I Sine laws of spherical singlets

  • I
  • Thread starter Thread starter difalcojr
  • Start date Start date
AI Thread Summary
Eugene Hecht's assertion that all aberrations cannot be made zero in systems with spherical surfaces is challenged, particularly for monocentric, biconvex lenses. The discussion centers on equations derived from sine laws that can theoretically yield zero spherical aberration (SA) under specific conditions, particularly at aplanatic points. However, participants agree that achieving a real image free of SA with a single spherical lens is impractical, as it effectively becomes a cemented doublet. The conversation also touches on the feasibility of using high-index materials to create such lenses, with costs and manufacturing challenges being significant considerations. Ultimately, the theoretical model presented could have implications for optical design, despite its practical limitations.
  • #51
difalcojr said:
Now I know why mathematicians are so special, and why I'm not a mathematician. Wow. Is that it what is takes to find exact position points for a parbolizing, second surface? And they can mass produce molds for those calculated shapes? Stupendous achievement of mathematics.
Looks like the spherical model could do some of the same things except for the real-real examples. Good problem for a cost analysis, maybe, for spheric/aspheric vs. spheric/spheric molds. For someone else.
I take that back, actually, after another look at the article. This model cannot duplicate the aspheric examples shown. They are all either real-real systems or have angle magnifications. This model has wave, but not angle, magnification, in real-virtual or virtual-real systems. Not sure what to make of that.
 
Science news on Phys.org
  • #52
hutchphd said:
You may find this treatment interesting (shaping aspheric second surface) .
https://royalsocietypublishing.org/doi/epdf/10.1098/rspa.2014.0608
In the royal society article, there is not a plane wave input, plane wave output example. Nor online that I can find. I don't think it is possible, even using all that math. So, the question arises: can an aspheric singlet do the same as this model's singlet (doublet)?
Image (121).jpg
Could it do a 16:1 exact reduction like this diagram above?
 
  • #53
No idea. This is a fair bit above my skill level in optics. Which is about two semesters of geometrical optics from five years ago.
 
  • #54
Yes, me too, no idea. Math in the spherics article was way over my head.
Still, if it was possible to reproduce a planar wave magnification without angle change, I think they would have included it in the article. All their models had angle changes.
My thinking is that, if the models shown in this post are unable to do any of the projections that the spheric-aspheric singlets in the royal society article can do, then probably the spheric-aspheric family models cannot do any of this model family's projections, either.
Good problem for all the mathematicians in your audience to solve.
 
  • #55
Hmmm. Does this not violate conservation of etendue?

Per wiki:
The etendue of a given bundle of light is conserved: etendue can be increased, but not decreased in any optical system. This means that any system that concentrates light from some source onto a smaller area must always increase the solid angle of incidence (that is, the area of the sky that the source subtends). For example, a magnifying glass can increase the intensity of sunlight onto a small spot, but does so because, viewed from the spot that the light is concentrated onto, the apparent size of the sun is increased proportional to the concentration.
 
  • #56
No idea. Don't know the term at all.
I wanted this post just to show a lens model family with constant magnifications and zero SA. A singlet in principle, a doublet in most all applications. To show the various wave and ray refractions, and to show the historical math connections to Huygens.
I think giving magnification values that I've calculated, and also showing optical path lengths, will further help to convince everyone that these theoretical lenses are really possible. I'll post that diagram and chart of values tomorrow. Thks for seeing this project along.
 
  • #57
Are the centers of curvature of each surface located at the same place? I was trying to do a paraxial raytrace, but I don't know how far apart the surfaces are.
 
  • #58
difalcojr said:
No idea. Don't know the term at all.
Probably worth looking at.
 
  • #59
Drakkith said:
Are the centers of curvature of each surface located at the same place? I was trying to do a paraxial raytrace, but I don't know how far apart the surfaces are.
For the single lens, yes. Monocentric.
Axial distance between surfaces is radius1 plus radius2.
For n0=1, and n2=square root of 2, and radius1=1, radius 2 is then (square root of 2)/2. So, adding radii., you get (1 + .7071...). The answer is 1.7071... for the thickness.
 
  • Like
Likes Drakkith
  • #60
Hmmm. Using this ynu ray trace calculator, I'm not getting what you're getting. My initially parallel ray diverges after exiting the 2nd surface. Not sure if a paraxial ray trace is inappropriate here, or if I'm doing something wrong.
 
  • #61
index2 should equal 2.
Try a marginal ray, maybe. y-axis height is 0.5 for the marginal ray. Everything else the same.
 
  • #62
Ah, I found the issue. The result was actually in engineering notation, but the number of digits after the decimal was so high the e-6 was hidden off to the right inside the text box. So it's very, very close to zero, close enough to agree with your analysis.
 
  • #63
Did it work for both the paraxial and the marginal ray? Ray angles set at zero degrees with the horizontal, and the exit angles the same?
 
  • #64
Just did a ray trace with an initial ray angle of 0.1 and ray height of 0 (hits the first surface in the center). The resulting exit angle was about 0.0999939. The difference from 0.1 is easily attributable to me only using 4 or 5 digits for the input values.

difalcojr said:
Did it work for both the paraxial and the marginal ray? Ray angles set at zero degrees with the horizontal, and the exit angles the same?
Yes, the first trace was a parallel ray with an angle of 0. The exiting angle was -1.4082972487512724E-06. Or very, very close to zero.
 
  • Like
Likes difalcojr
  • #65
Terrific! That's good news. You can get a number even closer to zero maybe if you keep lowering the initial ray angle too. So, you have ray trace solfware for thick lenses too. Excellent.
That's two points on the planar wave, exiting with the same angle, the wave reduced in half its size. All the other rays between horizontal axis and marginal ray will do the same if you check. Thanks.
 
  • #66
Image (113).jpg

Here's a last, spider-legged diagram and values. A good test for your software now.
Let me explain how to read this. It is to show ray/wave magnifications and optical path lengths. Under the minimum conditions. At the point when each wave starts to enter the lens, and at the point when each wave finally exits the lens. Same waves as from the other sequences shown. Marginal rays incoming, marginal rays outgoing. Here's some values you can check for the marginal and arbitrary rays. Minimum magnification ratios.

TYPE...............................ANGLE......................MAGNIFICATION.............................................................................................................
converging.......................30.................................10.24:1...........................................................................................................................
converging.......................15....................................3.97:1..........................................................................................................................
planar..................................0...........................................2:1...........................................................................................................................
diverging..........................15..............................................1:0.92.................................................................................................................
diverging..........................30..............................................1:1.71.................................................................................................................
diverging..........................45..............................................1:3.41.................................................................................................................

Hope this chart posts OK to read.
For optical path lengths only the lens itself needs be figured to get the OPL. This formula works at the minimum condition for the lens:

Minimum OPL=(radius1+radius2)(index1). Its value for the lens index of square root of 2 equals 2.4142...
This value is constant for every wave shown.
 
  • #67
Drakkith said:
Just did a ray trace with an initial ray angle of 0.1 and ray height of 0 (hits the first surface in the center). The resulting exit angle was about 0.0999939. The difference from 0.1 is easily attributable to me only using 4 or 5 digits for the input values.Yes, the first trace was a parallel ray with an angle of 0. The exiting angle was -1.4082972487512724E-06. Or very, very close to zero.
So, I know you are probably super busy with other projects, and I've overloaded everyone with this, and especially you and hutchphd. And it's the weekend now, and it's been 2 weeks with this problem.
When you get a chance, or anyone else please, even teachers needing a student project, maybe, set up this lens model with those easy equations given. Use the diagram for the plane model. Ray trace some different angled rays through the model lens. Convince yourselves that the incoming and outgoing angles are the same for this model's shape and choice of indexes.
I truly expect someone in this forum to be able to confirm or disprove my contention that this model has zero SA. It's only geometrical optics and very easy, right? Thks.
 
  • #68
I managed to slap together a quick design in Optical Ray Tracer (which I've never used before) but either limitations in the software or my own inexperience with it forced me to eyeball things. I was able to get the rays entering the second lens to look close to parallel. You can try it yourself, you just have to fiddle with the thickness of the first lens and the x-position of the 2nd lens to get the distances between the elements correct. Unfortunately it wasn't as easy as just inputting the values we had above, for reasons I don't know.

I basically set up two lenses, the first with asymmetric curvatures, setting the curvature of the 2nd surface (the 'left sphere' in the program) to 0.70710798, the curvature of the 1st surface of the second lens to -0.70710798, and making the 2nd lens very thick. The radius of each lens should be small, something like 0.5 or so. Smaller than the smallest radius of curvature. The second lens should have an x-position that puts it very, very close to the last surface of the first lens with only a tiny gap in between.
 
  • Like
Likes difalcojr
  • #69
Drakkith said:
I'd say do a full raytrace before trying to consider what the system would be good for. SA might be fully corrected for, but what about other aberrations?
There is free OSLO EDU raytracing software which is limited up to 10 surfaces for free educational version.
 
  • Like
Likes difalcojr and Drakkith
  • #70
Drakkith said:
I managed to slap together a quick design in Optical Ray Tracer (which I've never used before) but either limitations in the software or my own inexperience with it forced me to eyeball things. I was able to get the rays entering the second lens to look close to parallel. You can try it yourself, you just have to fiddle with the thickness of the first lens and the x-position of the 2nd lens to get the distances between the elements correct. Unfortunately it wasn't as easy as just inputting the values we had above, for reasons I don't know.

I basically set up two lenses, the first with asymmetric curvatures, setting the curvature of the 2nd surface (the 'left sphere' in the program) to 0.70710798, the curvature of the 1st surface of the second lens to -0.70710798, and making the 2nd lens very thick. The radius of each lens should be small, something like 0.5 or so. Smaller than the smallest radius of curvature. The second lens should have an x-position that puts it very, very close to the last surface of the first lens with only a tiny gap in between.
Valiant effort! Here's the model diagram again, explicitly, if you want to check that software against this one.
Image (122).jpg
 
  • #71
Gleb1964 said:
There is free OSLO EDU raytracing software which is limited up to 10 surfaces for free educational version.
Should only need two surfaces, thankfully.
Trace a planar ray through the planar model with the values shown in the last post's diagram, and trace any diverging or converging ray angle through it too. See if the input and output ray angles will be the same.
I have my doubts still on existing software being able to do thick lens, trigonometric tracing, as this requires.
 
  • #72
hutchphd said:
You may find this treatment interesting (shaping aspheric second surface) .
https://royalsocietypublishing.org/doi/epdf/10.1098/rspa.2014.0608
From what I can tell the family of aspheric model lenses free from all orders of spherical aberration as shown in the article, is akin to this family of spherical model lenses free from all orders of spherical aberration.
A modern, aspheric singlet and a classic, spherical singlet. Both with zero SA. They could be paired together, possibly.
The spherical model here is exceedingly simple, while the modern physics aspheric model is the opposite. From a geometrical optics point of view, the aspheric model seems similar to Descartes' ovals.
The family photo of spherical models is shown in that large array in a previous post here. If you turn those pictures sideways, you will see how highly groomed their "haircuts" are. An ordered universe.
 
  • #73
It turns out that the OP is trying to use PF for peer-review of his work, so this thread is closed per the PF rules.
 
Back
Top