MaxSonar-EZ1 (a great ultrasonic sensor) is featured in three magazines!

I'm always interested in ultrasonics,
and I've no doubt this is a good sensor,
but I don't believe all that's written in the FAQ ;-)
=== quote
The signal from the transducer is amplified by a
bandpass-filter/amplifier, followed by another bandpass
filter/log-amplifier, followed by an integrator with integrated gain,
followed by an analog to digital converter integrated in a
microcontroller. This system yields continuously variable gain from less
one to over 1000.  The microcontroller continuously performs signal
processing techniques to extract the distance.  Some users have asked
for finer resolution, but the time between readings is used by the
software to evaluate the signal.  This type of circuit performs much
better than simple comparator thresholds, single gain, or switched gain
type circuits.
=== end quote
I don't see a log amplifier in the circuit,
and can't see how the gain is controlled in a range of less than 1 .. 1000,
...
so if the "signal processing techniques",
and the rest of the FAQ is equally true ....

cheers,
Stef

Stef,

Thanks for your comments.

This is a completely new design and it may be hard for you to
understand.  Please enter the schematic into PSpice.  (The analog
section fits into the student version of Pspice.)  You will see a very
good log amplifier.  This sensor does work as advertised.  In addition,
some gain is digital and some is analog and so the sensor does function
as stated in the FAQ.  

Best regards,

Bob Gross
www.maxbotix.com

hi Bob,

Bob (Robot Wars Thumper 1997) wrote:

Sorry I doubt about your claims,
at as said, I believe it's a good sensor,
but I really like to understand it.
And the issue I really like is the 1 sensor approach !

   Please enter the schematic into PSpice.  (The analog

I looked into the schematic, but I can't find the logarithmic section.
Could you point me where it's located ?
Is it some non-linear (ff course it must be non-linear ;-) effect which
I can't discover ?

Another point I don't understand,
if it's so good,
why can't you measure with greater resolution than 1 inch ?

cheers,
Stef

Stef,
I am not sure what you are asking here.
If you put the schematic into PSpice the
log amplification will be very apparent.
The circuit will fit into the free student
version of PSpice.  Here you can see
for yourself the operation and how it
works.

Have you tried the sensor?  The precision
is what most people describe as incredible
especially if they have used the competitor's
sensors where readings are expected to
jump around.  The MaxSonar-EZ1 sensor
will output the same number until the object
you are sensing moves closer or farther from
the front sensor face.  This is a big deal, to
get real repeatable data, at full speed and
not having to deal with sensor non repeatability.
In addition, the beam width is controlled by the
variable gain, so this yields a very controlled
narrow beam while, still maintaining the very
close and long range operation.

People have really liked this sensor.  I have had
virtually no complaints about the operation, but
many many kudos!  Many people have especially
liked the very small size (the PCB is smaller than
"the worlds smallest"), and the user interface, the
narrow beam, the low power...

http://www.maxbotix.com/MaxSonar-EZ1___Others_Say_.php

www.maxbotix.com

I've to make my excuses,...
... I studied the schematics again,...
... and I overlooked the diodes in the feedback loop,...
... so indeed it's a real logaritmic amplifier.

Ever thought of increasing the accuracy ?
By accident I just read an article, which claims 0.05 mm accuracy.
They use FSK and use a simple 1 bit correlator.
I'm hoping to try this in the next couple of weeks,
using normal correlation of the measured zero crossings.
I'll keep you informed, might be an idea for your next version
(I'm just an interested hobbyist, no intentions at all to sell anything ;-)

cheers,
Stef Mientki

Awesome -- good to hear it, so are you looking at turning a profit (get
out of the $ hole) sooner than you were expecting? I think I remember
you were (from your estimates) looking at it taking at least a year if
all went well..

meanwhile I was using your sensors on a ground vehicle project (i made a
pic based multiplexer for the sonar: reconnsworld.com/serial_mux.html)
but then I graduated and the ground vehicle project (this year's team)
fell apart. oh well, it happens

-andrew
reconnsworld.com

Bob (Robot Wars Thumper 1997) wrote:

I really do *love* this sensor, except for one major problem and
one minor gripe:

Problem: at power-up the sensors should be at least 14 inches (I
may have this value slightly wrong, but it's in the neiborhood)
from any solid object. This is difficult for me to guarentee.

Gripe: I *hate* the inverse TTL-level serial. This is only minor
because I can use both pulse-width and analog voltage to read
the sensor.

And I also hope that you turn a profit as quickly as possible.

Andrew wrote:

--
D. Jay Newman           ! Author of:
jay@sprucegrove.com     ! _Linux Robotics: Programming Smarter Robots_
http://enerd.ws/robots/  ! "Heros aren't born, they're cornered."

I appreciate your comments.

The problem (having to keep objects away from the front face during
power up) you discuss was done to solve another problem, this, being
able to detect within the so called "dead zone".  Other single
sensors do not provide the ability to detect to the front face at all.
This is not really a problem if you don't want to detect closer than
6 inches, yet if you do, you can.

The serial output has been used by many individuals to evaluate this
sensor.  The ones with "stamp" type micro-controllers and in the
home automation area have really liked the serial output from the
sensor.  It has also really helped MaxSonar-EZ1 sales to OEMs and
distributors as they are able to easily evaluate the sensor by directly
connecting directly to a PC and using hyper terminal.   I also feel
that teachers will be able to demonstrate the sensor to students if
they would like.  Hyper terminal allows a very large font so one is
able to see the range at a distance.  Sorry that you would like the
other polarity.  I hope to accommodate your serial request sometime in
the future.  Right now we are growing very fast and this growth is the
area we have focused.

Regards,

Bob Gross
CEO MaxBotix Inc.
www.maxbotix.com

D. Jay Newman wrote:

For people on this list who are not familiar with sonar modules,
there are two broad classifications -- single sensor and
double sensor.  A single sensor module has a single ultrasonic
transducer that acts as both a transmitter and a receiver.
The old Polarioid sensors and the MaxSonarEZ-1 are examples
of the single sensor modules.  A double sensor double sensor
module has a dedicated transmitter module and a dedicated
receiver module.  Example double sensor modules are the
SRF04 and the Ping.  The single sensor modules have issues
associated with measuring objects up close, since they can
not transmit and receive at the same time.  The double sensor
modules do not have this particular issue.  The double sensor
modules tend to be larger since they have two transducers.


I, too, think the other polarity is preferable, but it is
ultimately your design choice.

Even if you reversed the polarity on the MaxSonar-EZ1, I
personally would not use the serial output, since it only
resolves distances to whole integer units of inches, no
fractions.  So for the increased resolution I desire, it
is necessary to use either the analog or pulse width interface
on the MaxSonar-EZ1.

----

Over the past several years the amateur roboticist has gone
from a lousy selection of ultrasonic sensors (basically just
the Polaroid sensor) to multiple sensors at competitive prices.
I hope the trend continues and I look forward to the next
products from MaxBotix.

-Wayne

[snip]

Yoy mention
"The single sensor modules have issues associated with measuring
objects up close, since they can not transmit and receive at the same
time.  The double sensor modules do not have this particular issue."

Measuring objects up close with dual sensor models
The dual sensor models, as one gets closer and closer to the front
surface, are forced to have a blind spot and dead zone because the sent
signal is not lined up with the receiver so refection does not make it
back so they also have a "dead zone".  The dual sensor models
become completely blind when an object is too close (and yes this can
be as close as 1cm to 3 cm), but when this happens the sensor is blind.


Measuring objects up close with single sensor models
While it is true that single sensor modules cannot transmit and receive
at the same time, they really do not have to.  They first transmit, and
then receive.  But there is the problem with ringdown (where you have
just deposited a great deal of energy on the sensor, and where you must
also be able to detect on top of that ringdown, a small return signal)
and how to deal with that.   What I found was that the ringdown pattern
is not consistent, day to day.  So although the MaxSonar-EZ1 can
measure within the presence of  the ringdown pattern, the ringdown, is
not consistent.  So the MaxSonar-EZ1 sensor records this pattern and
then detects changes to this pattern.  This way the MaxSonar-EZ1 can
detect all the way up to the front sensor face, even objects pressing
against the front face are detected.  This comes with the price of
needing to know what the normal ringdown pattern is, hence the required
calibration.

Bob Gross
CEO MaxBotix Inc.

Wayne C. Gramlich wrote:

That's a nice theory, but in practice the two dual sensor
modules I am familiar with (the SRF04 and the Ping) can sense
objects all the way to down to 0.  I've never had one give
me an infinity reading when an object was on top of it.
Does anybody else want to chime in with their SRF04 and/or
Ping experience?


Whatever you do, it works.

However, as Jay mentions, it is not always possible to ensure
that there is a blank space in front of the sensor on power up
(e.g. the robot parked next to a wall with the sensor pointed
at the wall.)

-Wayne

[snip]

The MaxSonar-EZ1 sensor may not exist today if my daughters robot was
not stuck on a wall, the SRF04 was blind, becasue it was too close.  My
initial experiments showed that this particular sensor would not detect
closer than about 1 cm.  I am going to repeat these tests and
eventually publish, but you will find that when you use the SRF04
sensor, it will stop detecting when you get too close.  If pressed
against the wall (or your hand)  the signal exiting the sender cannot
make it to the receiver.  Just physics.

Two cents...

My experience differs from yours.  I have never received
an infinity reading when an SRF04 is against the wall.
Likewise from a Ping.

With regards to physics, your average wall is not neither
a perfect ultrasonic energy absorber nor reflector.  Some
energy gets absorbed and rereadiated.  This is why you can
hear people on the other sides of walls.  Basic physics.

-Wayne

Again, these sensors even state in their own web sites, 1 to 3 cm for
the SRF04, and about ½ cm for the Ping.  I have measured both of these
sensors, many times, during my comparison tests.  I have both sensor
types, as well as the larger 2" round ones, but I guess you have
different experiences.

Bob Gross
CEO MaxBotix Inc.



Wayne C. Gramlich wrote:

The spec. sheets for both the SRF04 and the Ping specify
the range at which they can take accurate measurements.
They do not specify the behavior when out of range.

While the SRF04 spec. sheet does not specify its out of
range behavior at less than 3 cm, here is what Gerry Coe
(the person who designed the SRF04) says in the SRF04
technical documentation at the following web URL:

   <http://www.robot-electronics.co.uk/htm/srf04tech.htm>

To quote:

    It [the SRF04] will reliably measure down to 3cm and will
    continue detecting down to 1cm or less but after 2-3cm the
    pulse width doesn’t get any smaller.

The SRF04 still detects at less then 3cm, it just reports the
distance as about 3cm.  My experience is that the SRF04 goes
all the way to the face plate.  Ditto for the Ping.

In other words, the SRF04 behaves pretty much like the MaxSonarEZ-1
only its is last accurate measurement distance is 3cm vs.
6 inches for the EZ1.

-Wayne

[snip]

That's another reason to put a small piece of special formed material
between both the transducers, like here
http://oase.uci.kun.nl/~mientki/data_www/pic/projects/rapid_prototyping/rapid_prototyping_us_ranger.html

another 2 cents ;-)

cheers,
Stef

That's also an algorithm problem. You need to write code to
detect stuckness and exploratory (blind) algorithms to get
unstuck.

What I want to know is when is someone going to commercialise
my 3-sensor sonar? With a transmitter in the centre and two
receivers 5cm each side, I can resolve the angular position
of the echo to better than 3 degrees with ease. So when your
robot is trundling along and sees the leg of a chair, it knows
that it's off to the left a bit and can steer around it to the
right. It's really *very* effective.

Clifford Heath.

Clifford,

Get in touch with me, lets talk.

Best Regards,
Gerry.
http://www.robot-electronics.co.uk

------

At what distance ?

Anyway I like the idea of 3 sensors.

So when we put all the ideas proposed in this thread together,
we get the following specs:

- Multi object detection
- Detection range: 0 mm .. 5000 mm
- Measurement accuracy: 0.05 mm
- Angle detection -40 .. +40 degrees
- Angle accuracy: 0.01 degrees
- Output serial 19.2 / 115.2 kB
- Interrupt output, when object within a given range
- maybe it can even give some idea of the size of the obstacle
- temperature output 0 .. 50 Celcius,
- temperature accuracy 0.5 Celcius
- humidity output 20% .. 90%
- humidity accuracy 5%

So who is the first one to bring this on to the market ;-)

(I hope to test the high distance accuracy in the next couple of weeks,
pure for hobby ;-)

Stef Mientki