InfraRed Refective Long Distance Object Sensing

[Mark]

I am relatively new to Infra-Red Electronics and I need help please. I
need to "see" how fast an Aluminum Bar (unpainted/unpolished) passes
in front of a panel. The bar is about 25mm (1") in diameter and it can
pass in front of the panel at different distances from it, ranging
from 0 to about 250mm (10"). I need to do this with a row of Photo
transistors (I cant use Ultrasonic's in this application) and a
Microprocessor using the 10 bit ADC feature. My initial thoughts were
to mount the photo transistors in parallel mounted Black tubes,
pointing towards the target area, so that they had a limited viewing
angle, flood the area in front of the panel with Infra Red light, see
the reflection, let the Microprocessor decide if it is a valid
reflection or not, and then calculate the speed in the usual way. The
panel will be in daylight, but not pointing directly at any bright
light source. Cost is an issue. so this precludes me from using very
expensive devices.

I would welcome any advice people can give me, like is what I am
suggesting feasible? what size LED's/Photo-Transistors to use
(3mm/5mm), what wavelength of Infra-red light would yield the best
results, what would be a good starting length for the Black Tubes,
recommendations of which device's to use, is it worth flashing the
Infra-Red light source, to help with ignoring the ambient light
problem, would I need to amplify the output of the photo transistor or
will a simple resistor up to 5v be good enough with a 10 bit DAC.
Sorry for so many questions, but as I say, I have not done much work
with Infra-Red before.

Thanking you in anticipation
Mark in Spain
[email protected] (remove the X to reply)

 

[Product developer]

I am relatively new to Infra-Red Electronics and I need help please. I
need to "see" how fast an Aluminum Bar (unpainted/unpolished) passes
in front of a panel. The bar is about 25mm (1") in diameter and it can
pass in front of the panel at different distances from it, ranging
from 0 to about 250mm (10"). I need to do this with a row of Photo
transistors (I cant use Ultrasonic's in this application) and a
Microprocessor using the 10 bit ADC feature. My initial thoughts were
to mount the photo transistors in parallel mounted Black tubes,
pointing towards the target area, so that they had a limited viewing
angle, flood the area in front of the panel with Infra Red light, see
the reflection, let the Microprocessor decide if it is a valid
reflection or not, and then calculate the speed in the usual way. The
panel will be in daylight, but not pointing directly at any bright
light source. Cost is an issue. so this precludes me from using very
expensive devices.

I would welcome any advice people can give me, like is what I am
suggesting feasible? what size LED's/Photo-Transistors to use
(3mm/5mm), what wavelength of Infra-red light would yield the best
results, what would be a good starting length for the Black Tubes,
recommendations of which device's to use, is it worth flashing the
Infra-Red light source, to help with ignoring the ambient light
problem, would I need to amplify the output of the photo transistor or
will a simple resistor up to 5v be good enough with a 10 bit DAC.
Sorry for so many questions, but as I say, I have not done much work
with Infra-Red before.

Thanking you in anticipation
Mark in Spain
[email protected] (remove the X to reply)

Hi Mark,

What is the maximum speed that the bar (specimen) will traverse the
aperture or measurement window?

Will the specimen surface profile very in texture from specimen to
specimen?

Will the specimen always arrive in a fixed path?

If the device has to be outdoors you should chop your source to remove
DC effects of the sun. The black body or source should be a halogen
bulb chopped at a frequency high enough to yield several cycles during
the time that the specimen will be present.

You will have to be able to accurately detect the presence and absence
of the
Specimen as it traverses a fixed window or aperture so you may need to
have your device first validate the specimen then set a read mode
where you detect the leading edge of the specimen, start a counter,
and then terminate count upon detection of trailing edge as it passes
out of view.

I have designed coating systems where I had to open a shutter to
evaporatively coat a small 25mm lens or mirror on a spinning table
inside a vacuum chamber but I had the benefit of an absolute position
encoder for triggering. You need to be able to trigger your "time of
flight" counter upon specimen confirmation, which is going to be a
little tricky.

 

[Dave Garnett]

I am relatively new to Infra-Red Electronics and I need help please. I
need to "see" how fast an Aluminum Bar (unpainted/unpolished) passes
in front of a panel. The bar is about 25mm (1") in diameter and it can
pass in front of the panel at different distances from it, ranging
from 0 to about 250mm (10"). I need to do this with a row of Photo
transistors (I cant use Ultrasonic's in this application) and a

Can you back illuminate ? ie put the sources on the other side and look at
when the objects interrupt the beam. If you haven't got enough room for a
source on the far side, then perhaps a retro reflector (plastic device like
a bicycle reflector).

Depending on the trajectory/geometry of the bar, why do you need more than
two sensors ?

If you are operating outdoors or in an artificially lit environment the
using pulsed sources with synchronous detection is a good idea, as is using
infrared filters.

Dave

 

[John Fields]

I am relatively new to Infra-Red Electronics and I need help please. I
need to "see" how fast an Aluminum Bar (unpainted/unpolished) passes
in front of a panel. The bar is about 25mm (1") in diameter and it can
pass in front of the panel at different distances from it, ranging
from 0 to about 250mm (10"). I need to do this with a row of Photo
transistors (I cant use Ultrasonic's in this application) and a
Microprocessor using the 10 bit ADC feature. My initial thoughts were
to mount the photo transistors in parallel mounted Black tubes,
pointing towards the target area, so that they had a limited viewing
angle, flood the area in front of the panel with Infra Red light, see
the reflection, let the Microprocessor decide if it is a valid
reflection or not, and then calculate the speed in the usual way. The
panel will be in daylight, but not pointing directly at any bright
light source. Cost is an issue. so this precludes me from using very
expensive devices.

I would welcome any advice people can give me, like is what I am
suggesting feasible? what size LED's/Photo-Transistors to use
(3mm/5mm), what wavelength of Infra-red light would yield the best
results, what would be a good starting length for the Black Tubes,
recommendations of which device's to use, is it worth flashing the
Infra-Red light source, to help with ignoring the ambient light
problem, would I need to amplify the output of the photo transistor or
will a simple resistor up to 5v be good enough with a 10 bit DAC.
Sorry for so many questions, but as I say, I have not done much work
with Infra-Red before.

---
I suggest that you forget the reflective scheme and use beam-breaking
to do what you want.

But... what is the range of speeds you need to detect and how far
apart were you planning on putting the IR detectors?

 

[Mark]

---
I suggest that you forget the reflective scheme and use beam-breaking
to do what you want.

But... what is the range of speeds you need to detect and how far
apart were you planning on putting the IR detectors?

Beam breaking is not an option, I cannot get anything behind the
target object. The speed I guess is about 10-30 mph (I will verify
that with my customer later), I cannot attach anything to the target,
the path that the target takes is constant but varies in distance from
the panel by the amount already mentioned. I will be using about 20
sensors at a set distance from each other to read the speed of the
target (for reasons that is too complicated to explain at this stage),
the target is not a constant width/shape, but I was going to use an
averaging process (compute the peak refected light value, as being the
centre (the taget is always round/ovoid)) with all the sensors I have.
Accuracy needs to be within 1 or 2 mph, I will be running the
Microprocessors at 20Mhz, 1 for each sensor). Hope that clarify's
things a bit better.

 

[Jim Miller]

obtaining the reflection from the aluminum bar will be difficult. 3M makes a
reflective tape which acts as a retroreflector. can you apply this to the
aluminum?

jtm

---
I suggest that you forget the reflective scheme and use beam-breaking
to do what you want.

But... what is the range of speeds you need to detect and how far
apart were you planning on putting the IR detectors?

Beam breaking is not an option, I cannot get anything behind the
target object. The speed I guess is about 10-30 mph (I will verify
that with my customer later), I cannot attach anything to the target,
the path that the target takes is constant but varies in distance from
the panel by the amount already mentioned. I will be using about 20
sensors at a set distance from each other to read the speed of the
target (for reasons that is too complicated to explain at this stage),
the target is not a constant width/shape, but I was going to use an
averaging process (compute the peak refected light value, as being the
centre (the taget is always round/ovoid)) with all the sensors I have.
Accuracy needs to be within 1 or 2 mph, I will be running the
Microprocessors at 20Mhz, 1 for each sensor). Hope that clarify's
things a bit better.

 

[John Fields]

Beam breaking is not an option, I cannot get anything behind the
target object. The speed I guess is about 10-30 mph (I will verify
that with my customer later), I cannot attach anything to the target,
the path that the target takes is constant but varies in distance from
the panel by the amount already mentioned. I will be using about 20
sensors at a set distance from each other to read the speed of the
target (for reasons that is too complicated to explain at this stage),
the target is not a constant width/shape, but I was going to use an
averaging process (compute the peak refected light value, as being the
centre (the taget is always round/ovoid)) with all the sensors I have.
Accuracy needs to be within 1 or 2 mph, I will be running the
Microprocessors at 20Mhz, 1 for each sensor). Hope that clarify's
things a bit better.

---
Yeah, OK.

What I'd do would be to plan on using an individual IRLED for each
station, (instead of a light curtain) and collimating its output into
a nice, tight beam. That'll get more power on the target and give you
a better signal to noise ratio than if stuck the LED at the end of a
piece of tubing, and you might be able to get away with using just an
IR phototransistor in a tube and and a comparator as the receiver
instead of having to modulate the IR beam and use an IR receiver to
detect it. If you want to do it that way you can get some nice
Fresnel lenses to do the collimating from http://www.fresneltech.com.

 

[Mark]

---
Yeah, OK.

What I'd do would be to plan on using an individual IRLED for each
station, (instead of a light curtain) and collimating its output into
a nice, tight beam. That'll get more power on the target and give you
a better signal to noise ratio than if stuck the LED at the end of a
piece of tubing, and you might be able to get away with using just an
IR phototransistor in a tube and and a comparator as the receiver
instead of having to modulate the IR beam and use an IR receiver to
detect it. If you want to do it that way you can get some nice
Fresnel lenses to do the collimating from http://www.fresneltech.com.

OH shit, lenses etc will most probably not be an option, we are
looking to manufacture quite a few of these, and my subcontractor may
be able to solder boards on his flow soldering machine, in the shed he
has in his back garden, but glueing lenses into situ will be beyond
his reproducable scope. Just to clarify things a little further, the
black plastic tubes were in fact going to be a row of holes drilled
with a vertical pillar drill, at equi-distant spaces in a black
plastic bar, with the holes drilles to an exact depth, which the
Photo-Transistors would be mounted in.

 

[Product developer]

How would you get "behind" the aluminum bar to create an interruptable
beam and what if the bar changes profile longitudinally creating a
false profile?

Using the reflective approach permits distinguishing a face profile
reflection from the bar wall. If the bar arrives in varying angles
reflection is out unless he is familiar with using an array, a Fresnel
lens, or a huge integrating sphere!

If there is enough space between bars

 

[Rich Grise]

OH shit, lenses etc will most probably not be an option, we are
looking to manufacture quite a few of these, and my subcontractor may
be able to solder boards on his flow soldering machine, in the shed he
has in his back garden, but glueing lenses into situ will be beyond
his reproducable scope. Just to clarify things a little further, the
black plastic tubes were in fact going to be a row of holes drilled
with a vertical pillar drill, at equi-distant spaces in a black
plastic bar, with the holes drilles to an exact depth, which the
Photo-Transistors would be mounted in.

Put two bars alongside each other, so each PHT corresponds to an LED.
Pulse each LED at its own center frequency, and just watch. The maxima/
minima thing seems to be something that you were going to have to work
out anyway, right? 'Cause I have no clue about that. But it seems I've
seen LEDs/PHDs with a little collimating lens built into the package.

You say the distance will be constant, but varying? What does that mean?

Cheers!
Rich

 

[Rich Grise]

obtaining the reflection from the aluminum bar will be difficult. 3M makes
a reflective tape which acts as a retroreflector. can you apply this to
the aluminum?

Actually, aluminum's a pretty decent reflector in its own right. ;-)
Especially if there's a black background.
Might want to watch the skew angle, but shininess could narrow your
response pattern.

Cheers!
Rich