I've been wondering exactly how those traffic light sensors work.  These are
the inductive loops they put under the pavement to sense the presence of cars.
Now it isn't very hard to figure out that they can put a signal into the loop
and then look at whether or not it is attenuated (or perhaps boosted somehow?)
by the change in inductance caused by the presence of a big mass of steel
nearby.  

But what are the frequencies and protocols?  I've had this idea for a small
device that could trigger the sensors.  This would be useful when riding a
bicycle and needing to trigger a light change.

37 responses total.

Get out your frequency counter and go sniffing....

Neat idea, Scott!  If the details are universal enough, you could probably
sell quite a few of the things.  Unless the FCC clamped down on you.

(I don't have a frequency counter, or a spectrum analyser.  <sniff>)

I suspect the frequencies are rather low, but not too low.  Something in the
audio range?  Is there one loop or two?

I have had *exactly* the same thought!  From what  little reading I've done,
they are tuned systems.  There is a coil imbedded in the road who's
inductance changes when a car passes over it.  The changing inductance
is sensed and used to inform the light's CPU that there is a car present.
This is the simple system.  To reduce the problems associated with ice,
snow and rain, multipul coils are imbedded, one as a referance.  Motion
sensors will sometimes be mounted on light poles to sense traffic.  I
think thy are PID's.
 
My though was to suspend a coil from the bottom of a car such that it is
only a few inches above the road.  Then drive around and "sniff" traffic
loops using a scope.  I have a 200W inverter and a scope.  It's time I 
don't have much of these days.  I'm also not sure how one would go about
"pulling" the freq. of a loop with a small hand-held device.  Any thoughts?

Getting the frequency shouldn't be hard, it is likely playing it back that
would be difficult.  Fer instance, if the system works by sending a high freq.
thru the loop, then the presence of a car would increase the inductance,
raising the impedance at the freq. and therefore attenuating it.  Now how
would you make up a signal to attenuate the one in the loop?  Destructive
interference would be a bit hard to coordinate...

I think it is important to know how the sensor detects the car. If it
is just mutual inductance - you will need to have a *big* loop to cut
enough lines. However if it would respond to a signal induced into
the sensor, you hve a better chance. You can have a small coil, but
several hundred amperes (momentarily). Hmmm...which newsgroup would be
closest to this problem? Did you think it would fit in rec.home.automation?

Well, we might try the radio conf. with this.  I don't read Usenet.

That should have been comp.home.automation. I'll try there. (I suspect
everyone in the Grex radio cf is also here - hmmm, this is linked somewhere,
maybe radio, anyway.)

Carson linked it to the Cars conference.

ssh!

Saying it has to sense cars is only part true.  They also have to sense bikes.
Sensors, like the one by scott's apartment, that don't sense bikes can be
legally considered broken.

What?  You've found ones that do sense bikes?

They may be legally broken, but I doubt that you'll get them fixed any time
soon.  I've known too many that were pretty stuffy about sensing *cars* if
they weren't fairly big, and A2 was in no hurry to do anything about it.

Being a bicyclist, I know of what you speak.  It was the reason for my
wanting to build such a device.  I don't think a short duration current
pulse through a small coil will do the job.  The device seem to
use the element of time in deciding if the car is still there or if it
made a right on red.  A quick pulse will make it think you ran a red light
and the COPs will be dulley notified ;)
 
Ann Arbor seems to be a Beta test bed for traffic sensing lights.  It is
my opinion that the new ones don't work any better then the original ones
they were putting in in the 60's.  A four way Stop sign seems to be just as
effective, if not more so.
 
Yea, we need more specific detail on the nuts and bolts of the sensors.

Huh?  I've seen traffic sensing lights in other cities easily sense
a bicycle, and that was 10 years ago at least.  Ann Arbor has a dearth
of traffic sensing light, though if the ones they install don't work
well, I can see why.

There are a lot of them in Ann Arbor that can sense bikes, at least if I put
a foot down and lay the bike somewhat flat (to increase the amount of metal
near the road).  Being legally broken is an important condition, even if it's
not going to help getting them fixed.  It's legal to treat a broken traffic
light as a stop sign.  If such lights weren't considered broken, cyclists
would just be stuck until a car came along.

That is good to know, however it does not solve the problem.  The newer 
lights will not ever turn a given light green if it doesn't sense traffic
in the lane for that light.  Hence you have a light that is selectively
"broken" for bicycles but working for cars.  These lights are set up to
maximize traffic flow and maintain constant traffic flow through the in-
tersection.  As a result it can be difficult and dangerous to cross against
the red.  I pass through one of these lights on my way to and from work
at Industrial and Eisenhower.  It's a real pain to get through at times!
 
By law, all of these lights are required to sense bicycles.  Most of them
do, provided you know *exactly* where the hot-spot is or if you lay your
bike on the road where the sensing coils are.  One of these days I'll have
to mail my letter to the city concerning this issue.  Those hot-spots 
should be marked on the road.

According to an article I read a few years ago, the hot spot is generally a
couple feet behind the white line you're supposed to stop behind, and a couple
of feet to the right of the center line, but I don't remember what their
specific measurements are.  Putting the bike at about a 45 degree angle in
one of those spots usually works for me, although there are a few lights where
it doesn't.  Some lights are even sensative enough to sense a bike standing
upright.  I think that's what the law actually requires.  Sensors that were
installed after the road was paved are easy -- just look for cracks in the
pavement.

They certainly should sense an upright bike.  OTOH, it seems like the
technical problems of having *that* work but also not being triggered by
cars in the next lane or something might be difficult to work out.

A couple of feet to the right of center and a couple feet behind doesn't
work for my trouble light.  Not even a car will trip it if you follow that
rule.  The hot-spot should be marked.

That positioning seems like it would put the coil directly under the
engine of a car that's stopped just short of the white line.  Maybe  this
is to make detection of those new plastic-bodied cars more reliable?

The coils are pretty big; if you see the retrofit grooves at older
intersections, the coil is about 3-4 feet across and several more feet long.

At the Plymouth and Barton intersection, the coils seems to be at least 15
feet behind the line.

A good usenet group for this would be sci.electronics. Don't let the "sci"
part fool you, most of what gets discussed in that group is home hobby
type electronics. This discussion would be very much at home there.

Thanks Greg - I've been meaning to find out what group would be best,
but hadn't yet. I'll drop into sci.electronics, and see what's going on.
I know about sci. - sci.geo.satellite-nav has become nearly entirely
GPS use, and most of that just for outdoor recreation.

at some intersections there is a little radar unit or perhaps infrared
unit sitting on the crossbar, dunno which. But it works. I've spent
soem time recently with the traffic-signal-fixit supervisor getting
the low down on 'broken signals.' A2 is kinda hodge-podge but is
rather soon going to a fiber optic control system.
  
As far as the motorcycle sensing, it has always been a problem unless
you know rightwhere the sensor is adn bike directly on top of it.
  
I was wondering about some sort of ignition-noise pickup coil which
would make a lot of sense, it would seem. There must be something
wrongiwththe idea though, since it hasn't been used, i guess.

Perhaps the sheilding that protects radio reception attenuates the noise
too much.  It wouldn't work for diesels or electric cars, either -- no
ignition to generate the noise.

Wouldn't work with me on my bike either.  Hey should simply mark the 
Hot-Spot.
 
How does a fiber-optic cable sense a car?  Or is that for controlling the
light from traffic central of whatever they call the place with all the
antique light control computers.  In that case, how will switching from
wires to fiber optics help anything?

Heh.  A fiber optic cable wouldn't be very useful as a sensor; one of the big
advantages of fiber for some applications is that it is nearly impossible to
intercept communications without actually cutting into the thing.

Nice things about fiber optic cable are: it's very immune to noise,
non-conducting, light weight, vibration resistant, nearly impossible to
corrode, & supports high data rates.  Not all of these are useful for
traffic sensor purposes, but enough are that I suspect it's very
attractive.  Fiber optics are also attracting similar attention in
automotive applications.  In both these cases, the cables & tranceivers
used are almost certainly different than what would be used in long haul
land based uses; for instance, the cable is likely to be plastic rather
than glass.  The glass used in long haul cables is *very* transparent,
to reduce loss, but this isn't necessary just to cross the street or
travel the length of a car.

Just so long as City Concile hasn't been sold a bill of good that does
little, in reality, to help solve their traffic flow problems.  I figured
replacing the control computers was at the top of their list.  If light-
ning strikes are taking out equipment because it's all strung together
with wires, opto-isolators on either end would solve that along as being
cheaper than replacing all the copper with fiber.
 
(Once I built a liquid level detector out of a plastic optical fiber ;)

I agree about the computers.  As a programmer myself, I find that a lot of
lights have truly brain-dead operations, like ignoring sensors at certain
times.

Last I heard, Ann Arbor was still using a more than 20 year old computer to
control the traffic lights.

With fiber-optic links.

Shhh!  Don't tell the Nova it's pumping photons not electrons!

I think the fiber optics translate the relative glow of the vacuum tubes into
instructions for the local light control boxes.

Wow.  I thought they'd switched from their Nova a couple of years back.

guess i shoulda been more carefully explicit .. the fiber is for signal
control, not sensing. And teh computer that controlled the signals
died in Dec 93 .... and i have this Feb 94 problem......

Response not possible - You must register and login before posting.

- Backtalk version 1.3.30 - Copyright 1996-2006, Jan Wolter and Steve Weiss