Analog scaling difficulty

I have tried to solve this requirement on my own, but have not made progress all day.

The requirement is to place the scaled output from a Newport PD-25-TC process controller in an accessible register of a GE 90/30 311 plc.

The 90/30 does NOT support FPD.
The DP-25 has scalable analog Current output
This is a Temperature application over the range of -30C to +100C
The DP-25 provides set points and PID functions for an existing process tool. The data in the plc is for other process functions. The DP-25 is performing it's function PERFECTLY, There is no desire to replace the device in the process system, Only to have it's data in the plc.

PD-25 scaling is set for 4mA =-30C, and 19.99mA=100C
Output is enabled, and set to Current

An IC695ALG221 Analog input card is in the 90/30 and set without connection jumpers, thus enabling 4-20mA input. According to GE, this card sends 0-32000 counts to %AI over the full input span 4-20mA

Logic is
raw data in at %AI1 MOVE to %R1
%R1 SUBTRACT DINT 7385 (intercept) and output to %R3
%R3 MULT DINT by 13000 (130 scaled span keeping xx.xx significant digits)
output to %R5
%R5 DIVIDE DINT by 32000 (raw span) output to %R7

%R7 should be the desired number. But it's off but 1.6 degrees, quite constant over a test range of 0 C (ice water) and around 35 C .

I don't see where the scaling conversion is failing me..If I could I would correct it ;-)

The useful tutorial presented here http://www.plctalk.net/qanda/showthread.php?t=15069 has been studied at length.

The 7385 intercept was confirmed using the handi spreadsheet provided in the final post to that thread. Thanks!

I still must be missing something.

Second thoughts and opinions welcome!

Cheers

Cal

Controller reference"
http://www.omega.com/manuals/manualpdf/MQS3732.pdf

I took an example at 12mA (35.0°C), did your calc and found the error to be insignificant at 0.01°C

When you say "off by 1.6 degrees, quite constant over a test range of 0 C (ice water) and around 35 C", it sounds like that's an empirical test, the result of using an ice bath and seeing the resulting error between the readings on the controller and PLC, not from calculating the PLC value and coming up with an error.

The empirical test error could be
- analog output calibration
- analog input calibration
- combination of the two above
- ground loop offset

Here's my calc:
-30 to 100 = 130 Deg C span
1/2 of 130 Deg = 65 Deg C
-30° + 65° = +35 Deg C
35 Deg C = 12mA

The 'intercept' is the percentage of the total count span in the negative range
-30°C to 0.0°C = 30° range of negative values
range of negative values/span = 30/120 = 0.2307 or 23.07% of span.
23.07% of 32000 counts = 7,384 counts
12mA = 16,000 counts on a 0-32,000 scale

16,000 counts minus 7385 counts = 8615 counts

(8615 * 13000)/32000 = 3499.84
Somehow you move the decimal point: 3499 = 34.99 Deg C, error = 0.1 Deg C from 35.0°C

Your algorithm looks good to me.

Thanks for having a look

Dan,

Your considerations are appreciated!

The maths are not so much questioned as the set up.

When the cold junction calibration was being performed, I noted that the intercept was greater than the 7385 calculated value. More like 7685 counts for 0C.


The exact value is not so important, but may give a clue as to which aspect is being missed. IF the intercept is adjusted, the linearity goes out visibly even over this small range.

ps

The decimal point is not moved, and does not display, it is implied. No FPD with this plc. But without the 13000 multiplier, there are only two figures in a 32 bit register ;-)

The error is 1.6/130 = 1.2%.

There's a calibration error somewhere.

You can fix it by changing the intercept, or you can calibrate the analog ouput or analog input.

Statistically, I'd suspect the retransmit output on the controller first. I'm not sure those are even calibrated at the factory. In fact, when a re-transmit card is field installed, a Honeywell controller will not 'recognize' the card until a calibration is performed, which means to me that the D/A on the card lacks any factory setting. Honeywell unbundles the task of calibration onto the end user. Maybe it's unique to Honeywell or maybe it's a way of cutting costs across the industry . . .

Dan