The SCP instruction is just an easier version of the SCL instruction. They both do the same thing: y=mx + b
All you're doing is figuring out a line slope. The limits of that line need to be set with other instructions.
If you're getting overflow errors, you probably need to put the value into a larger-bit address. If you're using an N address, move it into an F address. F addresses are 32-bit and can handle any number from -2,147,483,647 to 2,147,483,647 as opposed to -32767 to 32767.
If you want to limit the output of the SCP, simply add compare and mov instructions after the SCP. If for example, you want to SCP to "Clamp" at 100. On another rung or branch below the SCP, do a GRT instruction to see if it's over 100, and if it is use a MOV instruction to move 100 back to the value. Of course, this won't help with your overflow problem. The internal calculation of the SCP will inevitably result in a number than overflows the address before the MOV can correct it, so I suggest if you're using a 16-bit address for the SCP result, change it to a 32-bit address.