John Baymore on thu 29 aug 96
----------------------------Original message----------------------------
From: Tom Gray
Subject: digital pyromoters
...... Why does a digital pyrometer register inaccurately at higher temps.?
............. At bisque temps my pyrometer reads pretty close to
accurate (1830)-but at ^10 it reads 2260. Any ideas?
Tom Gray
--------------------------------------------------------------------------------
---
Tom,
The accuracy of a pyrometer is affected by a LOT of things. The main corollary
is basically the more you pay, the more accuracy you get. About a month ago
there was a long, long discussion about this subject. Here is what I could
quickly find and "cut and paste" together from my own past comments...... you
could look at the old "Digest" to see those of others:
----------------------------old messages complied--------------------------
One factor is the "quality" of the meter itself...... the accuracy of the
electronics. An inexpensive meter is pretty inaccurate. A meter rated at +/-
2% of full scale at 2400F (a pretty good analog meter) is still +/- almost 50
degrees. So if it reads 2300 you only know it is somewhere between 2250 and
2350.
-----------next post--------------------
This thermocouple, pyrometer, and controller thing is sort of like the
discussion that's been going on about barium and glaze chemistry....... there is
a lot more to it than first meets the eye. Now we're leaning into the realm of
kilns......... as you know, that' s an area in which I have some expertise .
Many of the things we "assume" to be true from "experiential logic" are simply
not. A pyrometer seems like such a simple thing on the surface. Stick a probe
in the kiln and attach a guage to read the result. Yet there are complexities
untold in these devices. The pottery suppliers haven't helped the situation
much either. They have us expecting a cheap "magic bullet" in these things.
The controller on your kiln uses type K thermocouples, most likely, like
99.9999999999% of all equipment supplied to the craft potters. The Seebeck
coefficients (temperature change related to voltage change created) for Chromel
/ Alumel type thermocouples (K) goes somewhat non-linear at the top end of the
use range. Seriously oversimplified, this means that one unit temp change no
longer results in one unit voltage change. That makes the software for the
controller a bit difficult for the programmer, and the conversion from
microvolts to degrees gets less accurate as the temp goes up.
For example, the Seebeck for the 50 degree interval from 600C to 650C changes .3
microvolts per degree Centegrade. As you get up to 850C to 900C, it changes
more, from 40.5 uV/dC to 40 uV/dC....... a change of .5 microvolts. From 1150C
to 1200C it goes from 37.2 uV/dC to 36.5 uV/dC, a change of .7 microvolts. From
1200C to 1250C, it changes .8 microvolts. The software has to take this
increasing "curve" into account.
Worse, the ASTM recommends type K for use only up to a max of 1260C. So a
firing to cone 10 (150C/ hr....... 9 hours+/-) is pushing the absolute limit of
their usefulness by a great factor...... 105 degrees C over the max. use rating
of the sensor. So they simply should not be used for cone 9-10 kilns. Yet they
are all the time.
Continuing this complex picture, that rating is for PROTECTED (more on that
later) thermocouples of AWG (wire thickness) of 8 ......... yes EIGHT!!!!
(pretty darn thick) The maximum use rating goes DOWN to 1090C (1970ish F) for
AWG 14 (look at household 14-2 wire for a size comparison), 980C for AWG 20, and
870C for AWG 24.
So...... what guage (AWG) are your thermocouples? I think your kiln's
manufacturer uses 8 AWG...... they do on the single pyros they sell. Not sure
on the controllers though. Look at the household wire....look at your
thermocouples. If it is AWG 14, it is only recommended to be accurate by ASTM
(within all the other constraints) for use up to about cone 4!!!!!!
Enlightening, huh?
Next, you have to look at the design accuracy of the manufacturing process to
start with. ASTM specs that the initial calibration of type K is +/- 2.2C or
..75% (whichever is GREATER) between 0 and 1200C. That is the BEST it will ever
be....when it is brand new. At 1200C, that allowable .75% is +/- 9 degrees. At
cone nine, plus or minus 9 degrees C is about 1/3 a cone (ignoring the heat-work
concept for simplicity)!
Plus, you have to deal with the accuracy of the meter (or controller). Good
analog meters are only accurate within 2% of full scale to start with. If full
scale is 1300C then it is only accurate +/- 26 degrees C, and THAT is of the
thermocouple output which is also only +/- 9 degrees C!
Additionally, type K is subject to oxidation when used IN AIR above 750C. This
oxidation results in the electrical output changing, and the need for constant
recalibration. This recalibration requires some electronic equipment that costs
a lot of $, and some skill to use. (Industry recalibrates on a regular
schedule to maintain precise production control.) Type K should not be used in
reduction atmospheres or sulpherous ones, so ASTM also says they are best used
enclosed in a sealed tube with an inert gas........ (a thing you'll never see in
a potters kiln!) BTW..... A vacume on the protection tube is also a no-no.
In general, for cone 9-10 use, type K is useless for REALLY accurate work at the
top of the temperature range. They are used on "potters kilns" simply because
they are CHEAP, and most potters won't pay for the platnium/ 10%
platnium-rhodium (type S) thermocouples and control systems that really should
be used (and the more expensive plated switches and contacts that entails).
Type K simply are more resistant to oxidation than the really cheap types E,
J,and T at temps over 500C, so they got pressed into use.
So you DO just get used to the fact that the temperature shown is not really the
temperature in the kiln (pretty bad since that is what you are lead to beleive
you are paying for! :( You also get used to the fact that the accuracy
will change with time. You "work around" the inherent inaccuracy. So all the
fancy "scientific" sounding precision of these controllers supposedly offer is
subject to a lot of variables.
What these controllers SEEM to promise (incredible accuracy), CAN be done, but
not at the price point that they have to occupy to meet the budget of a studio
craftsperson. Industry has much better stuff.
YET.................................., it is still far better than the old "seat
of the pants", look in the kiln, sniff the wind, discuss it for a half-hour,
throw salt over your shoulder, stuff we used to do (I fire a wood kiln....I
still do all that stuff!).
Luckily, most potter's applications are not all that critical at the top end of
the firing for cone 9-10 work. So the manufacturers get away with type K in
most cases. If you are getting into trouble and need more accuracy, you will
have to change controllers to one that uses the platnium thermocouples. And
it'll cost you some $$$. But from the sound of it in the thread, you just need
to understand the limitations of the tool, and work with it.
PS: If your readings are that far off and you DO have type S (or type R or B)
thermocouples, call the kiln supplier. Something is really off there with the
controller.
---------next post -------------
You got good info from whoever you talked to. Yes, you will need to "match" the
circuit again to make the readings accurate (or at least as accurate as they can
be ....... see other post). If you are not into this "tech weenie" type
stuff, you probably will do best to have your supplier do the math and the
conversion for you. They gave the right answer, so they probably know how to do
it correctly.
The meter you have needs to see a specific resistance to electron flow in the
overall circuit to be accurate to it's calibrations printed on the face of the
meter. If you change the length of the wire, the total resistance changes. So
if you didn't do anything else, the readings would change. So you add a
resistor of the correct value (Ohms) into the circuit to adjust the resistance
the meter "sees". You will have to add the correct value resistor to the
circuit either in series or parallel to make the total resistance match what the
meter was designed to see.
Resistors in series with the circuit are simply additive in value. Rt = R1 + R2
+ R3 + .........Rx. Resistors in parallel are a little different. The
following expresses that relationship for two resistors in parallel: Rt = R1R2
/ R1 + R2.
So here is some more info about pyrometers to help out (or confuse ) in the
process:
You will also need to use the correct wire to extend the length. You can't just
use copper wire (or anything else). If you do, the place where you join the
thermocouple lead and the copper wire becomes another pair of thermocouples!!!
These junctions of dis-similar metals will create another millivolt signal that
will combine with the other "real" signal (either plus or minus depending on
polarity....or "direction"..... of electon flow). Worse yet, there will be TWO
new thermocouples (each junction)..... one Copper / Alumel and one Copper /
Chromel each producing different signals! As the the temperature at this
junction (probably the outside air around your kiln) changes, the signal will
change.
Also note that you need to use a junction between the extension wire and the
thermocouple that keeps both legs of the comnnection made out of the same
materials. If you don't go directly wire to wire (which is a bit of a bad
idea..... more in a second), but put a terminal block in the circuit, it's
contacts need to be rated for type K (or whatever). However, if you use some
sort of pressure connection that is made of copper (or whatever) that presses
the two wires directly onto each other, you can use other metals without any
appreciable problem. Otherwise you create more little thermocouples!
Now, on the the need for a terminal block between the extension wire and the
thermocouple. If you use a thermocouple extension wire that is not the exact
same composition as the thermocouples ("extension" grade wire is slightly
different than "thermocouple" grade wire for type K), there are those two
undesirable "thermocouples" formed at the junction. They are pretty "poor"
thermocouples since the metals are not really all that dis-similar. But they do
add error. If each of the legs of the wire is at a different temperature, that
complicates the inaccuracy! So you want a terminal block designed to conduct
heat to keep the two legs of the connector at about the same temperature. This
is called an "isothermal block", and it's job is not only to provide a
mechanical connection, but to keep the junctions at the same temperature.
While we are on the subject of the connections between the thermocouple and the
extension wire............ a pyrometer functions by measuring the difference
between the hot junction (the tip of the probe in the kiln chamber) and the cold
junction (usually thought of as the the connection point to the meter....or
where the system goes to copper wire). As the temperature of EITHER the cold
junction or the hot junction changes (or both), the reading changes.
In other words, to be as accurate as possible the cold junction needs to be kept
at a specific reference temperature! That reference temperature in industry is
called the "ice point", cause it is precisely 32 degrees F. You could keep the
cold junction in an ice water bath . Industry uses an electronic means to do
this same thing...... but it costs $$$. It uses a resistance temperature probe
to measure the temperature of the cold junction and then inserts a corrective
voltage into the signal to make the system "think " the clod junction is at 32
degrees F. (this is called hardware compensation) Unless you want to spend the
bucks, all us potters need to know is that if the cold junction gets hot, the
reading on the meter is not accurate.
(Industry now uses software compensation more often for the cold junction via
computers. It doesn't correct the voltage, it just knows how much the signal is
"off" and compensates for it in calculations. Very little industrial
temperature measurement is done with anything but digital systems now. The "ice
point" system is going the way of the dodo bird. Analog meters are pretty
ancient technology in industry.)
This need for like metals in self-powered systems (cheap analog) is true of any
switches placed in the system too, unless you start to use some more electronics
(no longer self-powered). There are special switches made for switching each
different type of thermocouple lead. Since the switches are made of the same
alloy as the wire, the temperature issue is not important.
So you need to buy what is called (appropriatly enough) "thermocouple extension
wire". It is made of the same stuff (basically) as the thermocouples
themselves...... at least for type K. So first of all, you need to know the
type of thermocouples you have. Probably, they are type "K", which is a
combination of Chromel and Alumel (synthetic alloys). The meter face may say
that right on it. Then you need to get extension wire for that type of
thermocouple. Your supplier probably can sell it to you in short lengths.
For the type of insulation you need on the wire, you should look at the
environment it will be in (is it mostly in the open, air cooled, or is it mostly
enclosed). If it is very hot, you'll need a wire that has insulation that will
stand that level of heat. I have used a polyvinyl chloride insulated wire in
most kiln applications where I can route the extension wire well "in the clear"
and it has worked fine. It's good for 220F temps, has good flame resistance,
and is pretty flexible. The next grade goes to teflon insulation, which should
be good for most any "potter type" applications. It is good to 500F. ( You can
even get extension wire with a ceramic fiber braid for insulation that is rated
for continious duty at 2000 degrees! BIG BUCKS!) You supplier probably has the
right insulation on what they get, unless your installation is very atypical.
The guage of the wire is sort of important. Use the thickest guage you can
afford (the thicker the more expensive). The signals in these simple
unamplified systems are extremely low....... you don't want to complicate that
with small diameter, high resistance wire. Plus it is possible for the
resistance in the wire to be great enough when compared to the resistance in the
thermocouple itself, that the meter starts to read a "false" thermocouple.....
the wire, not the real probe. I usually use AWG 14 stuff called "EXPP-K-14"
from Omega Engineering. It is an alloy that works with type K thermocouples,
but is what is called "extension grade". The code number stands for EXtension
grade wire, PPolyvinyl Chloride insulation, type K wire, 14 AWG thickness.
Next the issue becomes the resistance of the wire to electron flow. Generally,
the thicker the wire, the less resistance. To calibrate your "system", you'll
need to know the resistance per foot of the wire. Electrical resistance is
measured in units called Ohms. The EXPP-K-14 mentioned above has a resistance
of 0.146 Ohms per double foot (one foot running length, but two sections of
wire) at 68F (*Note-see below). So ten feet would have a total Resistance of
14.6 Ohms. So you will need to know how long a piece of wire you are using
also.
[* NOTE: As I mentioned in the other post, if you are dealing in fine accuracy
this stuff gets complicated beyond belief. The actual resistance of the
extension wire changes with the temperature of the wire! So the accuracy is
affected slightly as the temp of the wire changes from 68F. By itself, not a
significant change for potters, but just another variable that affects the
accuracy picture discussed in the other post.]
Another interesting fact to know is that if you route a long length of extension
wire parallel to and close to a live AC line, you can create a small signal in
the extension line by what is called capacitive coupling. Since we are dealing
with desirable signal levels of only millivolts here, this is a concern. In the
"trade" this type of erroneous, induced signal is called "noise". So route your
extension away from AC lines, and preferably at right angles to them if they
have to come close.
Next you need to know the resistance the meter likes to "see". This may be
printed on the meter as "external resistance". Might not be. If not, you'll
need the help of the meter supplier. You need to know this to get the new
system (meter, extension wire, and thermocouple) to be the same resistance
(Ohms) as the original meter was designed for.
Pyrometers are really just millivolt meters or milliamp meters. They measure
the "pressure" of electricity in the system, like a gas guage or the volume of
flow. The scale that is printed on the front is really a conversion from
electrical values to degrees that the manufacturer worked out for us.
So that is a bit more on pyrometers.
--------------------end of old messages -------------------
So hopefully that should help you out Tom.
.........................john
John Baymore
River Bend Pottery
22 Riverbend Way
Wilton, NH 03086
76506.3102@Compuserve.com
http://www.CraftWEB.com/org/jbaymore/rivrbend/shtml
Spehro Pefhany on fri 30 aug 96
Ok, here are are some more comments...
John Baymore (76506.3102@CompuServe.COM) wrought:
[remaining comments are from quoted articles]
: longer results in one unit voltage change. That makes the software for the
: controller a bit difficult for the programmer, and the conversion from
: microvolts to degrees gets less accurate as the temp goes up.
No, if the software is properly written, the accuracy isn't affected by
the linearization (though there may be poorly written software out
there). 0.5% of 2000'F is more than 0.5% of 1000'F, though. That has
nothing to do with any linearity curve though..
: 1200C to 1250C, it changes .8 microvolts. The software has to take this
: increasing "curve" into account.
Yes, of course it does.
: Continuing this complex picture, that rating is for PROTECTED (more on that
: later) thermocouples of AWG (wire thickness) of 8 ......... yes EIGHT!!!!
: (pretty darn thick) The maximum use rating goes DOWN to 1090C (1970ish F) for
This is limited on the basis of *life*. In a 24hr/day industrial situation,
the thermocouple can oxidize into dust in a short time. You don't want to
be replacing industrial thermocouples every few weeks. In a kiln
situation, the thermocouple doesn't spend many hours per week above those
temperatures. That's why they work, and why they are ok in most
situations and give a reasonable life span.
: Next, you have to look at the design accuracy of the manufacturing process to
: start with. ASTM specs that the initial calibration of type K is +/- 2.2C or
: .75% (whichever is GREATER) between 0 and 1200C. That is the BEST it will eve
: be....when it is brand new. At 1200C, that allowable .75% is +/- 9 degrees.
That is the *worst* it will be, for standard error of limits, when it is
new. The alloys are pretty good, and actual errors are considerably
better than this in practice. Special limit of error wire is available
with higher guaranteed accuracy at somewhat higher price.
: cone nine, plus or minus 9 degrees C is about 1/3 a cone (ignoring the heat-wo
: concept for simplicity)!
: Plus, you have to deal with the accuracy of the meter (or controller). Good
: analog meters are only accurate within 2% of full scale to start with. If ful
: scale is 1300C then it is only accurate +/- 26 degrees C, and THAT is of the
: thermocouple output which is also only +/- 9 degrees C!
Digital pyrometers are usually much better than this. Analog pyrometers
are better than peering into the hole and guessing from the color though..
: Additionally, type K is subject to oxidation when used IN AIR above 750C. Thi
: oxidation results in the electrical output changing, and the need for constant
: recalibration. This recalibration requires some electronic equipme
K thermocouples do not, in general, require re-calibration due to
oxidation. The wire just gets thinner, effectively. For a digital control
or pyrometer, *no difference*.
Reduction atmosphere is a Bad Thing and can screw up the sensor.
Thermocouples (type K) need oxygen or to be sealed in an inert
atmosphere.
: In general, for cone 9-10 use, type K is useless for REALLY accurate work at t
: top of the temperature range. They are used on "potters kilns" simply becaus
: they are CHEAP, and most potters won't pay for the platnium/ 10%
: platnium-rhodium (type S) thermocouples and control systems that really should
: be used (and the more expensive plated switches and contacts that entails).
R or S are better for operation at the top end, especially with large
duty cycles (spending a lot of time at the high temperatures). They are
prone to *contamination*, however, and the controls etc. do tend to be
somewhat more expensive (and can be prone to drift) due to the very weak
signals from the precious metal thermocouples. Of course, buying the
platinum alloys and the protection tubes adds to the cost as well.
Sensor design and placement in a complex environment like a kiln involves
many factors.. for example- immersion length and radiant heating of the
sensor.
: Type K simply are more resistant to oxidation than the really cheap types E,
: J,and T at temps over 500C, so they got pressed into use.
E, J, T are useless at much over 1000'F. K (and N) thermocouples are
useful, rugged and work well at higher temperatures. They are *very*
widely used in industry. R and S and B, much less so. For very high
temperatures we have to use Tungsten thermocouples, but the accuracy
isn't good at all.
: So you DO just get used to the fact that the temperature shown is not really t
: temperature in the kiln (pretty bad since that is what you are lead to beleive
Or, more importantly, the temperature of your ware.. on the surface or
internally..
Don't assume that by changing the thermocouple type it will solve all
your troubles either.. the next generation of controls/pyrometers will be
close to or at industrial accuracies, and most of the mainstream examples
are in the ball park today (except those rather questionable analog meters).
: ---------next post -------------
: You got good info from whoever you talked to. Yes, you will need to
"match" the
[very good post on resistance and extension leads deleted, with sadness,
to save bandwidth]
: --------------------end of old messages -------------------
S.P.
| |
|
