Searching \ for '[EE]: [PIC]: How to measure level of liquids and s' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/microchip/devices.htm?key=pic
Search entire site for: '[PIC]: How to measure level of liquids and s'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: [PIC]: How to measure level of liquids and s'
2000\06\02@061225 by Peter L. Peres

picon face
> Not that I've ever tried this, but I figure the temperature variation
> would distribute itself along the tube in some manner [to be measured
> and calibrated empirically], and the math would be some weighted
> averaging routine. Just a thought. The sorta thing you try once, and
> immediately get a strong hunch whether it will ever work or not.

The way to do this is with 2 thermistors buried in a full cylinder such that
their sizes are negligible wrt to the cylinder diameter and whose
(cylinder's) heat conductivity approximates that of the liquid. The relation
that governs the output is so hard to match to reality that a lookup table
must be used. The output depends on the heat conductivity of the liquid and
the thermistors must be run very hot to get decent results. At the same time
the liquid must not boil or convect on the cylinder. If the liquid moves
(pump, drain, convection) without level change then the readout changes (the
conductivity changes).

However, there is a similar method, that is used, and works well. It is the
one with the thermistor wire that is partially immersed. This is what
measures the fuel level in your car's tank probably. The sensors are
available in lengths of up to 10 ft (can you say marine diesel tank ?).
Check out a National catalog for chips that interface to these, although you
can roll your own. Note that most sensors are 'bare' wires but there are
such sensors with teflon or ceramic clad wire which can go into evil liquids.
These are expensive and fragile. The fun part is, that this method is
imprevious to sloshing as it integrates the 'waves' away ;-)

bye,

       Peter

2000\06\02@061235 by Peter L. Peres

picon face
Hi,

the method that uses an air column raised by the liquid is actually VERY
accurate. The output value (air pressure in tube wrt atmosphere above free
surface of liquid) will practically depend only on the liquid density (rho)
and the height. Since 10 meters of water column ~= 1 at you can use a
differential pressure gauge for atmospheric use to read the height in most
usual tanks, and it will be accurate to within the height of the measuring
tube inlet (i.e. the vertical size of the tube opening under the liquid
surface). The readings will be off if the liquid is moving (flowing river =
Bernoulli suction on the tube inlet). Keeping the plumbing clean is a pain
however. Overflowed washing machine anyone ? <g>

Afaik, the humidity of construction materials (such as sand) is indeed taken
by weighing, cooking, and weighing, on a probe taken by a worker. They also
determine the density of the material after cooking. The measuring recipient
is a small round cylindrical pot with a hemispherical bottom which is to be
filled to the upper rim and evened. With a kid's sand tools ;-). They use a
sieve to make the input even and they push it in to remove air pockets. The
vessel I say holds about 2 liters of water (~ 5-7 kg of dry sand). The
density can be very important when working with large quantities (ever run
out of some ingredient with 95% of the concrete poured ? <g>). For materials
that cannot be heated (coffee, tobacco, more) the water is removed by using
vacuum at room temperature (0.3 at will dry it in a few hours). The rest is
the same as above. The recipient is also the same afaik. Another thing they
determine for concrete is the Ph of the input materials. This is done on the
wet material. Having a bad Ph in concrete input materials can really ruin
your day afaik.

Peter

2000\06\02@121935 by Dan Michaels

flavicon
face
Peter L. Peres wrote:
....
>However, there is a similar method, that is used, and works well. It is the
>one with the thermistor wire that is partially immersed. This is what
>measures the fuel level in your car's tank probably. The sensors are
>available in lengths of up to 10 ft (can you say marine diesel tank ?).
>Check out a National catalog for chips that interface to these, although you
>can roll your own. Note that most sensors are 'bare' wires but there are
>such sensors with teflon or ceramic clad wire which can go into evil liquids.
>These are expensive and fragile. The fun part is, that this method is
>imprevious to sloshing as it integrates the 'waves' away ;-)
>

Peter, can you more fully describe the makeup of this 10' length
of sensor? I thought a thermistor was basically a resistive bead with
copper wires attached. Are you referring here to some kind of distributed
thermo-resistive element? Confusing.

- Dan Michaels

2000\06\02@225249 by Ian Wilkinson

flavicon
face
On Thu, 01 Jun 2000 in "Re: [EE]: [PIC]: How to measure level of liquids and sense", you wrote:
I don't know if this has been mentioned before as I've not been following this
until I noticed a system that uses two metal rods that are placed into the
liquid.  The resistance is measured and as the liquid rises the resistence
drops.  Now this will probably only work for water based liquids, but does look
to be a very quick solution...

Ian.
--
Hofstadter's Law:
The time and effort required to complete a project are always more than
you expect, even when you take into account Hofstadter's Law.

Uptime at  3:28am  up 1 day, 11:20,  4 users,

2000\06\03@150830 by Peter L. Peres

picon face
>Peter, can you more fully describe the makeup of this 10' length
>of sensor? I thought a thermistor was basically a resistive bead with
>copper wires attached. Are you referring here to some kind of distributed
>thermo-resistive element? Confusing.

Dan,

normally thermistors are indeed bead-type using discoidal or cylindrical
active elements, but the original [tm] - ca. 1800's ? - way to do this, is
thermistor wire. Which comes on reels and is spec'd in ohms/meter. This is
different than resistor wire, as the latter has a near-zero temperature
coefficient, whereas the thermistor wire has a positive (and very linear vs.
other methods) coefficient. Tungsten wire comes to my mind as an inexpensive
<grin> candidate (note: tungsten wire: not tungsten steel. Chemical symbol
W, aka Wolfram in other parts of the world). This is what incandescent bulb
filaments are made of usually. In real life, complex and proprietary alloys
are used for the wire. The W wire is very fragile (it shatters like a
ferrite) so it has very limited uses. The tempco (for certain W alloys) is
about +0.03-0.05 ohms/K at around room temperature. You should really get
hold of a National Data Book that features the relevant chips, they go into
some depth with this (Automotive Section).

The wire is spanned in a frame (or a tube with openings), and immersed
vertically in the liquid. The liquid wets a part of the wire, and this part
will be cooled very efficiently. The part that sticks out heats up due to
the current passed through and this is what you measure. The wires used in
automotive apps. have a very high tempco (can't run them hot). The accuracy
depends directly on the temperature of the exposed wire and on the tempco of
the wire (both as high as possible to make it more accurate).

Peter

2000\06\05@141742 by Craig Lee

flavicon
face
I use a conductivity system with two probes referenced to a common probe
for sensing and controlling liquid levels.  The system is designed using
AC voltage to limit mineral plating to the electrodes.  I wouldn't want
to use it in a flamable liquid! It acts as kind of a Schmitt Trigger.
An output relay is turned on once the top electrode is emersed, and the
output is turned off once the lower electrode is out of the liquid.

Is this of interest to anyone?

{Original Message removed}

2000\06\05@143702 by Craig Lee

flavicon
face
Why not just use a float in a tube connected to an indexed rotary
encoder.  Counterbalance the float to a spring on the shaft of the
encoder, so that the float will extend the length of the tube unless
liquid is present.  Or instead of a spring, just counterbalance with
another weight.

Then just read the gray code.  Some of them even come with their own
counter and SPI interface, making it even easier to connect to a pic.
I'm using a unit that has 4096 transitions per revolution and there
are standard models that do 16384, so resolution certainly shouldn't
be an issue.

See http://www.opticalencoder.com for rotaries that might work for ya.

No Bernoulli issues, no ADC noise problems, no lookup tables, no calcs,
no plating, or liquid specific issues....


Craig


{Original Message removed}

More... (looser matching)
- Last day of these posts
- In 2000 , 2001 only
- Today
- New search...