Minimal inductive proximity sensor?

[Dave]

I have a situation where an 8-pin PIC would monitor an inductive
proximity sensor through a 3 ft wire and control a relay. The proximity
sensor would detect the spokes of a cast iron wheel at maybe 5mm
distance. Since the PIC is available is there any cheaper alternative
to the inductive proximity sensor? Permanent magnets can't be used in
this application.

Thanks!

 

[John Fields]

I have a situation where an 8-pin PIC would monitor an inductive
proximity sensor through a 3 ft wire and control a relay. The proximity
sensor would detect the spokes of a cast iron wheel at maybe 5mm
distance. Since the PIC is available is there any cheaper alternative
to the inductive proximity sensor? Permanent magnets can't be used in
this application.

 

[[email protected]]

Reflection sensor? Farnell has about half a page of "reflective optical
sesnsors", most of them costing a bit under $10 each in small
quantities.

If the cast iron wheel had any significant residual magnetism, you
might be able to do something with a linear Hall effect sensor and
comparator, but it would be more expensive,

 

[Dave]

Reflection sensor? Farnell has about half a page of "reflective optical
sesnsors", most of them costing a bit under $10 each in small
quantities.

If the cast iron wheel had any significant residual magnetism, you
might be able to do something with a linear Hall effect sensor and
comparator, but it would be more expensive,

Gritty, dirty, oily environment...

 

[Dave]

Capacitive proximity sensor. Basically, just a plate that senses
when another plate (the spoke) goes by with air as the dielectric.

What would be the basic setup for this approach? An oscillator on the
remote board that the PIC would monitor the frequency of?

 

[Ken Smith]

John Fields wrote: [...]

Capacitive proximity sensor. Basically, just a plate that senses
when another plate (the spoke) goes by with air as the dielectric.

What would be the basic setup for this approach? An oscillator on the
remote board that the PIC would monitor the frequency of?

The low cost way to do it is to make the oscillator at the PIC side of the
capacitor and have the object provide a ground that increases the
capacitance.

The PIC is wired to the plate at two points. One is an output from the
PIC that drives the plate through a resistor and the other is an input.
The PIC asserts a high for a long time and then a low. It then measures
how long it takes for the input to go low too. It compares this value to
a value from longish ago to remove drifts. If the time gets longer, the
object has arrived.

 

[quietguy]

Maybe the old fashioned approach would work for you - microswitch with
wand protruding into wheel arch - reliable, dust and water etc proof units
available - very reliable - use the PIC to count and de bounce etc

David

 

[[email protected]]

Right - that wipes out optical sensors, unless you can get the system
to squirt filtered oil at your sensor all the time.

 

[Dave]

Maybe the old fashioned approach would work for you - microswitch with
wand protruding into wheel arch - reliable, dust and water etc proof units
available - very reliable - use the PIC to count and de bounce etc

David

I considered that. The wheel speed is slow, but it can be at 100RPM for
long periods so I thought the resulting reliability would be
questionable.

 

[John Larkin]

I have a situation where an 8-pin PIC would monitor an inductive
proximity sensor through a 3 ft wire and control a relay. The proximity
sensor would detect the spokes of a cast iron wheel at maybe 5mm
distance. Since the PIC is available is there any cheaper alternative
to the inductive proximity sensor? Permanent magnets can't be used in
this application.

Thanks!

Sounds like you need a nonmagnetized variable reluctance pickup, like
the ones used for aircraft fuel flow sensors. It looks like a spark
plug, sort of; just a ferrous metal tube with a center post barely
peeking out, and a coil inside. Inductance increases as a gear
tooth/spoke/turbine blade comes close.

You can signal condition this with an oscillator that uses the coil in
its tank (observe frequency shift) or you can put it in an AC bridge.

Google "magnetic proximity sensor." You can buy these with the
necessary electronics inside if you don't want to do it yourself.
You'd have to make a lot of these to justify homebrew.

John

 

[Dave]

John Fields wrote: [...]

Capacitive proximity sensor. Basically, just a plate that senses
when another plate (the spoke) goes by with air as the dielectric.

What would be the basic setup for this approach? An oscillator on
the remote board that the PIC would monitor the frequency of?

The low cost way to do it is to make the oscillator at the PIC side
of the capacitor and have the object provide a ground that increases
the capacitance.

The PIC is wired to the plate at two points. One is an output from
the PIC that drives the plate through a resistor and the other is an
input. The PIC asserts a high for a long time and then a low. It
then measures how long it takes for the input to go low too. It
compares this value to a value from longish ago to remove drifts.
If the time gets longer, the object has arrived.

I can't imagine this working when the probable capacitance change would
only be a few dozen picofarads on the sensor board. I'm more inclined
to believe a seperate monitored oscillator might work.

 

[John Fields]

What would be the basic setup for this approach? An oscillator on the
remote board that the PIC would monitor the frequency of?

---
No, I think that method would be too "drifty", and the frequency too
high because of the size of the plate.

What I'd do is something like this:


+HV +V +V +V
| | | |
| | | [R7]
[R1] [R2] | |
|| | | | +--[R6]--+
|| | | | | | |
|| |--+---[C2]---+------|----+--|+\ |
|| | | | | >--+
|| C1 [R3] [POT[<----|-/
___||___ | |
|___ ___|--+--------+------+
|| |
|| GND
||
||
||<--SPOKE
||

The fixed plate of C1 (the capacitor formed by the plate and the
wheel spoke) is charged up to some relatively high voltage through
R1, a high-valued resistor, and connected to C2, which is used to
block the DC bias, but to let through the change in voltage which
occurs because of the change in capacitance as the spoke passes by
the plate. R2 and R3 are also high-valued resistors which comprise
a voltage divider which serves to bias the comparator's
non-inverting input and set it at a particular voltage. The pot is
used to set the trigger threshold of the comparator, and when the
voltage on the non-invertiing input becomes more positive than the
voltage on the inveting input, the output of the comparator will go
high, then when the voltage on the + falls to below the reference
voltage on the - input, the output will go low. Properly adjusted,
there should be a single pulse out of the comparator every time a
spoke passes the plate. R6 is used to set the hysteresis in the
circuit, which will keep the comparator from chattering around the
switching point and generating multiple outputs for a single spoke
passage.

The nice thing about this circuit is that since a _change_ in
capacitance is all that's being detected, instead of some phenomenon
which depends on the absolute value of the capacitance, the drift in
capacitance (from whatever source) becomes unimportant. Also, the
amplitude of the output from the capacitor can easily be adjusted by
changing the value of the high voltage bias on the cap, which can
probably be used to advantage in dirty environments. Also, the cap
ought to be a snap to clean; just blow it off with an air hose when
it gets dirty.

 

[Dave]

What would be the basic setup for this approach? An oscillator on the
remote board that the PIC would monitor the frequency of?

---
No, I think that method would be too "drifty", and the frequency too
high because of the size of the plate.

What I'd do is something like this:


+HV +V +V +V
| | | |
| | | [R7]
[R1] [R2] | |
|| | | | +--[R6]--+
|| | | | | | |
|| |--+---[C2]---+------|----+--|+\ |
|| | | | | >--+
|| C1 [R3] [POT[<----|-/
___||___ | |
|___ ___|--+--------+------+
|| |
|| GND
||
||
||<--SPOKE
||

The fixed plate of C1 (the capacitor formed by the plate and the
wheel spoke) is charged up to some relatively high voltage through
R1, a high-valued resistor, and connected to C2, which is used to
block the DC bias, but to let through the change in voltage which
occurs because of the change in capacitance as the spoke passes by
the plate. R2 and R3 are also high-valued resistors which comprise
a voltage divider which serves to bias the comparator's
non-inverting input and set it at a particular voltage. The pot is
used to set the trigger threshold of the comparator, and when the
voltage on the non-invertiing input becomes more positive than the
voltage on the inveting input, the output of the comparator will go
high, then when the voltage on the + falls to below the reference
voltage on the - input, the output will go low. Properly adjusted,
there should be a single pulse out of the comparator every time a
spoke passes the plate. R6 is used to set the hysteresis in the
circuit, which will keep the comparator from chattering around the
switching point and generating multiple outputs for a single spoke
passage.

The nice thing about this circuit is that since a _change_ in
capacitance is all that's being detected, instead of some phenomenon
which depends on the absolute value of the capacitance, the drift in
capacitance (from whatever source) becomes unimportant. Also, the
amplitude of the output from the capacitor can easily be adjusted by
changing the value of the high voltage bias on the cap, which can
probably be used to advantage in dirty environments. Also, the cap
ought to be a snap to clean; just blow it off with an air hose when
it gets dirty.

Thanks, that looks like a pretty reasonable approach. I'll do some
experimenting...

 

[Ken Smith]

Ken Smith wrote: [...]

the PIC that drives the plate through a resistor and the other is an
input. The PIC asserts a high for a long time and then a low. It
then measures how long it takes for the input to go low too. It
compares this value to a value from longish ago to remove drifts.
If the time gets longer, the object has arrived.

I can't imagine this working when the probable capacitance change would
only be a few dozen picofarads on the sensor board. I'm more inclined
to believe a seperate monitored oscillator might work.

10pF times a 1Meg resistor gives a 10uS change in the time constant so
whats the problem?

 

[Dave]

Ken Smith wrote: [...]

the PIC that drives the plate through a resistor and the other
is an input. The PIC asserts a high for a long time and then a
low. It then measures how long it takes for the input to go low
too. It compares this value to a value from longish ago to remove
drifts. If the time gets longer, the object has arrived.

I can't imagine this working when the probable capacitance change
would only be a few dozen picofarads on the sensor board. I'm more
inclined to believe a seperate monitored oscillator might work.

10pF times a 1Meg resistor gives a 10uS change in the time constant
so whats the problem?

Ok, now I understand what you're suggesting. Somehow I was thinking you
meant a split plate.