Electrodeless Conductivity Sensor circuit question

[Erikk]

I have to build a conductivity sensor with very low accuracy
requirement (50% is good enough in the 10 to 10,000 uS/cm range). I
am not all that familiar with the inductive method using 2 torroids,
one for the excitation, and the other for detection.

What is a good starting point for coil inductance, coil dimensions,
separation, excitation frequency, etc?

If anyone has a good reference book or informative web site on this
topic, please let me know.

Thanks.

 

[Erikk]

On 25 Mar 2005 12:01:24 -0800, dajpe@aol.com wrote:

This method is often used to measure the PH of a liquid. I have been
looking into this for one of my clients and have conducted some
experiments. Try an inductance of about 10mH. Drive that coil through
a capacitor with a single pulse, whose pulse width is picked to equal
one half of the period of the frequency of this series resonant LC
circuit. This pulse will cause the network to ring for many cycles.
You can use a CMOS transistor gate driver IC or you can gang several
CMOS inverters to form the driver. The pulse rate can be pretty slow,
about 50Hz or so. This keeps the power drain low but does launch a
hefty signal. Use a second coil of similar inductance, to form a
parallel resonance circuit with a capacitor. Place a resistor in
parallel with the network to limit the Q to about 5 or so. The ring
signal from the exciter coil will produce a nice ring signal in the
receiver coil, which lasts several cycles, provided there is a
conductive fluid through the center of the two toroid coils. Use a
peak detector to capture the ring amplitude. Do a little filtering and
you will then have a nice DC voltage, which is proportional to the
conductivity of the fluid.

David A. Johnson, P.E. --- Consulting Engineer
http://www.djandassoc.com
Home of http://www.discovercircuits.com A collection of over 9,000
schematics.
Home of the Imagineering on-line magazine:
http://www.imagineeringezine.com

Sounds like a good idea. I was actually hoping to use a continuous
excitation at about 10KHz and use a synchronous detection (lock-in
amp) on the receiving side.

Did you have about 10mH on both coils?

 

[Tony Williams]

In article <1111779897.530075.238650@o13g2000cwo.googlegroups.com>,
<dajpe@aol.com> wrote:

This method is often used to measure the PH of a liquid. I have
been looking into this for one of my clients and have conducted
some experiments. Try an inductance of about 10mH. Drive that
coil through a capacitor with a single pulse, whose pulse width
is picked to equal one half of the period of the frequency of
this series resonant LC circuit. This pulse will cause the
network to ring for many cycles. You can use a CMOS transistor
gate driver IC or you can gang several CMOS inverters to form the
driver. The pulse rate can be pretty slow, about 50Hz or so.
This keeps the power drain low but does launch a hefty signal.
Use a second coil of similar inductance, to form a parallel
resonance circuit with a capacitor. Place a resistor in parallel
with the network to limit the Q to about 5 or so. The ring
signal from the exciter coil will produce a nice ring signal in
the receiver coil, which lasts several cycles, provided there is
a conductive fluid through the center of the two toroid coils.
Use a peak detector to capture the ring amplitude. Do a little
filtering and you will then have a nice DC voltage, which is
proportional to the conductivity of the fluid.

I wonder if it would be viable to resistively load (or
even short circuit) the secondary and then look at the
Q, or watts-loss, of the primary. That would give a 2-
-wire sensor.

--
Tony Williams.