4-20mA Analog Output

It depends on the device you've connected. I've never seen a mA analog module that is bipolar, but they may exist. There are definitely Voltage output modules that can go + or -.

The trick is that, for any analog FieldBus module (either input or output), the analog value is converted into a digital equivalent for transmission over the bus. Just how this occurs depends on the specific module.

For example, a 12bit 0-10V module (unipolar) might ignore the sign bit entirely, or incorporate it into the entire I/O word. An unsigned 12bit value maxes out at 4095, so the resolution of the module would be 2.5mV.

OTOH, a 12bit -10 to +10 module from the same manufacturer would use the 12th bit as the sign bit, meaning the integer word would swing from -2047 to +2047 (with some caveats for different encoding schemes), so the resolution would be 4.8mV.

The math works the same for any analog module attached to a KRC's ANOUT or ANIN. The encoding differences for the different bit ranges and voltage/current ranges is handled at the bus/module configuration level. When set up properly, the full scale of the analog module (5V, 10V, 20mA, whatever) should map to a simple 0-1.0 or -1.0 to +1.0 ANOUT/ANIN value.

Basically, you can treat the ANOUT/ANIN values as a % of the maximum signal of the analog module. A 10V module should put out 5V for an ANOUT of 0.5, and a 5V input module should report an ANIN of 0.5 when detecting 2.5V.

Sending a negative value to a module that accepts only positive values could have unpredictable results. Not "blow up" unpredictable, just mathematically unpredictable. It depends both on the module, and on exactly how you configure the module's conversion factor in IOSYS.INI or in the I/O mapping of WorkVisual.

Well, when you grouped the signals in WV, did you set them as I/O, INT, UINT, DINT, etc?

For some bipolar devices, the MSB is the "sign" bit and the other bits are "magnitude", so (for a bipolar 16bit device) a binary value of 1000000000000001 is the same as 000000000000001, just opposite polarities. Assuming a +/-10V device, this would be -0.009mV and +0.009mV, respectively.

However, some devices encode differently -- 1000000000000001 would -10V, and as the binary value "below" the sign bit gets bigger, the analog value decreases, so 1111111111111111 would be -0.009mV.

That's what I meant about "unpredictable" behavior -- if you send a negative value to your unipolar device, just what would come out would depend on which encoding it was expecting you to use.

For testing, I usually just make a program that increments the ANOUT by one bit at a time (I usually map my analog devices as digitals, so I can directly control the "raw" binary), and record the output on a multimeter at each step. Then I plot both values against each other in a scatter plot in Excel, or something similar.

For analog inputs, pretty much the same, except I wire a potentiometer into the input and slowly dial it up/down, while recording the ANIN and the multimeter reading, for a similar plot.

That will usually show me pretty quickly how the combination of the device and WV configuration behaves.

Yeah... the WV docs don't go into any detail concerning exactly how SINT/USINT/DINT/etc are defined at the binary level. And I usually didn't want to be bothered with it, so I usually went for "raw" I/O and did the encoding/decoding in KRL, did everything in "digital," and avoided the ANIN/ANOUTs almost entirely. Probably not the most elegant way to proceed, but I'm an Old Fart.