# Validation of transmitter signals

P

#### P.Samuthira Pandi

I have three PTs to measure steam pressure in boiler. I have to take this signal to control loop. I am doing that by averaging the signal. I want to know that how to eliminate the faulty signal out of three.

Can anyone help me?

PSP

N

#### Nirbhay Gupta

The simplest way to validate the signals is to use the median of the three pressure transmitter signals. This way you will be able to discard one faulty signal. The procedure to make a median selector is simple.

First group three signals (say A, B and C)into three pairs. (Say AB, AC and BC). Pass each pair of inputs to the low selectors (Three Low selectors). Pass the outputs of these three low selectors through a High Selecor. Presto It is the median value of the three inputs viz. A, B and C.

All the best.
Nirbhay

D

#### Daniel Chartier

Hello PSP;

How do you determine that a signal is faulty?? Do the PTs have underrange/overrange detection? If so, use the signal in your logic to do your calculations without this signal. If you do not have a range detection in the PTs, use a compare instruction (GEQ, LEQ) to determine the validity pf all 3 signals; them implement the adequate logic.

Hope this helps,
Daniel Chartier

P

#### Peter Green

There is an algorithm to do this. It is credited to Sargent and Lundy. You could do a web search for these people or if you can send me your e-mail address, I can send you a function block implementation.

Regards

Peter Green
[email protected]

S

#### sreid

The name for this type of function is "Voting Logic." Basically, signals are checked to see if they are within a percentage range of each other. For three signal, if one is out of range, it is thrown out; the other two signals "Out Vote the third signal."

R

#### Robert Scott

How about using the median of the three readings? That will be better than averaging if you suspect that a transducer can suffer a catastrophic failure. Also, you can validate readings by comparing them to previous readings. If you process does not generally produce sudden jumps in pressure, then you can use a sudden jump in the reading as an indication that something is wrong with that reading.

Robert Scott
Real-Time Specialties
Embedded Systems Consulting

S

#### Steve Myres

Look for two of the three signals to be more similar in value and rate-of-change to each other than either is to the suspect sensor. Eliminate the third signal. Don't forget to annunciate an alarm so the third one can be fixed before one of the two good ones fails.
--
Steve Myres, PE
Automation Solutions
(480) 813-1145

Y

#### Y.K.JARIWALA

Dear PSP
Earlier there was product series called Split architecture, normally used in power plants
& Marine industries or cargo ships, They had
standard hardware module called 2 of 3 voting.

The logic was , taking average of 3 inputs
compare each input with average of 3
decide the dead band [ eg 3% ],ANY of the transmitter devaites it is isolated & average
of 2 signal is taken as process input.
when second fails [ criteria by user ], loop
goes into manual mode.

I think Taylor instruments , Foxboro & baily
used to supply this kind of products & approved
by LLOYDS

Jari
[email protected]