As jseavers says, the raw data from the analog input will range from 0 at 4 mA to 32000 at 20 mA. This corresponds to a range of 0 - 5" water column. Unfortunately, the Scale instruction that jseavers mentions is not available in the 90-30, so you will have to do the arithmetic yourself.
Using Tom's method:
DataOffset = Data in input register at 4 mA (or zero analog signal of whatever range - for example 0 VDC on a 0-10 VDC transmitter)
DataOffset = 0
DataSPan = Data at 20 mA - Data @ 4 mA
DataSpan = 32000
FACTOR = arbitrary factor (multiple of 10) needed to get proper resolution and accuracy
FACTOR = 500
EU = Engineering Units x FACTOR
EUOffset = Engineering Units @ 4 mA x FACTOR
EUOffset = 0
EUSpan = (Engineering Units @ 20 mA - Engineering
Units @ 4 mA) x FACTOR
EUSpan = 500
DATA = Actual data reading in input register
EU = ((EUSpan x (DATA - DataOffset))/DataSpan) + EUOffset
ProcessValue = EU / FACTOR
Example: Analog data value = 16000 (50% of full scale)
EU = ((500 * (16000 -0))/32000) + 0 = 250
ProcessValue = 2.50"
You may find it easier to convert the data value from the analog input to a REAL data type before doing the calculations. Use the INT_TO_REAL instruction and don't forget that REAL variables use two consecutive %R addresses in the 90-30. Be careful that you don't overlap the addresses you assign to variables. The easiest way to avoid overlaps is to let Machine Edition pick the address for you. You do this by entering %R when inputting an address (instead of, for example %R00001). Machine Edition will select the next available %R address for you.