Vous êtes sur la page 1sur 6

Comparing Transmitter

Performance Using Total

TPE
Probable Error

Reference-65 Reference-65
Comparing Transmitter Performance Using Total Probable Error

INTRODUCTION This does not mean that every specification on a


data sheet needs to be included in a comparison.
When choosing transmitters, how do you know
What to include depends entirely upon the
which one will be the best for the job? Will a
application. The important point is to be consistent
particular type of transmitter fit every application?
and to compare the same set of specifications for
Will its performance be consistent from one
each transmitter.
application to another? How do you decide?
The directory of the Instrument Society of America Reading the specifications
lists 66 manufacturers of electronic differential Once you have all the pertinent specifications, you
pressure transmitters. This is a significant number need to compare them equally among the various
TPE

of transmitter specifications to review. transmitters. Equal comparisons require that you


Along with the basic function and features of the need more than the numerical portion of the
transmitters, the accuracy specification is the most specification. You need to read the wording that goes
common one used to narrow the choices. To choose along with the number. Is the specification
transmitters in today’s demanding marketplace, you expressed as 0.2% of calibrated span, 0.2% of URL,
need to look critically at other specifications and be or 0.2% of reading? Per 50°F, per 50°C, or per 100°F?
able to compare them on equal terms between For example, suppose you are looking at a 0.2%
various transmitters. With the choices for accurate transmitter that has an Upper Range Limit
transmitters ranging from low-cost throwaway (URL) of 150 inH2 O. You will be calibrating it to
transmitters through smart transmitters, 0-100 inH2 O and typically reading it at 80 inH2 O. If
consistently comparing various transmitter the transmitter’s accuracy specifications is read as:
performances for a particular application becomes
even more important in the selection process. 0.2% of calibrated span, then the reading will be
80 ± 0.3 in H2 O (0.2% of 100 inH2 O = 0.2 inH2 O)
A method for doing a consistent performance
comparison between various transmitters is to do a 0.2% of URL, then the reading will be 80 ±0.16
total probable error analysis based on the published inH2 O (0.2% of 150 inH2O = 0.3 inH2 O)
performance specifications of each transmitter. This 0.2% of reading, then the reading will be 80
analysis gives a reliable indication of the ±0.16 inH2 O (0.2% of 80 inH2 O = 0.16 inH2 O)
performance that can be expected of a transmitter.
As you can see, there is quite a bit of difference
Although both reliability and performance of a
between these transmitters which all have “0.2%
transmitter are dependent on the design, they are
accuracy.” The only time these errors are all equal is
not interchangeable and reliability should be
when the transmitter is calibrated and read at the
considered separately from performance.
upper range limit.
Which specifications are important?
In order to determine which specifications are
important in the transmitter comparison, you need
to look first at the application and determine to what
±0.2% of
conditions the transmitter will be exposed. Will the Reading
process temperature or the ambient temperature
Milliamps

vary? Will a differential transmitter be exposed to


high line pressures? Will the line pressure vary? Will
the power supply vary a great deal and if so, will it
affect the transmitter’s output? Will the transmitter ±0.2% of Calibrated Span
be located where it will be exposed to a lot of
vibration? Will output noise be a problem? ±0.2% of Upper Range Limit

These are all factors which can have an effect on Pressure


the transmitter’s performance. These are also
factors which commonly will have a performance FIGURE 1. Error as Percent of Span or Reading.
specification written around them. If a
manufacturer does not list a certain specification, it What are the different types of errors that
does not always mean the transmitter is unaffected
may be seen?
by that condition. Call the manufacturer and ask
how the transmitter performs under that condition if When reading specifications, you may find them
it is felt to be a critical parameter of the application. expressed as a variety of errors. These are all
reflections of how the performance is affected. The
three types of error commonly used in data sheets
are: zero error, span error, and systematic error. An
unmentioned one is turndown error.
Reference-66
Rosemount Inc.

Zero error is a shift of a constant magnitude Error sources


between the measured variable and the ideal variable. Sources of errors for the transmitters are found in
It is normally measured at the zero reference point. the process conditions. Temperature and static
With this error, the starting point of the measured pressure are the two most common error sources and
curve is offset slightly and thus causes the entire the largest ones. They will be discussed below. Other
curve to be offset by an equal amount. error sources which occasionally occur are vibration,
power supply changes, and RFI. Any of these can
4–20 mA OUTPUT affect the overall performance.
A high static pressure can create some
4–20 mA OUTPUT performance errors. Often, the zero effect can be

TPE
calibrated out under a stable line pressure, leaving
4–20 mA OUTPUT only the span error, which may or may nor be
systematic. When static pressure varies, however,
Ideal Span the zero error becomes more important and should
0 to 100 inH2O input be considered in the overall performance.
Span error is the difference between the actual span Temperature error can result from both process
and the ideal span. With this type of error, the actual and ambient temperature changes. Generally,
span may be slightly larger or smaller than the ideal process temperature is usually stable while ambient
span. The variables within the curve then would be fluctuates. It is not always simple to determine
proportionately larger or smaller. which one causes the temperature effect errors. It is
a fairly simple matter though, to re-zero a
4–20 mA OUTPUT transmitter after it has reached operating
temperatures, assuming the process temperature is
4–20 mA OUTPUT constant. This would eliminate the zero error and
leave only the span error. The transmitter is still
4–20 mA OUTPUT
subject to ambient temperature changes.
An example of the effect to temperature and static
Ideal Span pressure is as follows:
0 to 100 inH2O input
On the bench, a transmitter is calibrated 0 to 100
Zero and span error are sometimes expressed together inH2 0 at 75°F and no static pressure. The accuracy is
as a total error. This implies a maximum amount of ±0.2% of span giving an error of ±0.2 inH2 0. The
error that the measurement could have. It does not transmitter will be used at a line pressure of 1500 psi and
necessarily mean that the two errors will be both in in a location where the temperature may vary by 50°F.
the same direction, nor does it mean that they are
divided equally. The error contributed by a 50°F temperature
change from calibration conditions is predicted by its
A systematic error is one that occurs during a
temperature effect specification of ±1.0% span/
number of measurements made under the same
100°F. This yields an additional error of ±0.5 inH2 0
conditions and to the same magnitude. It is a
(1% of 100 inH2 0 50/100)/ The error contributed by a
predictable, repeatable error.
line pressure of 1500 psi is predicted per 1000 psi.
Turndown error can occur when the transmitter is This yields another error of ±0.37 inH2 0 (0.25% of
used at a span other than its maximum span. It can 100 inH2 0 1500/1000). If we were to add all these up,
be expected to occur when the specification is there is a worst case error of:
expressed as a percentage of upper range limit (URL)
Error allowed within accuracy limits:
or a percentage of maximum span. Turndown error is
±0.2 inH2 0
proportional in magnitude to the amount of turndown.
It is a constant error along the whole range of the + error contributed by 50°F temperature shift:
measurement. It is the same for all readings. ±0.5 inH2 0
If an error, such as temperature effect, is + error contributed by 1500 psi static pressure:
expressed on a data sheet as 0.2% of URL and the ±0.37 inH2 0
URL of the transmitter is 150 inH2O, then at 150
= Worst case error
inH2 O, the error would be at 0.2% of URL (150) or
±1.07 inH2 0
±0.3 inH2 O. This equates to 0.2% of the actual
reading. If the same transmitter is read or calibrated This error is much worse than the expected
at 50 inH2 O, the error would still be ±0.3 inH2 O, but accuracy of ±0.2 inH2 0. A worst case error (total
at that reading, the error is ±0.6% of the reading. error) assumes that the errors will all be at the
maximum amount in the same direction. This is
Manufacturers may use any or all of the above
unlikely to happen and would not represent typical
terms when citing transmitter specifications; thus it
performance of a transmitter.
is imperative the entire specification is read and
understood when evaluating transmitters.
Reference-67
Comparing Transmitter Performance Using Total Probable Error

Overall Performance variables (specifications) are independent of each


In order to tie all these specifications together in a other and that the mean of each error term is zero.
meaningful and complete term, an overall When combining all the error factors, the units of
performance profile of the transmitter needs to be measure must all be the same. For example, compare
determined. This is done by using a root-sum-square the two transmitters shown in the example
technique to determine a total probable error. The calculation below.
total probable error is a more realistic number than Using the root-sum-square method, it becomes
would be obtained by simply adding up all the evident that Transmitter A would be better choice for
possible errors, since it is unlikely that all the errors that application even though the numbers associated
would go in the same direction (+ or -) from their
TPE

with Transmitter B appear better. If the transmitters


means. The root-sum-square (RSS) method were to be used at the upper range limit, then
determines the total probable error by summing the Transmitter B would be the better choice (1.56% for
squares of the individual errors and taking the Transmitter B vs. 1.67% for Transmitter A)
square root of the total. It is assuming that the

List Transmitter Specifications


SPECIFICATIONS TRANSMITTER A TRANSMITTER B
Upper Range Limit (URL) 300 inH2O 300 inH2 O
Accuracy 0.2% of span 0.1% 0f URL
Temperature Effect
Zero 0.5% of URL per 100°F
Span 0.5% of span per 100°F
Total* 1.0% of URL per 100°F
Static Pressure Effect
Zero 0.25% of URL per 2000 psi 0.25% of URL per 2000psi
Span 0.25% of reading per 1000 psi 0.25% of span per 1000 psi
Total*

Define Operating Conditions


Calibrated Span 0 to 100 inH2O
Expected Temperature Change 50 °F
Expected Static Change 500 psig
Expected Reading 75 inH2 O

Convert all of the errors into common terms:


SPECIFICATIONS TRANSMITTER A TRANSMITTER B
Accuracy 0.2% 3 100 = ± 0.2 inH2 O 0.1% 3 300 = ± 0.3 inH2 O
Temperature Effect
Zero 0.5% 3 300 3 50/100 = ± 0.75 inH2 O
Span 0.5% 3 100 3 50/100 = ± 0.25 inH2 O
Total* 0.75 + 0.25 = ± 1.00 inH2O 1.0% 3 300 3 50/100 = ± 1.5 inH2O
Static Pressure Effect
Zero 0.25% 3 300 3 500/2000 = ± 0.19 inH2 O 0.25% 3 300 3 500/2000 = ± 0.19 inH2 O
Span 0.25% 3 75 3 500/1000 = ± 0.094 inH2 O 0.25% 3 100 3 500/1000 = ± 0.12 inH2 O
Total*

__________________
Calculate Total Probable Error (TPE =√ A2 + B2 + C2...)
TRANSMITTER A TRANSMITTER B
________________________________ ________________________________
√ (0.2) + (1.00) + (0.19) + (0.094) = ± 1.04 inH2 O
2 2 2 2
√ (0.3) + (1.5) + (0.19)2 + (0.12)2 = ± 1.55 inH2 O
2 2

TPE = ± 1.04% for 100 inH2 O span TPE = ± 1.55% for 100 inH2 O span

*When one manufacturer gives only a “total” specification, and the other manufacturer breaks it down between zero and span, only the total values should be
used in the TPE calculation. Zero and span should be calculated and then stated. If both manufacturers give zero and span specifications, then the individual
values should be used in the TPE calculation.

Reference-68
Rosemount Inc.

Temperature Transmitters
Total Probable Error can also be used to compare performance of temperature transmitters. The same
principles of reading the specifications apply to the temperature transmitters as well. The most prominent
source of error is the ambient temperature’s effect on the electronics, yet vibration and load could also have
some influence.
The temperature transmitters should be evaluated independently of the temperature measurement device
unless one is interested in doing a “system analysis” of the probe and transmitter combined. The probe has its
own accuracy limits.
An evaluation of two temperature transmitters may consist of the following:

TPE
List Transmitter Specifications
SPECIFICATIONS TRANSMITTER A TRANSMITTER B

Accuracy 0.2% of span 0.1% of span


Temperature Effect
Zero ±0.3 °C per 100 °F ± (0.004 degree + 0.002% span) per degree
Span ± 0.4% of span per 100 °F ± 0.002% of span per degree
Total (or other) ± 0.15% of base temp., °C, per 100 °F

Define Operating Conditions


Calibrated Range 75 °F span, from 75 to 150 °F
Ambient Temperature Change 100 °F

Convert all of the errors into common terms:


SPECIFICATIONS TRANSMITTER A TRANSMITTER B

Accuracy 0.2% 3 75 = ±0.15 °F 0.1% 3 75 = ±0.075 °F


Temperature Effect
Zero ± 0.3 °C 3 9/5 = ± 0.54 °F ± (0.004 + 0.002% 3 75) 3 100 = ± 0.55 °F
Span ± 0.4% 3 75 = ± 0.30 °F ± 0.002% 3 75 3 100 = ± 0.15 °F
Total (or other) ± 0.15% of [(75 – 32) 3 5/9] = ± 0.04 °F

Calculate Total Probable Error (TPE = √ A2 + B2 + C2 ...)


TRANSMITTER A TRANSMITTER B

√ (0.15) + (0.54) + (0.30) + (0.04) = ±0.64 °F


2 2 2 2
√ (0.075) + (0.55)2 + (0.15)2 = ± 0.57 °F
2

TPE = ± 0.85% for 75 °F span TPE = ± 0.76% for 75 °F span

Summary
Transmitter specifications are often confusing to read and can be deceptive if not read carefully. In order to
evaluate different transmitters, the specifications and their impact on
_________________________________ performance must be understood. The
__________________________
specifications can be incorporated into a total probable error analysis allowing a more complete and accurate
comparison to be made between transmitters. The total probable error analysis not only assures that
transmitters are evaluated equitably, but also provides a more accurate picture of how a transmitter can be
expected to perform under specific conditions or changes in conditions.
Although total probable error can be a valuable part of an overall transmitter comparison, it should not be
the only aspect of comparison. The reliability of the transmitter to continue to perform may be more desirable
than the actual level of performance. The degree of performance, reliability, the appropriateness of the
transmitter to the application, the features included and the overall cost of ownership should all contribute to
the final decision.

Reference-69
TPE

TRANSMITTER #1 #2
Model Type

Reference-70
Transmitter Range
Calibration

LISTED SPECIFICATIONS
Accuracy:
Temp. Effect
Zero:
Span:
Total:
Static Press. Effect
Zero:
Span:
Total:
Other Specs (List)
1.
2.
3.

CONVERT ALL LISTED SPECIFICATIONS TO ACTUAL VALUES


Accuracy:
Temp. Effect
Zero:
Comparing Transmitter Performance Using Total Probable Error

Span:
Total:
Static Press. Effect
Zero:
Span:
Total:
Other Specs (List)
1.
2.
3.

TOTAL PROBABLE ERROR CALCULATION


√ A2 + B2 + C2 + D2 + . . . .

List any assumptions made for each mfg.

Vous aimerez peut-être aussi