Use the surface tilt in Y to model wedge (you only need one surface per singlet) and then use TETZ to rotate the whole element randomly around 360 degrees. This won’t do anything for the sensitivity analysis as 0 = 360 degree rotation, but during Monte Carlo you will get every rotation angle represented.

• Mark

Here’s an example:

Surface 2 is wedged with respect to surface 3 by line 10 in the tolerance editor. Line 9 TETZ rolls the component formed by 2 and 3 to be anywhere between 0 and 360 degrees.

• Mark

That’s a good point Jeff, thanks.

Another thing just occurs to me. When we design a doublet, we can use a short cut that sequential ray tracing provides and model the doublet using only three surfaces. But when tolerancing, in other to get the kind of offset Jeff shows in his reply, it’s essential that you model the doublet as two separate singlets: in other words you need to use four surfaces to define the doublet, not three. Each singlet uses two independent surfaces.

Now during optimization, we can use only three surfaces because the rear of element 1 must have the same radius of curvature (etc) as the front surface of the second element. But for tolerancing the doublet, you must allow them to be different, so they can have slightly different curvatures as well as positions.

OpticStudio does this automatically for you when you export to CAD or convert to NSC, as these are true 3D solid models. But you have to do it yourself for tolerancing. There’s a long-standing feature request to add this capability to the Design Lockdown tool, but for now you have to do it yourself.   

Thank you, Jeff and Mark! They are very helpful. I figured it out based on your suggestions.

Best,

Xiaolei

Mark.Nicholson wrote:

Here’s an example:

Surface 2 is wedged with respect to surface 3 by line 10 in the tolerance editor. Line 9 TETZ rolls the component formed by 2 and 3 to be anywhere between 0 and 360 degrees.

• Mark

TIRX 2 is the surface wedge

However, TETZ 2 3 refers to the whole element clocking

Does it matter if you use TETZ 2 2 ?  

When Zemax implements this, is the surface or element tilted before the rotation? 

Do these two commands need to be on adjacent lines, or in any particular order?

I think most people don’t know this clocking trick, and just use both TIRX and TIRY together, which gives inaccurate results for rotationally symmetric systems. 

Also, what if I also have element tilt?  I assume that I need a second TIRX (or TEDX if it’s an aspheric surface) but do I also need two separate TETZ commands?

I think a cleaner way to do this would benefit a lot of people…  :)  I’ve used Zemax for a long time and I still find this confusing...

It looks like the nesting rules answer some of my questions.