Scaling a zero center pressure sensor

ssarora · Sep 13, 2021

I have a zero center pressure sensor where 0.0 is 12mA, 0.5 is 20mA and -0.5 is 4mA. I cant figure out a way to get accurate results

I have attached some screen shots of how I am doing it. But need more help it seems. let me know if I'm missing something.

mylespetro · Sep 13, 2021

It might be worth grabbing one of the SCP (Scale With Parameters) AOIs from the Rockwell Sample Code Library. You would put the input minimum and maximum in raw data counts (not sure what kind of analog card you have, but could be -16383 min, +16383 max for example) and output minimum and maximum (-0.5 min and 0.5 max) and it should handle itself.

boneless · Sep 13, 2021

Easy way would be to scale from 0 to 1, then just subtract 0.5 after scaling.

Ken Moore · Sep 13, 2021

I agree with boneless.
Scale your span, then adjust for zero offset.

Helliana · Sep 13, 2021

(((Scaled Max-Scaled Min)/(Max Raw Input-Min Raw Input))*(Raw Input-Min Raw Input))+Scaled Min = Scaled Value

OkiePC · Sep 13, 2021

I have yet to find a Logix5k project where the SCP Add-on wasn't well worth the Add-in.

Sometimes I think that A/B left it out on purpose to get more folks to look at the shiny new AOI concept, because every program with any amount of analog will want and need it.

busarider29 · Sep 13, 2021

Helliana said:
(((Scaled Max-Scaled Min)/(Max Raw Input-Min Raw Input))*(Raw Input-Min Raw Input))+Scaled Min = Scaled Value

This is a calculation that every PLC programmer should commit to memory. At the very least, have it written down somewhere for quick reference.

As for the SCP instruction, the above calculation is essentially it. Just take that and then make your own SCP with Logix Add-On-Instruction. Function blocks and/or functions for CODESYS/TwinCAT users.

drbitboy · Sep 13, 2021

Summary

Use 0.000034241 instead of 0.00003 in the CPT formula.

TL;DR

In the first image below, which calculates

-0.039519995 for Prime2BoothPressure
From the that datum, plus the supplied CPT formula, i.e.
- 0.00003 * AirBalanceRemote1.0:I.Ch5Data - 0.5014,
we can back-calculate that the raw counts value of AirBalanceRemote1.0:I.Ch5Data on that scan was
- 15396
  - i.e. = (-0.039519995 + 0.5014) / 3e-5

However, OP says that value of Prime2BoothPressure should be 0 at the analog input value of 15396, and the calculation above assumes the 0.00003 scaling factor is correct.

The other images from post #1 suggest a slightly different scaling factor, i.e. an independent least-squares linear fit of the data from the second image below (the eXcel linear fit) suggests the slope parameter is actually

0.0000324241111 = 3.24241111e-05

and not just the

0.00003
- i.e. the "3e-05" that the eXcel chart displays. That would make the CPT formula be

3.24241111e-05 * AirBalanceRemote1.0:I.Ch5Data - 5.01428070e-01

So

-0.50081201 = -3.24241111e-05 * 19 - 5.01428070e-01
0.00163201 = -3.24241111e-05 * 15515 - 5.01428070e-01
0.49918 = -3.24241111e-05 * 30860 - 5.01428070e-01
-0.002226 = -3.24241111e-05 * 15396 - 5.01428070e-01

So both 15515 and 15395 counts yield pressure values within two milli-whatever units the W.C. pressure Prime2BoothPressure is using.

drbitboy · Sep 13, 2021

busarider29 said:
This is a calculation that every PLC programmer should commit to memory. At the very least, have it written down somewhere for quick reference.

I don't want to be rude, but I have to disagree and say it is even more important than that: this is a calculation that every person on the planet should understand in their bones.

That way, they don't have to rely on memory, they can derive it from first principles.

To OP's credit, from their sketches they do appear to have the concept correct in a graphical sense, but they cannot yet translate that to numbers and a formula.

Anyone without that understanding has no sense of proportion.

.

pturmel · Sep 13, 2021

drbitboy said:
has no sense of proportion.

Hah! Indeed.

ojz0r · Sep 14, 2021

Helliana said:
(((Scaled Max-Scaled Min)/(Max Raw Input-Min Raw Input))*(Raw Input-Min Raw Input))+Scaled Min = Scaled Value

Invaluable for sure.
One thing to be careful with is the datatypes and the theoretical range vs. real world range.

I encountered volume flow sensors that were scaled (lets say) 0..20 m3/h - 4..20mA. Some sensors could go a bit below 4mA like 3.9mA which would indicate that there is a small back flow or bad calibration, and the same at the top value 20mA with the value going slightly above to maybe 20.1.
The biggest problem i had was that i used UINT's so when the sensor was at 3.9mA my calculation would roll over to 65535~ish. Rest assured, I will not make that misstake again.

So my point is, don't use unsigned datatypes or put in protection for under/over-range.

busarider29 · Sep 14, 2021

drbitboy said:
That way, they don't have to rely on memory, they can derive it from first principles.

Agreed.

Scaling a zero center pressure sensor

ssarora

Member

mylespetro

Member

boneless

Lifetime Supporting Member + Moderator

Ken Moore

Lifetime Supporting Member

Helliana

Member

OkiePC

Lifetime Supporting Member

busarider29

Lifetime Supporting Member

drbitboy

Lifetime Supporting Member

drbitboy

Lifetime Supporting Member

pturmel

Member

ojz0r

Member

busarider29

Lifetime Supporting Member

Similar Topics