Maker Pro
Maker Pro

UltraSound Transmitter ?

S

Stef Mientki

Jan 1, 1970
0
hi,

I'm thinking of making a simple 3D location circuit, for multiple target
points.
I've the feeling that Ultrasound is the cheepest and a (quiet) reliable
way to perform that task.
I once saw a system, called "graphbar" (almost 15 years ago), that
performed 2D measurements, in a area of 40*40 cm, with a resolution of
about 0.1 mm (I'm not sure about that, but it was quiet remarkable).

One of the problems with ultrasound seems to make a "wideband" pulse, or
in other words creating a puls with a fast rising edge.
The trick the "graphbar" used was to make sparks over electrodes a few
millimetres apart.

Are there any other ways (and don't like sparks) to create fast rising
ultrasound pulses, with enough power ?
Does anyone know some interesting links about this subject ?

thanks,
Stef Mientki
 
R

Robert Lacoste

Jan 1, 1970
0
I's not directly what you are looking for, but I've published some time ago
an aritcle in Circuit Cellar on a "3D scanner" using ultrasounds (July 2001,
Circuit Cellar Ink #132). The idea was to mount a pair of 40KHz tranducers
on two servo motors, and to scan the environnement. A PC software then built
the 3D image. It was working correctly, however the key limitation was the
low angular resolution (may be 5-10°, dépending on the objects). It was
enough for robotics applications, but not for actual imagining. Moreover the
ultrasounds are not really easy to work with, in particular incident objects
could be completely missed (beam reflected anywere but not on the
receiver...).

Friendly yours,
Robert Lacoste - ALCIOM : The mixed signals experts
http://www.alciom.com
 
R

Robert Monsen

Jan 1, 1970
0
Stef Mientki said:
hi,

I'm thinking of making a simple 3D location circuit, for multiple target
points.
I've the feeling that Ultrasound is the cheepest and a (quiet) reliable
way to perform that task.
I once saw a system, called "graphbar" (almost 15 years ago), that
performed 2D measurements, in a area of 40*40 cm, with a resolution of
about 0.1 mm (I'm not sure about that, but it was quiet remarkable).

One of the problems with ultrasound seems to make a "wideband" pulse, or
in other words creating a puls with a fast rising edge.
The trick the "graphbar" used was to make sparks over electrodes a few
millimetres apart.

Are there any other ways (and don't like sparks) to create fast rising
ultrasound pulses, with enough power ?
Does anyone know some interesting links about this subject ?

thanks,
Stef Mientki

I built an ultrasonic ranger system. One way to generate reasonable edges is
to use a motor driver chip. I used an SN754410 running at 12V, which worked
pretty well. Its got a power and logic voltage input, so you can clock it
using TTL levels.

Regards,
Bob Monsen
 
S

Stef Mientki

Jan 1, 1970
0
hi Robert,

maybe not exactly what I'm looking, but it still sounds very interesting.
I tried Circuit Cellar on the web, but your article doesn't seems to be
on line.
Is there a way to view the arcticle somewhere ?

thanks,
Stef Mientki.
 
S

Stef Mientki

Jan 1, 1970
0
I built an ultrasonic ranger system. One way to generate reasonable edges is
to use a motor driver chip.
hi Bob,

That might be a good idea.
What kind of transducers did you use ?
And can you tell anything about the accurarcy ?

thanks
Stef Mientki
I used an SN754410 running at 12V, which worked
 
R

Robert Monsen

Jan 1, 1970
0
Stef Mientki said:
hi Bob,

That might be a good idea.
What kind of transducers did you use ?
And can you tell anything about the accurarcy ?

I used cheapo electronics goldmine 40kHz transducers, about a buck each,
IIRC. It was actually only used to detect motion, so I don't know about the
accuracy. Relative changes were detected by a little opamp circuit. However,
viewing the thing on my oscilloscope, I would trigger on the transmitter,
and display the receiver's waveform. It was easy to see that the relative
phase shifted very rapidly as I moved an object towards and away from the
system. 1127m/s means the wavelength at 40kHz is about 28mm. So, the phase
would shift through 2pi for every 14mm I moved the object.

Your problem, however, is timing accuracy, and being able to generate enough
energy to get a good return. Also, detection of the wave. For actual
ranging, I'm not sure how to detect the edge, unless its so large in
relation to the ambient noise that it wakes up all the dogs in the
neighborhood... Waiting for a PLL to lock takes too many cycles, I think, so
an LM567 is out.

Post the circuit to alt.binaries.schematics.electronic once you get it
working, it would be fun to see it.
 
A

Anthony Fremont

Jan 1, 1970
0
Robert said:
I used cheapo electronics goldmine 40kHz transducers, about a buck
each, IIRC. It was actually only used to detect motion, so I don't
know about the accuracy. Relative changes were detected by a little
opamp circuit. However, viewing the thing on my oscilloscope, I would
trigger on the transmitter, and display the receiver's waveform. It
was easy to see that the relative phase shifted very rapidly as I
moved an object towards and away from the system. 1127m/s means the
wavelength at 40kHz is about 28mm. So, the phase would shift through
2pi for every 14mm I moved the object.

I've recently been tinkering with some 24kHz transducers that I found at
a local surplus shop. Using 5V and driving directly from a PIC, I can
ping the ceiling with good success. ;-) I'm not sure what the max
range is yet as it's still on a breadboard with probes and clip-leads
attached all over it. I'm going to change it so that I have a
transistor in typical grounded emitter between the PIC and the
transducer. I'll use a 5mH inductor in parallel with the transducer.
This serves the purpose of actually allowing the transistor to conduct
since the transducer doesn't pass DC. It will also pump the 5V to about
40V really giving some kick to the transducer.

You can read all about how it should work here:
http://www.ee.latrobe.edu.au/~djc/UltraSonics/Ultrasonics.htm

Obviously not my idea, but hey that's what the internet is for, right?
;-)

BTW, I know what you mean about the phase shifting based upon the
distance. I wasn't expecting to see that, but I figured that the
reflecting sound was canceling
Your problem, however, is timing accuracy, and being able to generate
enough energy to get a good return. Also, detection of the wave. For
actual ranging, I'm not sure how to detect the edge, unless its so
large in relation to the ambient noise that it wakes up all the dogs
in the neighborhood... Waiting for a PLL to lock takes too many
cycles, I think, so an LM567 is out.

I use a 741 op-amp to get some voltage gain from the other transducer
(which is connected to both inputs of the 741 in differential fashion).
That then feeds directly to an LM393 comparator that is biased about
10mV higher than the idling output of the 741. The 393 gives me a nice
5V square wave to feed into the PIC. It seems to work pretty well, but
a more modern replacement for the 741 would do a better job. The Q of
the transducers is so high that they are virtually deaf to sounds that
are off the resonant frequency. False triggering is not near as much of
a problem as you might think, but a well timed loud hand clap close the
transducer can mess things up. As far as accuracy goes, it seems fairly
consistent but there is a definite but likely quite predictable
propagation delay thru the op-amp and comparator. I guess the biggest
problem with this way of doing it is that I can't tell how strong the
incoming reflections are.

Using a 16bit timer incremented every 1uS gives a max range of about
10mtrs, but very high resolution. I guess theoretically that would give
me close to .01" resolution, but I haven't done any real consistancy
testing yet.

michael
 
A

Anthony Fremont

Jan 1, 1970
0
Oops, I should proof read more.

Anthony Fremont said:
BTW, I know what you mean about the phase shifting based upon the
distance. I wasn't expecting to see that, but I figured that the
reflecting sound was canceling

I meant to say that I think the reflected sound cancels the outgoing
sound when it's out of phase. BTW, I send 5 cycles of sound and usually
capture 5-30 reflections depending upon the distance. The tricky part
is figuring out which one to pay attention to. ;-)
 
S

Stef Mientki

Jan 1, 1970
0
Anthony said:
I've recently been tinkering with some 24kHz transducers that I found at
a local surplus shop. Using 5V and driving directly from a PIC, I can
ping the ceiling with good success. ;-) I'm not sure what the max
range is yet as it's still on a breadboard with probes and clip-leads
attached all over it. I'm going to change it so that I have a
transistor in typical grounded emitter between the PIC and the
transducer. I'll use a 5mH inductor in parallel with the transducer.
This serves the purpose of actually allowing the transistor to conduct
since the transducer doesn't pass DC. It will also pump the 5V to about
40V really giving some kick to the transducer.
Looks indeed a quiet good idea.
You can read all about how it should work here:
http://www.ee.latrobe.edu.au/~djc/UltraSonics/Ultrasonics.htm

Obviously not my idea, but hey that's what the internet is for, right?
;-)
Yes !!
I think that's even the reason why the evolution of mankind and animals
differ ;-)
And indeed it's an interesting page !
I use a 741 op-amp to get some voltage gain from the other transducer
(which is connected to both inputs of the 741 in differential fashion).
That then feeds directly to an LM393 comparator that is biased about
10mV higher than the idling output of the 741. The 393 gives me a nice
5V square wave to feed into the PIC. It seems to work pretty well, but
a more modern replacement for the 741 would do a better job. The Q of
the transducers is so high that they are virtually deaf to sounds that
are off the resonant frequency.
That's dilemma I noticed too.
For having high accuracy, you need a very low Q (or in other words a
very high bandwidth, or still in other words a very fast edges).
On the other hand, to suppress unwanted signals you want a small
bandwidth. So I'm not sure what's the optimum.
False triggering is not near as much of
a problem as you might think, but a well timed loud hand clap close the
transducer can mess things up.
Yes, but that kind of "noise" could be simple removed by repeated
measurements.
As far as accuracy goes, it seems fairly
consistent but there is a definite but likely quite predictable
propagation delay thru the op-amp and comparator. I guess the biggest
problem with this way of doing it is that I can't tell how strong the
incoming reflections are.
I don't think you're interested in the strength of the signal, only in
the time delay.
That reminds me of another trick: "time controled gain" (I cann't
remember the correct term. The idea is: after sending an US pulse, you
should gradually the gain of the input amplifier. The longer the sounds
travels, the more it's amplitude drops.
Using a 16bit timer incremented every 1uS gives a max range of about
10mtrs, but very high resolution. I guess theoretically that would give
me close to .01" resolution, but I haven't done any real consistancy
testing yet.
Yes theotically ...
but then you've to measuere temperature and humidity very accurate,
because they influence the speed of sound considerable !
speed = alfa* SQRT ( T / M )
see for instance this page
http://hyperphysics.phy-astr.gsu.edu/hbase/sound/souspe3.html#c1

thanks,
Stef Mientki
 
A

Anthony Fremont

Jan 1, 1970
0
Anthony Fremont wrote:
Looks indeed a quiet good idea.

It's surely simple and effective. What more could one ask for? ;-)
Yes !!
I think that's even the reason why the evolution of mankind and animals
differ ;-)
And indeed it's an interesting page !

I thought you might like it.
That's dilemma I noticed too.
For having high accuracy, you need a very low Q (or in other words a
very high bandwidth, or still in other words a very fast edges).
On the other hand, to suppress unwanted signals you want a small
bandwidth. So I'm not sure what's the optimum.

I don't know how much actual delay is caused by the transducer vs. the
delays in amplifying it. At any rate, I would suspect that they should
be predictable. It seems that maximum resolution would come from using
as high a frequency as possible. I see that the "professional" sensors
use frequencies far in excess of 100kHz.
False triggering is not near as much of
Yes, but that kind of "noise" could be simple removed by repeated
measurements.

This is true.
I don't think you're interested in the strength of the signal, only in
the time delay.
That reminds me of another trick: "time controled gain" (I cann't
remember the correct term. The idea is: after sending an US pulse, you
should gradually the gain of the input amplifier. The longer the sounds
travels, the more it's amplitude drops.

Yes, I've seen this done and it's really the right way to do it IMO.
Polaroid does this IIRC. I just have a simple 1.5mS blanking period to
avoid detecting the transmitted signal directly and to allow the sending
transducer to stop ringing.

In one implementation I saw, the guy had a target placed at a fixed and
known distance from the sending transducer. This afforded him great
accuracy and adaptability to current conditions. He would just ping it
a few times and come up with a constant that was applied to all
following calculations. I would think that he still needed to treat
transducer reaction time and amplifier delays as a seperate constant in
the equation. I could see that temperature would even modify that to
some extent. I guess you just have to figure all these things out and
see just how much of an affect they have on the big picture.
Yes theotically ...
but then you've to measuere temperature and humidity very accurate,
because they influence the speed of sound considerable !
speed = alfa* SQRT ( T / M )
see for instance this page
http://hyperphysics.phy-astr.gsu.edu/hbase/sound/souspe3.html#c1

If I ever even achieved .1" accuracy, it would still be far better than
my needs. I was doing this as the groundwork for some simple robotic
"eyes" to avoid running into things, so the first reflection pulse (that
isn't caused by the receiver directly hearing the sent signal) is all I
needed.
 
Top