Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Ultrasonic Analyzer

Status
Not open for further replies.

dknguyen

Well-Known Member
Most Helpful Member
Hi.

I was thinking about building an ultrasonic sinusoidal "chirping rangefinder"/"doppler speed-measure-thingy".

All of the circuits I have found drive the transducer with a square wave which won't allow nice tone generation for chirping or doppler measurement. I have a feeling this is more akin to audio drivers which I don't know much about.

I was thinking about driving the transducer (electrostatic/capacitive) with a
DSP->DAC->amplifier->matching transformer->transducer
Is this the right way to go about the driver? I heard some mention of "biasing" capacitive transducers, although I don't know why you would need to bias a speaker...wouldn't that just limit it's dynamic range?

For the receiver, I was thinking
transducer->transformer->amplifier->ADC->DSP

And then having the DSP do some fourier analysis on the whole thing. Am I on the right track with this?

Since I want an sample rate of 16x the max frequency (60kHz), with 8-bit resolution and the maximum wait time is 0.04s, I need about 40k of memory to store all the numbers while the DSP runs through the signal processing. Not sure how long it would take for a DSPIC. To do a fourier analysis of 40,000 data points.

Thanks.
 
Last edited:
Gosh, no need to re-invent the wheel hehe

Check this ultrasonic range finder out

https://www.sparkfun.com/commerce/product_info.php?products_id=639


Its features:
# 0 to 255 inches (0 to 6.45m) in 1 inch increments with little or no dead zone!
# 42kHz Ultrasonic sensor
# New LV version operates from 2.5-5.5V
# Low 2mA supply current
# 20Hz reading rate
# RS232 Serial Output - 9600bps
# Analog Output - 10mV/inch
# PWM Output - 147uS/inch
 
I know. I've used all those Devantech-type sensors and all they do is let you find the average distance of the objects in front of you. THis is meant to be more a substitute for a laser imager or laser rangefinder (think bat sonar). The simple sonars don't let you do things things multi-frequency or multi-receiver tricks s like chirping, doppler shifts, phase delays, or pinpointing where things are a bit more.

With something like sonar where long-range operation implies wide beamwidth, differntiating the range readings to get velocity does not work to determine if something is coming your way (to get out of the way) since the readings are not necessarily coming from the same object (unless it's a really big object). Kind of need dopppler for that. Phase delays let you better know the direction where each distance measurement is coming from and the chirping let's you better pick up your signal in wind or to see softer/harder objects than a single frequency would let you. I kind of need something reinvented- a simple pulse to indicate detection of a return signal just won't suffice.
 
Last edited:
Dunno for certain, but if I had to design an electrostatic membrane speaker, I'd think that I might have a bias voltage in order to tension the membrane.

A couple years back I ordered some some of those 1.5" diamater kodak-camera style ultrasonic transducers, but I don't remember where I got them from. It might have been directly from the manufacturer.

James

Correction: POLAROID cameras. Geez, brainfart.
 
Last edited:
Hmm, that's a good point. I always assumed the membrane was already mechanically tensioned by default.

I initially said I was going to generate the ultrasonic chirp with a DAC- but my hope was to have an 8-bit parallel input DAC with an update rate that was 16x the frequency of the highest pulse. That would have meant 1/(60000x16) = 1.04us settling time. All the DACs I have found that have that settling time are serial- and although they react that fast enough to changes, they cannot accept data fast enough to define 16 points per cycle in a 60kHz sinusoid.

So...I was looking at what some other people had mentioned earlier about filtering square waves. For some reason I crossed this off my list early on since I didn't think I could get a multi-frequency bandpass filter that was accurate enough...but then I remembered about switched-cap filters. And it also seems that I do not need a bandpass filter either, just a lowpass filter right? Since there are no subharmonics in a square wave signal. Actually, I don't think I even need a switched cap filter. If my sinusoid is only ever going to be 40kHz-60kHz, I could just get a lowpass filter that has sufficient attenuation above 120kHz, and any square wave sent into the filter would exit as a sinusoid with the fundamental of the input, right?

The thing I am having the most trouble with now is figuring out hwo to bias the transducer. it recommends a max AC signal of 200Vp and a DC bias of 200V. I can see how to get 200VAC from regular logic with a transformer (assuming I can find a high frequency one that works 40kHz-60kHz). But I don't understand where you would get 200VDC from, and apply it to the transducer, especially if there was a transformer also connected to the transducer leads (wouldn't that destroy the transformer?)
 
Last edited:
Why do you need such an insanly high sample rate? I thought twice the frequency was normally high enough.
 
Hero999 said:
Why do you need such an insanly high sample rate? I thought twice the frequency was normally high enough.

It's pretty hard to find an ADC with a 120kHz sample rate. For 8-bits, might as well go with 1MHz. A lot of this is just me sitting down at a table with and drawing a single sine wave, and then dividing it up into dots until I can obviously tell that it's a since wave, and then some- 16 dots for me.

Also, the fourier spectra is a lot cleaner. Two samples per cycle = a triangle wave for the highest frequency. Lots of other frequency components that might affect the cleanliness of the spectra- harder to pick out what is due to the chirp, doppler, noise, or just sampling distortion.
 
Last edited:
Get one of these (or one of it's relatives) and be done with it. Built in SIN rom and voltage output, and frequency sweeping support:
**broken link removed**
 
Perfect! Now I have to find a way to provide this 200V DC bias and a transformer that works at 40-60kHz to step up this signal. What section of Analog's website was this IC under? I couldn't find anything like it the last time I searched.
 
dknguyen said:
Perfect! Now I have to find a way to provide this 200V DC bias and a transformer that works at 40-60kHz to step up this signal. What section of Analog's website was this IC under? I couldn't find anything like it the last time I searched.

What sort of transducer are you using?, I've not heard of high bias voltages and transformer drive for decades - how far are you expecting it to work?.
 
https://www.electro-tech-online.com/custompdfs/2007/03/60020environmental20spec.pdf

I was quite suprised when I saw the voltage requirements myself- seeing as how everythingn I've seen before that has been piezo rather than electrostatic. But it must be possible, since SensComp has "Smart Transducers" where all the circuitry is crammed onto a PCB that sits on the back of the transducer.

I was hoping to get a range of 6m- this was typical with the piezo sensors I've been using. But these transducers seem to be able to work out to 10m since they are more sensitive than piezos.

This diagram seems to be representative of how to DC bias the transducers-
**broken link removed**

Seems like I might need a 5V to 200V DC-DC converter not to mention transformers which I have been unable to find (SensComp also has some but they have so few specs I can't figure out how to use them). BUt the Smart Transducers don't have anything as big as DC-DC converter on them, so I'm wondering how they did it!
 

Attachments

  • wiring.gif
    wiring.gif
    7.1 KB · Views: 257
Last edited:
When I said that I "use piezo transducers", I meant I used them, but didn't build the circuitry behind them. This is a first time thing for me. It's not an augment onto the ones I have, it's a completely new one that's going to be built (hopefully more of a spectral analyzer for the echo, than a timer).

With the piezos, they work at 20V, and all you really have to do is apply a square wave to them (or a sinusoidal signal) and they ring at resonance. But I'm not using those.

With these electrostatic transducers that I am using, they require some much higher bias voltages which I am having trouble dealing with.

Right now, this is the leading contender for my final year project next year, but it's be nice if I had something to use on my robot before then.
 
Last edited:
I didn't think piezo required a bias voltage. That looks more loke a condenser mic working in reverse.
 
These are electrostatic <condensor-type> transducers. If it was a piezo, it would be easy. No bias needed and only low voltage AC is required. But they have non-existent bandwidth and crummy sensitivity.

I don't think the bias is a huge problem now (though I still don't know how the guys at SensComp got it so small). A really low power 200V DC-DC converter should work since it's really just a capacitor so there aren't any real current requirements apart from charging up the transducer.

The 60kHz transformer is still a problem though.
 
Last edited:
dknguyen said:
Perfect! Now I have to find a way to provide this 200V DC bias and a transformer that works at 40-60kHz to step up this signal. What section of Analog's website was this IC under? I couldn't find anything like it the last time I searched.
RF/IF components, DDS chips
 
Last edited:
Direct Digital Synth. The main idea is to have an overflowing counter which increments at some user defined rate, and the counter is fed directly into the ROM/DAC. The main distinction is that there are no analog PLL's in the signal path. (Although these parts do use a PLL to multiply the frequency up). The advantage is that you can have instantaneous frequency/phase changes without having to monkey with other control loops.
 
No wonder I've had a hard tiem finding ultrasonic transformers. I always assumed that audio transformers cut off at a bit above 20kHz. Most RF ones that can step up any amount of voltage start around 100kHz, or have really bad attenuation at 50kHz. Turns out that audio transformers easily go up to the 100kHz range...I guess that's just a result of physics. Neat.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top