From: John De Armond
Subject: Re: Is my 1 ton Dodge van/RV overheating?
Date: Wed, 22 Dec 1999 15:10:41 EST
Newsgroups: rec.autos.tech,alt.rv,rec.outdoors.rv-travel

Allen Ashley wrote:

> Where is a good place to get thermocouples? I know they have to be
> welded from two different types of wire. Do you make your own?

TCs are simply two dissimilar metals in contact.  Actually this
definition is now extended to semiconductors.  The process of
dissimilar metals in contact generating a voltage while experiencing
a temperature differential is called the Seebeck effect.  The
reverse, causing a temperature differential by the flow of current
across dissimilar metals is the Peltier effect.  This is the well
known method of cooling portable solid state ice boxes.  The peltier
pile inside the refrigerator consists of dozens of semiconductor
junctions hooked in electrical series and thermal parallel.  Little
known is the fact that these piles are also efficient generators. 
My old peltier-cooled ice box will continue running its fan when
unplugged from the voltage generated by the pile as it rapidly pumps
heat back into the cool compartment.  If you want to make an
emergency battery charger, get a couple of these junctions (All
Electronics, etc or strip 'em out of an old icebox.), mount 'em on a
heat sink and connect them together to make an about 14 volt pile. 
To generate up to a couple of amps of current, simply heat the other
side of the junctions with your RV's stove or a propane torch while
keeping the heat sink cool.  Works like a champ! Keep the hot side
under 300 deg F or you run the risk of loosening the very soft
solder used to attach the junctions to the substrates.

The voltage vs temperature curve is very nonlinear, being an nth
(9th for common TC materials) order polynomial.  The coefficients of
the polynomial are different for each pair of metals.  Therefore for
convenience and economy, only certain metals have been characterized
for use as TC metals.  Copper, iron, platinum, tungsten and several
proprietary alloys dominate.  Polynomial terms and tables of
temperature vs voltage are available from the National bureau of
standards (or whatever they call themselves this week) web site.

One can read the millivoltage directly from the thermocouple and
look it up in the NBS table to derive the temperature but it is much
easier to use a converter.  The converter implements a linearization
function and standardizes the output to typically 1 mv per deg. 
Again, because the linearization is different for each TC type, the
converter must be matched to the TC type.  Standard TC types are
known by their code letters.  Type K, for instant is chromel-alumel,
two formerly proprietary alloys that have become ubiquitous.

Older pyrometers (the term is traditionally used to describe analog
meters without any electronics) linearized the reading by simply
having the scale be non-linear.  Analog pyrometers do a good job but
they require that the combination of TC and lead wire have a
specified resistance.  Typically a fixed resistor is attached to the
rear of the meter and is selected to make it and the TC equal to the
specified total resistance.  This is usually 500 ohms.

Thermocouples have become ubiquitous in recent years.  Type K
(chromel-Alumel) has become the defacto high temperature standard
while type T (copper-constantan) is the one for low temperature
work.  Type K is all you really need.  I have seen ready-made type K
thermocouples for sale everywhere from Home Depot to the local auto
parts store.  Many digital volt meters have the ability to directly
read a type K TC.  And the aforementioned converter is inexpensively
available.  Another TC that used to be very popular was type J
(iron-constantan.)  Because the iron rusts and corrodes, it has
fallen out of favor.  Pyrometers can be found on the used/salvage
market very cheap.  I recently bought two type J pyrometers at a
flea market in Alabama for $5 each.

The granddaddy of TC suppliers is Omega Engineering
(http:/www.omega.com).  Also the most expensive.  but they have
EVERYTHING having to do with temperature measurements.  Another
source is Graingers (http://www.graingers.com).  Fluke offers
thermocouples with a Fluke part number so any outfit that sells
Fluke instruments will have 'em or can order them.

TCs can be bought ready-made in a wide variety of forms but it is
also trivially easy to weld your own using TC wire.  A car battery
and a carbon rod from a dry cell battery are all the materials
required.  Twist the two TC wires together and connect the wire to
the positive terminal of the car battery using an alligator clip
placed as close to the weld as possible.  Connect the carbon rod
from a dry cell battery (NOT alkaline) to the negative.  Touch the
TC to the rod, hold until color appears and then moderately slowly
withdraw.  An arc will form.  It may take some practice to learn
just how fast to withdraw but when you get it right, there will be a
small round ball of welded TC material on the end of the twist.  For
low temperature work, the junction can be silver soldered but the
soldering does introduce some error because it alters the metal
alloys.  Not terribly serious unless you're doing something
critical.

A few more TC facts.

Leads and jackets are color-coded.  Rather than recite the codes,
check with omega (URL above).  Type K uses red and yellow wires with
a brown jacket.  If the wire is thermocouple and not thermocouple
lead wire, the jacket will have a white tracer.  For all common TCs,
rEd is always nEgative.  The difference between thermocouple wire
and thermocouple lead wire is the level of purity and thus
accuracy.  TC wire is very pure and meets the requirements of the
various standards organizations for accuracy.  Type K is about +-2
deg if I remember correctly.  TC extension wire is of lower purity,
costs less and is allowed to generate a higher error, typically +-5
deg.  It is used to bring the TC signal from the field back to an
instrument room.  Since the actual TC is what is generating the
millivoltage, the error term of the lead wire is relatively
insignificant.  For less critical applications, I buy TC lead wire
because it is much cheaper.  Over the range of temperatures we're
interested in, the error can be detected and either calibrated out
or allowed for using ice and boiling water baths.  Calibrations at
higher temperatures are done in freezing metal baths.  Anyone who is
interested in this should contact me directly.

If you want to learn more than you ever cared to know about TCs,
study Omega's web site or even better, request a free catalog.  If
you go full boat with Omega, UPS will handcart about 60 lbs of
hard-bound catalogs to your door!  All you really need is their
Temperature Handbook.  This is effectively a graduate course (in
plain english) on temperature measurement.  All of this is free.

John