever being able to work with my target samples, and so is the physics. 2. Use the freebie PZT plate to build transducer that is excited at 50kHz. This essentially means operating the plate in it's 'static mode', since the driving frequency is too far removed from the resonant freq. for mechanical harmonic motion to play any part. I'm trying to work out the relative acoustic output intensity of such an arrangement. I have a feeling that, even without any additional dampening, the wave intensity might be insufficient. 3. Save the freebie for later, try a different alternative. You've all seen piezo buzzers and their u-sonic siblings. They work on a totally different principle to the piston transducer discussed above. The piezo material is usually very thin (~0.3mm). It is mounted on a brass diaphragm. As a voltage is applied across the material, the disc thickens as expected, but this isn't important. The key is, it also expands/contracts sideways (the radius changes). This causes the diaphragm to flex in and out. The actual change in the thickness of the piezo sheet is too small to be measured, but the diaphragm flex enough to blast your ears out. AFAIK, this method is commonplace for air transducers but is not used in surface contact NDT. My guess is that pressing such a TX head against any material (even via a coupling liquid) would quench the delicate flexing osillations, because solid matter compresses so little compared with air. I think in acoustic terms this can be boundary reflection. There is a huge difference between the wave velocities of air and steel (300 m/s vs. 6km/s), hence they respond diffferently to. But what about apples? Perhaps a bender(diaphragm) transducer can be modified to work. However I have found no information on the quantitive design of diaphragm transducers, so I can only make educated guesses (this subject is not-surprisingly not covered in the Krautkramer book) Can you give me some help on how I can go about this dilemma? I believe the electronics to follow would be the easier part, a good challenge but defeatable in due course, with patience. Peter L. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email spam_OUTlistservTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

Peter, see my comments in your text. -- Rich Peter Lau wrote: {Quote hidden}> > Hi, > > For a school project, I want to investigate the use of ultrasonic waves to > determine the state of various items of food (eg. fruit). > > I have done quite extensive research, and determined that I need a piston > transducer of the type used in NDT (non destructive testing) of metal > components and rails. > > The problem is this, such units are ludicrously expensive. A ready-made unit > is definitely out of my reach. However, having read the authoritative volume > on NDT by the Krautkramers, I think a home made version is not out of > question, in fact, it should be quite doable, assumming pinpoint accuracy > and cutting edge efficiency is not required. Afterall, I plan on measuring > attenuation and transit-time in apples and pears, not detecting microscopic > cracks inside hunks of steel. > > Such piston transducers as I have described rely on the mechanical resonant > frequency of the piece of PZT (or equivalent piezoelectric material). For a > thin disc, this is calculated easily from a simple formula, using a > frequency constant for the material and the thickness of the disc. The disc > is held in place with a knife-edge ring and dampened by either a Helmholtz > chamber (a smaller version of those found in studios) or a backing material > (eg. rubber), this is done to even out the transducer's bandwidth (decrease > Q). I don't know anything about the piston type of transducer so bear that in mind when I make my other comments. > > I've managed to acquire free from a large manufacturer as a sample a disc > 2mm thick x 20mm dia. Such a disc has a resonant frequency of around 1Mhz. > This is a pretty standard frequency for NDT in metals. > > The catch is, a 1Mhz pulse would not survive very far into such a high-water > content, porous material as an apple, let alone hitting the other end and > reflecting back. The water content is not a problem, in reality it helps. The speed of sound is about 1500 meters/second in water. This gives a wavelength of about 1500m/s / 1MHz = 1.5mm. The resolution will always be worse than the wavelength. Maybe 5mm resolution is possible with some care in design. Sound is attenuated at about 1db/cm/MHz. As I remember, this is round-trip attenuation. For an apple 10cm in diameter, that would give 1db * 10cm * 1MHz = 10db of attenuation. Even if I am wrong, 20 db of attenuation is not a problem with a low noise amplifier. Since the amplifier can be narrow band, the noise should not be a problem. I don't know how much the porous nature of the apple will affect the attenuation. My gut feel is that the porousness is not a problem. >Apparently, 50kHz is more reasonable. Why not build a 50kHz > transducer then? The second catch is, a piece of PZT having a natural > resonate frequency of 50kHz would have to be 50mm thick(that's two whole > inches!), with a diamter greater than that still. OK then, so what? > > Well, unfortunately this isn't compatible with my concept of economy :-). I think that the wavelength at 50KHz (about 30mm) is WAY too long. In fact you may need to have more like 10MHz to get the resolution to see fine details of the fruit's interior. > Furthermore the drive voltage would have to become impossible high. > Commercial transducers (read $$$) get around the latter limitation by > sandwiching together a stack of plates, electrically wired in parallel. This > is not feasible in my case. Since the attenuation is only 10db to 20db I don't see that you need very high drive voltages. A few volts should be more than enough. > > And now comes the part where any help from you techno gurus would be > tremendously appreciated. > I have a few options here: > > 1. Use the freebie PZT plate to build a 1Mhz transducer and use it at > that frequency. However, the odds are stacked against it ever being able to > work with my target samples, and so is the physics. I think that you must start with this transducer. You may want to try your hand at making your own high frequency transducer with the Kynar film materials at a later time. Another possibility is to find some old or surplus diagnostic ultrasound transducers. > > 2. Use the freebie PZT plate to build transducer that is excited at > 50kHz. This essentially means operating the plate in it's 'static mode', > since the driving frequency is too far removed from the resonant freq. for > mechanical harmonic motion to play any part. I'm trying to work out the > relative acoustic output intensity of such an arrangement. I have a feeling > that, even without any additional dampening, the wave intensity might be > insufficient. See comments about wavelength. > > 3. Save the freebie for later, try a different alternative. You've all > seen piezo buzzers and their u-sonic siblings. They work on a totally > different principle to the piston transducer discussed above. The piezo > material is usually very thin (~0.3mm). It is mounted on a brass diaphragm. > As a voltage is applied across the material, the disc thickens as expected, > but this isn't important. I believe that the thickening _is_ the sound producing mechanism that you _do_ want. > The key is, it also expands/contracts sideways > (the radius changes). This causes the diaphragm to flex in and out. The > actual change in the thickness of the piezo sheet is too small to be > measured, but the diaphragm flex enough to blast your ears out. > > AFAIK, this method is commonplace for air transducers but is not used in > surface contact NDT. With a transducer that changes in thickness you will couple from the flat surface into the material being tested. >My guess is that pressing such a TX head against any > material (even via a coupling liquid) would quench the delicate flexing > osillations, because solid matter compresses so little compared with air. Noncompressablity means that the sound will be transmitted better! To get good transmission of the high vibrations you will have to have a fluid filling the air gap from the transducer face to the surface of the fruit. Putting both the transducer and the piece of fruit in a fluid is one way of doing this. Another advantage of having the transducer and fruit in a tank of water is that the transducer can be father away from the fruit to put the fruit at the focal point of the transducer. This allows the transducer to be larger than the fruit. A large transducer can have a tighter focus than a smaller transducer. {Quote hidden}>I > think in acoustic terms this can be boundary reflection. There is a huge > difference between the wave velocities of air and steel (300 m/s vs. 6km/s), > hence they respond diffferently to. But what about apples? Perhaps a > bender(diaphragm) transducer can be modified to work. However I have found > no information on the quantitive design of diaphragm transducers, so I can > only make educated guesses (this subject is not-surprisingly not covered in > the Krautkramer book) > > Can you give me some help on how I can go about this dilemma? I believe the > electronics to follow would be the easier part, a good challenge but > defeatable in due course, with patience. Good luck with your project. It sounds like fun. > > Peter L. > > -- > http://www.piclist.com#nomail Going offline? Don't AutoReply us! > email .....listservKILLspam.....mitvma.mit.edu with SET PICList DIGEST in the body -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email EraseMElistservspam_OUTTakeThisOuTmitvma.mit.edu with SET PICList DIGEST in the body

> For a school project, I want to investigate the use of ultrasonic waves to > determine the state of various items of food (eg. fruit). Sounds like you've done your homework. It also sounds like you now know a lot more about ultrasonic transducers than I do, so I can't help there. However, what about hydrophones? I don't know if they go to the frequency range you want, but they are certainly intended to be coupled to water. You probably don't want the ones the Navy uses. At $1,000 per toilet seat, I can't even imagine what the taxpayers are shelling out for a hydrophone. There must be consumer applications that use hydrophones. Cheap fish finders, maybe? Maybe you can find something useable as surplus. I imagine you've looked into this, but I'm surprised you need 50KHz to measure an apple. I would have thought audible frequencies would be sufficient. Why not emit a step or impluse, then look at the frequency content of what comes back or goes thru. I bet you could tell a lot by looking at the absorption spectrum up to 20KHz. When I was little, my father taught me how to tell what was inside a watermellon without hurting it. You hold one end up to your ear, then tap on the other end with your fingernail. With a little practise you can tell how ripe it is and whether it has cavities in the middle. People look at you funny when you do this in the store, but I've never bought a bad watermellon yet. My sense is that all the information is well under 5KHz in this case. ***************************************************************** Olin Lathrop, embedded systems consultant in Devens Massachusetts (978) 772-3129, olinspam_OUTembedinc.com, http://www.embedinc.com -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email @spam@listservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

Peter: I may have jumped too quick. :-) I have worked with ultrasonic imaging so I just assumed that you would want enough resolution to get an image. I agree that you may get useful information at lower frequencies. I will be interested in the results of your investigation. If my attenuation formula applies to apples then you may not get very much loss at 50KHz. It sure is hard to resist grabbing an apple and a transducer (or 2) and trying it myself... -- Rich Peter Lau wrote: {Quote hidden}> > It is true that 50hHz can't produce any resolution. But that's exactly why I > thought It'd be suitable. I want to measure the propagation speed of the > compression wave through the apple, and the attenuation of the signal. > Whilst 1Mz will be reflected strongly by the fine details of the apple's > interior (with a wavelength of 0.3mm), a 50kHz wave would pass unaffected. > > On the other hand, if, as you suspect, a 1Mhz pulse can make it thorugh the > apple without being entirely absorbed, I might be able to take advantage of > the 1Mhz resolution and make analytical use of those 'spurious' reflections. > > > I believe that the thickening _is_ the sound producing mechanism that > > you _do_ want. > > > > This is correct. Change in the thickness of the PZT plate is exactly how > piston type transducers work. They usually have plates around 2mm thick. > Just imagine a flat cylinder getting thicker or thinner as the voltage > across the top and bottom is varied. This generates the compression > (longitudinal) waves which is coupled into the medium for testing. > > Note that with a plate disc polarised top to bottom (the case most of the > time), there is also a _minor_ change in the radius as the voltage is > varied, that takes place concurrently with the thickness change mentioned > above. This is usually neglected in piston type transducers. > > The bender types (e.g. common audible piezo),however, have a PZT layer so > thin (0.2mm vs. 2-5mm common for piston types) that the thickness variation > becomes insignificant. The radial expansion, is not affected by the > thickness of the disc. When a voltage is applied, the disc expands in > radius. It is glued to a brass backing which obvious isn't piezoelectric and > doesn't follow in the expansion. So the backing starts to warp. These work > the same way as do bimorph strips, only that the bending is quasi-spherical > rather than linear. > > These modes of US generation are incompatible. The element of a piston > transducer obviously doesn't flex very much, conversely, thickness mode > oscillations cannot be utilised in a bender element (eg. if the brass plate > is held rigidly still, the thickness oscillation alone would be > infinitesimally weak, the glue might give way. > > My strategy would probably be to try a bit of everything in hope of coming > across a variation that works. > > Thanks to all those who have responded, I look forward to more input. > > Peter L. -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email KILLspamlistservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

>For a school project, I want to investigate the use of ultrasonic waves to >determine the state of various items of food (eg. fruit). I want to use ultrasound to explode pumpkins 8^) >The catch is, a 1Mhz pulse would not survive very far into such a high-water >content, porous material as an apple, let alone hitting the other end and >reflecting back. Apparently, 50kHz is more reasonable. Why not build a 50kHz >transducer then? The second catch is, a piece of PZT having a natural >resonate frequency of 50kHz would have to be 50mm thick(that's two whole >inches!), with a diamter greater than that still. OK then, so what? Why not use a cheap piezo disk vibrating in the shear mode? If you look up acoustics on the web, their are pages that show the drum-head or disk mode. The best I found is Maple's animations of the Bessel functions. If you buzz a piezo disk, you will find multiple resonances of these various modes. Scratch or electrochemicaly etch off the disk's plating to drive these mode more effectively, and you should be able to get up to a watt out of the disk. You can find a powerfull mode up around 30 KHz on a 2 KHz piezo disk. >3. Save the freebie for later, try a different alternative. You've all >seen piezo buzzers and their u-sonic siblings. They work on a totally >different principle to the piston transducer discussed above. The piezo >material is usually very thin (~0.3mm). It is mounted on a brass diaphragm. >As a voltage is applied across the material, the disc thickens as expected, >but this isn't important. The key is, it also expands/contracts sideways >(the radius changes). This causes the diaphragm to flex in and out. The >actual change in the thickness of the piezo sheet is too small to be >measured, but the diaphragm flex enough to blast your ears out. If you use a laser you can well see it. Or you can sprinkle a dust and drive it with a few watts to image the mode. > My guess is that pressing such a TX head against any >material (even via a coupling liquid) would quench the delicate flexing >osillations, because solid matter compresses so little compared with air. Solid is much better than air- you will have a much better acoustic match. Are you going to try to measure the complex acoustic impedance of food, or just attenuation? You might need to examine over a broad bandwidth. Kynar film would be great for that, I doubt much power would be required. I strongly recommend a kilo-joule shock to disintegrate the vegitable under test, then spatter-pattern analysis 8^) **************************************************************************** Freedom is pursuing your carrot, not running from somebody's stick Does society make you enthusiastic, or fearful? The mob rules only what its members achieve. **************************************************************************** -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu