R

#### Richard

- Jan 1, 1970

- 0

things, but what I'm fighting with is the analog section. Standard 0-20ma

inputs and , Thru a Precision 250 ohm resistor to convert to 0-5 volts for

a ADC Card.

ADC Card is a 16 channels single ended 12 bit analog.

Problem is, that no matter what the source is, the voltage always seems to

be approx. 50mv lower than actual. I have an Omega MA Simulator in order to

test this. If I send 4 ma then the ADC Card reads 0.950 volts and 20 ma =

4.950 volts.

It appears to be pretty linear. Not sure what is causing it, because I have

not seen this on my other boards designed with 250 ohm resistors.

Anyway, since it's linear, and we built abunch of these boards, I need a

correction factor in my scaling in order to make the engineering units

correct. I would like to scale it on the voltage side since we know the

offset is the same on all the channels and not offset the scaled value.

I have the following for each channel:

Factor and Offset which works normally (When I don't need an offset)

Vin * Factor + Offset

I can't add the millivolt difference to Offset because offset is adding to

the scaled value. But I could add it to Vin like this....

CrFactor = 0.50

((Vin + CrFactor) * Factor) + Offset

The above works for Sensors 4-20ma, because we are actually scaling 1 - 5

volts or 0.950 to 4.950 and by adding the correction factor it brings it

to 1.00 to 5.00

Follow?

Okay, But now, we actually have a sensor that is 0-20ma or 0-5volts, the

above formula and correction factor does not work because Zero volts is

already zero volts (no more voltage drop) so the actual voltage shown is

0 to 4.950

Using the above correction factor, we end up with 0.050 to 5.000 which is

incorrect and 0 to 4.950 is also incorrect.

Would it be 0 - 4.900 ?

Not sure, Anyone have an idea how I can handle this voltage drop for all my

channels?

My software "could" look at MinVoltage and MaxVoltage and determine if the

scalling starts at Zero or Negative and use a different formula for each

case?

Richard