"ehsjr" wrote in message
news:ALthj.4679$tZ6.1291@trndny03...
Oppie wrote:
"ehsjr" wrote in message
news:3qihj.3848$K%6.3695@trndny04...

Joerg wrote:

ehsjr wrote:


You call that a coil? That's not a coil.
*This* is a coil!


No, it's a petunia. Or whatever. (I am not a botanical expert)

So, what's that super-stretch edition of coil for?


For a permeability tuned oscillator. It's an experiment to
measure water level, which changes only 3 feet nominal at
the site.

The finished coil is 42" of wire on 46" of PVC. A float
makes the coil ride up and down on a galvanized rod as
the water level changes.

It is accurate to within an inch, with +/- 3" being acceptable,
so that part is good. There is a big (ie 3" change in reading
with a 30F change in ambient) temperature problem to be solved,
but I ran out of warm weather to work on it. Looking for +/- 3"
accuracy between 50 and 100 F.

Ed


Where do you think the temperature effect is coming in - the oscillator
portion or the coil?
Some freeze spray is usually a good way to find gross effects of
temperature.

Does the inductance change also with temperature? An inductance bridge
would be helpful here.

How about using a LVDT approach instead?


Your question is right on target - it is "where I'm at"
in the experiment. Really, "where I got to" - but now
I have to wait for May with day and night temp over 50.

I think at least some of the temperature effect is coming
from the oscillator, in spite of the fact that it is
temperature compensated. (The output variation with
temperature is "solvable" at this point with a lookup table
and temperature input, but that is really addressing the
symptom, not the cause and adds complexity. Aside from that,
I'm more interested in pursuing it from the cause side of
things.)

The whole thing, at this stage, has to be outdoors - and that
includes the test equipment. The inductance measurement changes
with temperature, AIRC, but I don't know yet how much of that,
if any, is test equipment error. Using the frequency change
approach vs delta L makes a wider input, and small changes
in level were easier to see. I hadn't thought of pulsing a
current and measuring V.

I may have to jury rig something for testing to bring the
equipment inside where the temperature remains the same.

Ed

You've also got a lot of copper wire there in the coil that has a positive
tempco. Does a + tempco figure into your experimental results?

Does the frequency output stabilize immediately after power up or does it
drift to a 'stable' point? Perhaps pulsing the operation of the sensor is a
reasonable strategy.

What range of frequency are you expecting? You mentioned that the coil
assembly is attached to a float and rides over the fixed iron rod.
Hysteresis effects in the iron and capacitive effects between the coil and
surrounding water/air should be considered.

LVDTs are interesting since they ammount to a balanced bridge and the output
is the ratio of coupling between an excitation coil and two sensing coils.
Being balanced, common mode stuff falls out of the equation.