-
Categories
-
Platforms
-
Content
I see it can resonate still.. ok.
I was thinking of putting a variable capacitor in series with the transducer to allow some tuning.
My guess is commercial solutions don't tune it. I'm not sure how a transducer resonates compared to a normal capacitor anyway for resonance.
I'm also thinking it is impossible to tune electrically for both transmit and receive.
I was thinking of putting a variable capacitor in series with the transducer to allow some tuning.
My guess is commercial solutions don't tune it. I'm not sure how a transducer resonates compared to a normal capacitor anyway for resonance.
I'm also thinking it is impossible to tune electrically for both transmit and receive.
A piezo transducer is a little more complicated than you might think. It has mechanical and electrical resonances. It is effectively a capacitor in series with an inductor in series with a resistance with a capacitor in parallel with the lot (if I remember correctly). It is possible to ascertain its resonant frequency with a fair degree of accuracy. To do this you need to determine the frequency at which voltage and current are in phase at which point the load becomes resistive. To do this you will need an oscilloscope and a signal generator. Place a small value resistor in series with the piezo then connect the lot to your signal generator. Now connect one channel of your scope between ground and the signal input and the other channel across the resistor. Compare the phase of the 2 signals as you vary the signal generator frequency. When the phases are the same, you have resonance.
Incidentally, you don't need to drive the transducer at 40KHz, a short pulse of 25uS or so will suffice and will yield much shorter distances.
Incidentally, you don't need to drive the transducer at 40KHz, a short pulse of 25uS or so will suffice and will yield much shorter distances.
I may be way off base then.. I was hoping to improve the efficiency of the driver by utilizing resonance. The idea was to recover some of the energy in the capacitance of the transducer each cycle rather than wasting it.
In any case, I have 2 transducers. I plan to start out with one driving and one receiving at first before working toward using a single transducer.
The plan is to drive at 40khz in 20uS pulses once per second. I will adjust frequency to tune it. I believe they have secondary resonance frequencies that are much higher, like 170khz or so, so that might be interesting as well.
I have no oscope or signal generator, but this is so simple I doubt I need one... I do have stm32 which can be turned into a 250khz scope (1.1msps adc)
In any case, I have 2 transducers. I plan to start out with one driving and one receiving at first before working toward using a single transducer.
The plan is to drive at 40khz in 20uS pulses once per second. I will adjust frequency to tune it. I believe they have secondary resonance frequencies that are much higher, like 170khz or so, so that might be interesting as well.
I have no oscope or signal generator, but this is so simple I doubt I need one... I do have stm32 which can be turned into a 250khz scope (1.1msps adc)
I have not read the whole thing, so forgive me if I blunder.
Seems the circuit has seen much attention. Not my forte, so I won't go there.
Don't know much about your application area, but ...
If 40kHz has a period of 25 micro seconds, your 20 micro second pulses are each almost one cycle. The transducer resonance will help, but not have enough time to build up. The resulting pulse will get you a little bandwidth, if you are looking to use it.
20 micro seconds in water at about 1500 m per second is about 30 mm, or at least 15mm blind time if you are monostatic/colocated transmitter &receiver. One second return time gets you about 750 meters, if you don't have much reverberation. Any idea of the transducers' parameters?
If you have any sort of matched filtering in mind, longer pulses get you more energy in the water & hence more SNR out of the matched filter. If you don;t mind the longer blind time, while transmitting. Similarly, more frequent pulsing will get you more energy if you don't need the range & have a way to integrate it.
40kHz is quite low for underwater, will tend to get you range, but be bulky, especially if you want any directivity. Directivity will help with range/SNR. 40kHz is a traditional choice for air. You have some matched for water or a way to couple them?
Those filters you mentioned are usually to gain some SNR. No idea how your link budget looks, as in expected SNR. "The sonar equation" is not so fierce after you stare at it a bit.
Generally as soon as you sample, you alias all noise into the band you are using. If you're planning to use baseband sampling, that is, sample at least twice the highest frequency expected, you'll bring all noise present into this band. So digital filtering will narrow that a bit, but you can't exclude any higher frequency noise, as it is now in-band. If you run the sampler faster, you can then reduce bandwidth digitally, but your gains are limited unless you oversample by quite a lot, eg 3dB per doubling/halving.Multi-rate filtering/decimation can reduce your computational load for the filtering. For many serious applications you just can't do it without some filtering prior to sampling.Badpass may be needed. You could look at bandpass sampling if you're that way inclined.
There have been some dead simple sonar circuits that use an RF receiver chip for the receiver. The RSSI is your output signal, being nicely log already. These are cheap & can be worth looking at.
Sorry, too long ago for me to remember numbers.
If those cleaning transducers are omnidirectional, a lot of your power goes into spherical spreading.
Seems the circuit has seen much attention. Not my forte, so I won't go there.
Don't know much about your application area, but ...
If 40kHz has a period of 25 micro seconds, your 20 micro second pulses are each almost one cycle. The transducer resonance will help, but not have enough time to build up. The resulting pulse will get you a little bandwidth, if you are looking to use it.
20 micro seconds in water at about 1500 m per second is about 30 mm, or at least 15mm blind time if you are monostatic/colocated transmitter &receiver. One second return time gets you about 750 meters, if you don't have much reverberation. Any idea of the transducers' parameters?
If you have any sort of matched filtering in mind, longer pulses get you more energy in the water & hence more SNR out of the matched filter. If you don;t mind the longer blind time, while transmitting. Similarly, more frequent pulsing will get you more energy if you don't need the range & have a way to integrate it.
40kHz is quite low for underwater, will tend to get you range, but be bulky, especially if you want any directivity. Directivity will help with range/SNR. 40kHz is a traditional choice for air. You have some matched for water or a way to couple them?
Those filters you mentioned are usually to gain some SNR. No idea how your link budget looks, as in expected SNR. "The sonar equation" is not so fierce after you stare at it a bit.
Generally as soon as you sample, you alias all noise into the band you are using. If you're planning to use baseband sampling, that is, sample at least twice the highest frequency expected, you'll bring all noise present into this band. So digital filtering will narrow that a bit, but you can't exclude any higher frequency noise, as it is now in-band. If you run the sampler faster, you can then reduce bandwidth digitally, but your gains are limited unless you oversample by quite a lot, eg 3dB per doubling/halving.Multi-rate filtering/decimation can reduce your computational load for the filtering. For many serious applications you just can't do it without some filtering prior to sampling.Badpass may be needed. You could look at bandpass sampling if you're that way inclined.
There have been some dead simple sonar circuits that use an RF receiver chip for the receiver. The RSSI is your output signal, being nicely log already. These are cheap & can be worth looking at.
Sorry, too long ago for me to remember numbers.
If those cleaning transducers are omnidirectional, a lot of your power goes into spherical spreading.
Sorry, this was a mistake. I meant 1ms or 1000us. This should give 40 cycles or so. I think I need 3-4 cycles minimum to get a rise in amplitude. I might get away with less. Of course I can tune it much like the frequency to see what works best
So at 1ms, I should be able to measure 1 meter. I don't care about less than that. I need to measure in the 1 meter to 10 meter range, although deeper returns would be nice.
Yes, well, 50 watt 40khz transducers are about $8. I read somewhere they can operate at 170khz as well. Eventually I will probably try other transducers up to 200khz. I have even read about some systems using 800khz or higher frequencies to get very high detail.
I think you measure time? Without knowing the water temperature and other parameters it will be slightly inaccurate, but this is a start.
The stm32 has a 1.1msps adc. So in theory I could be sampling up to 50x the frequency. Hopefully this will allow all the filtering to be done in software.
I am interested in any circuits, but again, I'm trying to use minimal component count, and make the whole thing as cheap as possible.
Despite spherical spreading, I believe at such low frequency and enough power, they should still work for basic use.
Is it possible to bolt something to them to aim the beam better?
Apologies if I was covering things you'd already covered well. Difficult to know the level to use.
Agreed 50 watts is a good start. Transducers already matched to water, a good start.
50 times oversampling will get you a few dB. The noise accumulation will depend rather obviously on how much noise there is where you want to operate, eg outboard motor noise,
The thing about the single RF IC for the receiver was exactly as you say, low parts count & cheap. I guessed if it was good enough for a low-end commercial product, must be cheap.
The temperature will make precious little difference. For short ranges, far field, you can basically ignore attenuation & look just to spreading & directivity. The variation with temperature & salinity are mostly small.
Also, I guess you have it covered, but the pulse length in normal carrier-wave, pseudo-single frequency systems will determine your range resolution as well as blind time. Not sure whether you want to distinguish multiple targets at different ranges. If you have three transducers, you could look to use two receivers and get some spatial awareness
Applaud the to digital as soon as possible.Sometimes there's more to the DSP side than people anticipate, if they want to see the maximum advantage from it.
Transducers are not my area, another guy. If you want directivity, generally you need aperture size, extent. Acoustic lenses are not impossible. Maybe someone else can help with that. I guess the answer in normal, air, acoustics would be a horn, but can't say I've seen one under water, but my experience is limited. I guess they work by changing impedance. Might mess up your match.
Enjoy.
Agreed 50 watts is a good start. Transducers already matched to water, a good start.
50 times oversampling will get you a few dB. The noise accumulation will depend rather obviously on how much noise there is where you want to operate, eg outboard motor noise,
The thing about the single RF IC for the receiver was exactly as you say, low parts count & cheap. I guessed if it was good enough for a low-end commercial product, must be cheap.
The temperature will make precious little difference. For short ranges, far field, you can basically ignore attenuation & look just to spreading & directivity. The variation with temperature & salinity are mostly small.
Also, I guess you have it covered, but the pulse length in normal carrier-wave, pseudo-single frequency systems will determine your range resolution as well as blind time. Not sure whether you want to distinguish multiple targets at different ranges. If you have three transducers, you could look to use two receivers and get some spatial awareness
Applaud the to digital as soon as possible.Sometimes there's more to the DSP side than people anticipate, if they want to see the maximum advantage from it.
Transducers are not my area, another guy. If you want directivity, generally you need aperture size, extent. Acoustic lenses are not impossible. Maybe someone else can help with that. I guess the answer in normal, air, acoustics would be a horn, but can't say I've seen one under water, but my experience is limited. I guess they work by changing impedance. Might mess up your match.
Enjoy.
Afterthoughts.
Another frequency for transducers may be a different oscillation mode, eg transverse instead of axial. The efficiency or directivity of radiation may be quite different. The resonance will get you amplitude, but will lengthen the ringdown depending on the Q factor.
The transducer, if driven near resonance, will change resonance frequency as soon as the drive is removed, series & parallel resonances & all that. Usually ignored.
The approximate equivalent bandwidth will be one over the pulse length. If you're looking only for amplitude, won't matter.
If you are generating the carrier for excitation & therefore know it's timing exactly, you could sample at a known related frequency.
Another frequency for transducers may be a different oscillation mode, eg transverse instead of axial. The efficiency or directivity of radiation may be quite different. The resonance will get you amplitude, but will lengthen the ringdown depending on the Q factor.
The transducer, if driven near resonance, will change resonance frequency as soon as the drive is removed, series & parallel resonances & all that. Usually ignored.
The approximate equivalent bandwidth will be one over the pulse length. If you're looking only for amplitude, won't matter.
If you are generating the carrier for excitation & therefore know it's timing exactly, you could sample at a known related frequency.
I was trying to calculate the size of toroid needed. I find it needs a cross sectional area of at least a square centimeter which is a large toroid. I may have calculated wrong.
I could change the driver schematic. Instead use a boost converter to generate 150 volts, and charge a 10uF capacitor. Then a single mosfet (or igbt?) could drive the transducer from it. Would this be as efficient? I think it's slightly more complicated, but maybe even cheaper and smaller.
This would allow for a much smaller inductor in the boost converter than the transformer needed. Would it be more efficient or less?
I could change the driver schematic. Instead use a boost converter to generate 150 volts, and charge a 10uF capacitor. Then a single mosfet (or igbt?) could drive the transducer from it. Would this be as efficient? I think it's slightly more complicated, but maybe even cheaper and smaller.
This would allow for a much smaller inductor in the boost converter than the transformer needed. Would it be more efficient or less?
