1% of Full Scale

[xnetfly]

So I have been reading specifications for some time now and the controls language still seems to baffle everyone. Can someone tell me what the exact meaning of 1% of full scale. Let's say I have a 50-100F temperature transmitter. If I measure 57 and the sensor is reading 55 that is a 2 degree absolute error. So when you 2F divided by 100F you get 2%. This is greater than 1% and is out of cal in accordance to the requirement of 1%of full scale. Does this mean that you take 2F divided by 100F and get 4%. Just trying to figure out what 1% of full scale means.. Thanks for the info

 

[lacajun]

1% of full scale means your maximum error from the transmitter should be no more than +/- 0.01 times the full scale value of the transmitter. If it is greater than that, your transmitter has drifted and needs calibrating. In your example, the transmitter output can be 54 deg. F - 56 deg. F, which is 55 deg. F +/-1 deg. F. If the transmitter has not drifted, any reading between 54 deg. F and 56 deg. F is within the tolerance of the transmitter.

Be certain of your measurement against the temperature transmitter. Make sure the installation is right, if it's a new installation. If the element is a thermocouple in a well inserted into the pipe, the transmitter is probably OK. I don't know the history of that loop so I can't say more. The process engineers I've worked with wouldn't worry about that loop yet, if your measurement is correct. Yours may worry about it so ask them.

Contrast that spec with 1% of reading, which is more accurate. I've always looked for % of reading rather than full scale.

I don't quite grasp your calculation of 2 deg. F divided by 100 deg. F to get 4%.

I'm sure others will have better ideas.

Pamela K. Quillin, P.E.
Quillin Engineering, LLC

 

[danw2]

Full Scale (FS) is typically used on pressure gauges.

A 0-100 PSI gauge would have a ±1psi error, which is huge error as a percentage of the reading at the lower end of the scale, say at, 3 psi, ±1psi, the error is 33% of the reading. But at 85 psi, ±1 psi, the percentage error is far less.

Temperature transmitters, on the other hand, are frequently spec'd for error in percentage of span, where span is Upper range value (URV, the 20mA eng value) minus Lower Range Value (LRV, the 4mA eng value), because the lower range value (LRV) is frequently not at zero like the typical pressure gauge whose bottom range is zero.

A 50° to 100° measurement output range is a 50° span (100°-50°). If you have 2°error for a 50° span, I would say that you do indeed have a 4% error, 2/50.

If your comparison is to a traceable standard, I'd say something is out of whack.

FYI, thermocouples themselves have a fairly wide range of uncertainty, out-of-the-box, per the ANSI standard that states standard and special limits-of-error. For example, a type K standard limit of error in the 0 to 293°C range has an acceptable uncertainty of ±2.2°C. The special limit-of-error is ±1.1°C for the 0 to 275°C range.

 

[lacajun]

Darn, I am rustier than I thought! :-(

Pamela K. Quillin, P.E.
Quillin Engineering, LLC

 

[IRstuff]

While all of the math above is essentially correct, measuring temperature can sometimes be tricky. What is the accuracy of your independent measurement? When, where, and how, was it done, relative to the other measurement that you are questioning? I think you have to do the stack up quite carefully to ensure that you are not muddying the waters. One obvious issue is that you've posted integer readings, whereas real measurements generally have fractional portions as well. I can certainly see a case to made that one sensor was rounded up and the other was rounded down to artificially increase the apparent difference. Unless you are using temperature standards and a reference thermometer with less than 0.1° uncertainty, a more detailed analysis is dictated.

TTFN
faq731-376

 

[xnetfly]

Thanks so much for the replies and it does help.