@g-andresen wrote:
Hi,
Please illustrate this statement with an example.
günther
EDIT 03 APR 2024: I have since settled on a different formula for theta:
acos(Scale_Length / ( sqrt(( Scale_Length ^ 2 + ( ( String_Width_Saddle - String_Width_Nut ) / 2 ) ^ 2 )) )) * 2
Here is a calculation of the angle of the sides of a quadrilateral, in this case a trapezoid. It is the angle that the E strings of a guitar are from each other, as they are not parallel. The formula is:
180 deg - ( 2 * atan(Scale_Length / (String_Spacing_Saddle - Spring_Spacing_Nut / 2 )))
In the first image, I have successfully created this parameter using the above formula, with unitless parameters as inputs:

In the next image, I have replaced with formula with 0, and then added units to two of the inputs. I then pasted in the same formula:

This was overlaid, but was not included in the screenshot:

Other attempts with a different strategy also failed. I first attempted to simply add units to the unitless parameters shown earlier, but they were locked out. Normally a down caret and a drop down menu appears if you hover over the unit, but this did not appear. This also seems to happen to the first parameter added, whether it was used in a formula or not; it is forever locked out of changing units.
In the following image, a formula without trig functions was successfully created using parameters with units as inputs. Also, note that units are mixed. Additionally, the parameter "Scale_Length" is FOREVER locked to inches as units. Whether other parameters depend on it as an input or not, it cannot be changed. This is something I have been able to repeat across new, blank designs.

In the following image, I removed the units from the formula "Fret_01", and is no longer valid.
Note the error message:

In other words, parameters are not consistent or robust. I haven't shown this, but it is impossible to create a new unitless parameter using an existing parameter with units as an input, as a workaround. This means if I foresee possibly needing Trig in my model, I have to start out of the gate with unitless parameters ONLY. It's weird.
This makes it very difficult to work with mixed systems, such as a guitar neck with a 24-inch scale length but modeled in metric, or any number of other scenarios. It is far simpler and more robust to model with native units, for example a 650mm scale length of a classical guitar fretboard and a 25-1/2" scale length of a Fender Jazzmaster fretboard, as configurations of the same model. With the current limitations I described here, this is impossible if parameters have units.