1. Mount the probes from the side rather than the top if possible to prevent overheating from rising hot air.
A little bit of theory here might help.
The output of a thermocouple is proportional to the DIFFERENCE in temperature between the hot and cold junctions. The hot junction is usually the one measuring the process and the cold junction is at the head itself. If the temperature of the head changes then the output from the thermocouple will change even if the temperature of the process has not changed. To compensate for this most temperature transmitters incorporate a cold junction compensating circuit (CJ). The non linearity of the thermocouple characteristic means that this circuit is usually an approximation and has an error usually of the order of 1deg. for an ambient change of 25deg.. This is particularly significant if the process temperature is close to ambient but becomes less noticeable at very high or very low temperatures.
The CJ sensing device must be mounted as close as possible to the cold junction, however in practice it is embedded in a resin and there is usually a small gap between the two. This means that there will be a slight delay in compensation as the ambient temperature changes. This can usually be demonstrated by blowing warm air across the thermocouple terminals.
As mentioned above this effect can be minimised by keeping the transmitter at as constant a temperature as possible (which is a good idea anyway) but basically what it boils down to is that a thermocouple is not a good idea for monitoring temperatures close to ambient if accuracy is required.
A resistance sensor, such as a Pt100, has a resistance which depends on the ABSOLUTE temperature and no cold junction compensation, with its possible error, is required.
As a rule of thumb a Pt100 will always give better results than a thermocouple (and with the correct construction a Pt100 probe can cover the range from -200deg.C to +600deg.C) and is the preferred device for process temperatures within about 200deg.C of ambient.
The disadvantages of a resistance bulb are usually physical strength and, that because it is physically larger than a thermocouple junction, it tends to have a slower response time.
Pt100 elements used to be made by winding
Platinum wire on a ceramic former. These days they are made by deposition
of a film onto a ceramic substrate. The latest Pt100 elements are smaller
than 3mm square by 1mm thick. They are available in various grades, 0.1%
and 0.2% accuracy being probably the most common.
CHOOSING A THERMOCOUPLE
Most articles about thermocouples give a lot of detail about which type of thermocouple should be used or not used in different atmospheres. In practice, as most thermocouples are in sheaths and can be sealed from the environment, this may not be as important as it sounds. If in doubt get advice from your local supplier.
The following information is a guide for use with temperature transmitters and does not take these effects into account.
Bear in mind that thermocouple errors, usually because of variations in the composition of the alloys used, tend to be larger than any error in the electronics.
Type J - Iron/Constantan
Large output - Approx. 52microV/deg.C.
Useful range +20 to +700deg.C continuous.
-180 to +750deg.C intermittent.
Reasonably easy to linearise over any part of range. Best choice for lower spans as larger output implies lower drift error on instrument.
Type K - Chromel(NiCr)/Alumel(NiAl)
Medium output - Approx. 40microV/deg.C
Useful range 0 to +1100deg.C continuous.
-180 to +1350deg.C intermittent.
Very easy to linearise up to about 500deg.C and over about 800deg.C but the characteristic does a bit of a dogleg between these two regions. This is not normally a problem, however recalibrating an instrument from say 0-600deg.C to 0-700deg.C could give linearity errors.
Type T - Copper/Constantan
Useful range -185 to +300deg.C continuous.
-250 to +400deg.C intermittent.
Type R - Platinum/Platinum 13% Rhodium
Low Output - Approx 10microV/deg.C
Useful range 0 to 1600deg.C continuous.
-50 to 1700deg.C intermittent.
Type S - Platinum/Platinum 10% Rhodium
Similar to type R but slightly lower output. The same comments apply.
Type B
- Platinum 6% Rhodium/Platinum 30% Rhodium
Very low output - Approx 6microV/deg.C
Useful range +100 to 1600deg.C continuous.
+50 to 1750deg.C intermittent.
Practically no output below 400deg.C. Very difficult to linearise, in fact just about impossible below 500deg.C. Recommended for high temperature work only. Pick a range with a large zero offset for best accuracy eg. 700 - 1700deg.C.
Type N - Nicrosil/Nisil
Medium output - Approx. 38microV/deg.C
Useful range 0 to 1100deg.C continuous.
-270 to +1350deg.C intermittent.
This is a fairly new type with characteristics
similar to the type K. Both alloys contain silicon which improves long
term stability. Possibly the way to go in future.
TO ISOLATE OR NOT TO ISOLATE
The majority of head mounting temperature transmitters are non-isolated, which means that there is an electrical connection between the sensing element and the 4-20mA loop.
In most cases the sensing element is contained within a sheath and is surrounded with an insulator such as aluminium oxide. In such cases the element can not come into contact with anything which may be connected to the loop and isolation is therefore unnecessary. Exposed sensing elements also normally cause no problem if used on individual loops.
The major problem occurs when several transmitters are used with one receiving instrument such as a PLC. If grounded tip thermocouples, or other exposed sensors are used with a single power supply then isolation is probably a must, however it could be less expensive to look at three alternatives to buying an isolated transmitter.
1. Replace the sensors with insulated types.
2. Use individual power supplies for each loop.
3. Use a separate isolating module mounted in the control panel.
WARNING. Some ceramic type thermocouple
sheaths become electrically conducting at high temperatures, giving intermittent
ground loop problems on multiple installations.
RADIO FREQUENCY INTERFERENCE
Any instrument measuring PT100 or thermocouple signals is prone to receive interference from radio signals because the probe leads make nice antennae and they are designed to amplify low level signals. Careful design can minimise these effects however the use of walkie talkies or similar devices in close proximity to temperature transmitters is definitely to be avoided.
A well designed temperature transmitter will often show little effect with 2 watt Walkie talkies and cellular telephones at distances of 500mm but 1 metre is the recommended closest distance.
Note that IEC803 and IEC1000 specifications refer to a maximum field strength of 10V/m which is roughly equivalent to 2 watts at 1 metre. Normal RF testing is performed in a linear field however transceivers closer than about 1 metre are likely to generate a field with a large gradient and this is what causes the problems.
Screening in metal control panels can often be a solution to RF
pickup in instrumentation. Ferrite beads on the thermocouple leads and
a small non-inductive capacitor (100pF is often a good starting point)
across the instrument terminals can give dramatic improvements at UHF.
LONG TERM CONSIDERATIONS
Most manufacturers attempt to minimise this effect by accelerated ageing during which process the units are heated and cooled repeatedly over a period of 24 to 48 hours. This is a big help but does not totally eliminate the problem.
Generally speaking the drift in calibration with ageing is at its worst when new and decreases as the instrument ages. Obviously user requirements for accuracy vary and it is difficult to be specific, but it is good policy to re-calibrate a unit after it has been in use for perhaps three months, again about six months later and thereafter about once a year. If the transmitter is subject to wide extremes of ambient temperature, more frequent re-calibration may be necessary.
When calibrating a temperature transmitter the usual considerations should be taken into account, such as correct polarities for compensating cables for thermocouples and corrections for lead resistances for Pt100. Note that for optimum overall tracking accuracy the calibration may be slightly offset at zero and span.
If the process which is being monitored only covers a relatively narrow temperature range then the calibration can be optimised over this part of the range of the instrument rather than trying to obtain a good accuracy over the whole range.
I hope this information will enable users of temperature transmitters to make a more educated choice and to anticipate where problems could arise.