PDA

View Full Version : Space Between Lenses



seawolf66
17-Apr-2012, 09:26
How does one compute the space between lenses is it hard or just a simple formula
See attached : Thanks a lot :

genotypewriter
17-Apr-2012, 10:11
I don't think there's only one exact number. But at least one will be better than the others.

E. von Hoegh
17-Apr-2012, 10:37
The spacing is part of the design computations. If remounting a lens in shutter, it is best to measure the spacing and make adapters if needed to preserve the spacing in the new shutter.

domaz
17-Apr-2012, 16:13
I think this falls into the "very hard" category. The space between the lens elements is an integral part of the lens design. Finding the number would kind of be like finding what the formula for a film developer after it's been mixed.

Leigh
17-Apr-2012, 18:26
I believe you're asking about the separation between the front and rear lens cells.

Calculating that is a complex task that requires many parameters that you don't have and can't get.

Measuring it requires expensive optical test equipment that you probably don't want to buy.

I have the test equipment, but it's not set up to do inter-cell spacing tests.

I suggest you send it to SK Grimes and ask them to mount it for you. They have the equipment to do it right.

- Leigh

www.skgrimes.com

Michael Clark
17-Apr-2012, 20:30
I measured the distance from the front of the lens to the back of the lens while it was still mounted on the old shutter it came with, then installed it in the new shutter and measured it was the same distance. Not that precise but seems ok to me, there is always SK Grimes.

Mike

Leigh
17-Apr-2012, 20:35
I measured the distance from the front of the lens to the back of the lens while it was still mounted on the old shutter it came with, then installed it in the new shutter and measured it was the same distance.
That will work.

The desired precision would be ±0.1mm (±0,1mm for our European friends).
Looser precision will work but might degrade performance.

- Leigh

Michael Clark
17-Apr-2012, 21:02
it was a 150 syminar in an old broke down number 0 sycro-compur shutter to a newer copal #0 press shutter. Looked up the specs for them and they were the same as far as the two tread distances, there were no shims in the old shutter. I think they were #0 shutters.

Mike

genotypewriter
18-Apr-2012, 02:29
Calculating that is a complex task that requires many parameters that you don't have and can't get.


Not saying it's easy but, if you're interested in finding the "correct" spacing for infinity, there are only two parameters: (1) spacing between the two cells and (2) the back focal length.

The range of values for these can be further narrowed down by using domain knowledge. For example, intuition says the spacing is most likely not going to be massive like 10 inches. So you can put it between 0-10" (or say 1-250mm). For the second parameter, BFL, again some educated guess about focal length can narrow the range down. Let's say 1000mm (which is most likely grossly over in reality) so that puts the range between 1-1000mm.

Assuming an initial precision of 1mm, we have 250*1000=250,000 combinations to try out. This is where a heuristic is going to help. If properly done, a Monte Carlo search should allow the OP to estimate the spacing (and BFL) within a hundred or so attempts. But I think it's going to be even quicker in practice since this is not a very stochastic problem and the parameter ranges are continuous.

After finding a rough (1mm precision) configuration, fine-tuning can be done by hand easily to whatever the possible precision is.

ic-racer
18-Apr-2012, 06:01
How does one compute the space between lenses is it hard or just a simple formula
See attached : Thanks a lot :

Measure the entire assembly = distance A
Measure the front cell from filter ring to the area that touches the shutter (not including the threads) = distance B
Measure the rear cell same as above = distance C

The space between the cells is then = A- (B + C)

E. von Hoegh
18-Apr-2012, 06:42
Not saying it's easy but, if you're interested in finding the "correct" spacing for infinity, there are only two parameters: (1) spacing between the two cells and (2) the back focal length.

The range of values for these can be further narrowed down by using domain knowledge. For example, intuition says the spacing is most likely not going to be massive like 10 inches. So you can put it between 0-10" (or say 1-250mm). For the second parameter, BFL, again some educated guess about focal length can narrow the range down. Let's say 1000mm (which is most likely grossly over in reality) so that puts the range between 1-1000mm.

Assuming an initial precision of 1mm, we have 250*1000=250,000 combinations to try out. This is where a heuristic is going to help. If properly done, a Monte Carlo search should allow the OP to estimate the spacing (and BFL) within a hundred or so attempts. But I think it's going to be even quicker in practice since this is not a very stochastic problem and the parameter ranges are continuous.

After finding a rough (1mm precision) configuration, fine-tuning can be done by hand easily to whatever the possible precision is. And how does one know when the spacing is correct? Your exercise is based on two guessed parameters. Unless you have one known, say the original measured EFL, you're throwing darts in a darkened room, and you don't know where the target is.

E. von Hoegh
18-Apr-2012, 06:45
Measure the entire assembly = distance A
Measure the front cell from filter ring to the area that touches the shutter (not including the threads) = distance B
Measure the rear cell same as above = distance C

The space between the cells is then = A- (B + C)

The OP is asking how to determine the spacing when he doesn't know what the shutter dimensions were. If you have the lens in shutter, all you need to do is unscrew the cells and measure the distance between seating surfaces on the shutter, one measurement, not three.

genotypewriter
18-Apr-2012, 09:01
And how does one know when the spacing is correct?

Well it's a photographic lens... the objective is to get something in focus, right?



Your exercise is based on two guessed parameters. Unless you have one known, say the original measured EFL, you're throwing darts in a darkened room, and you don't know where the target is.

Not guessed... empirically calculated. Also, there is no dart throwing of any sort. Monte Carlo methods are very successful in just this sort of thing.

There's going to be at least one configuration in which sharp focus will be obtained. If there's more, one can pick which ever suits them the most. Can't guarantee that the methods brings the lens to the original design/configuration. Can't guarantee that it's going to be worse either.

E. von Hoegh
18-Apr-2012, 09:16
Well it's a photographic lens... the objective is to get something in focus, right?




Not guessed... empirically calculated. Also, there is no dart throwing of any sort. Monte Carlo methods are very successful in just this sort of thing.

There's going to be at least one configuration in which sharp focus will be obtained. If there's more, one can pick which ever suits them the most. Can't guarantee that the methods brings the lens to the original design/configuration. Can't guarantee that it's going to be worse either.

Please explain how one "empirically calculates" these unknowns....when a lens can easily be + or- 5% from the marked focal length. Your best bet would be to set the cells up on a simple optical bench and set the spacing at the distance that gives you the flattest field. Monte Carlo methods....... try the experimental method, it's even more succesful in this sort of thing.

ic-racer
18-Apr-2012, 11:37
The OP is asking how to determine the spacing when he doesn't know what the shutter dimensions were. If you have the lens in shutter, all you need to do is unscrew the cells and measure the distance between seating surfaces on the shutter, one measurement, not three.

That is not a parameter that is "computed" it is established by trial and error.

seawolf66
18-Apr-2012, 11:44
Thank you all for your time and effort here to answer my crazy question<
One being it came as just a pair lenses marked 175mm at F4.7 so its going to be send it to
Grimes or find a shutter they will and play with till the image is sharp: Thank you All

E. von Hoegh
18-Apr-2012, 11:47
That is not a parameter that is "computed" it is established by trial and error.

Sorry, but it's part of the design process of the lens, no trial or error about it. One resorts to trial and error when one has lost the original number.

genotypewriter
18-Apr-2012, 18:26
Please explain how one "empirically calculates" these unknowns....when a lens can easily be + or- 5% from the marked focal length.

I have already explained most of it in my previous post. For the rest, please refer to:
http://en.wikipedia.org/wiki/Search_algorithm
http://en.wikipedia.org/wiki/Monte_Carlo_method
http://en.wikipedia.org/wiki/Evolutionary_algorithm



Sorry, but it's part of the design process of the lens, no trial or error about it. One resorts to trial and error when one has lost the original number.

Lens design also uses these exact techniques:

http://spiedigitallibrary.org/oe/resource/1/opegar/v43/i2/p432_s1?isAuthorized=no
http://www.opticsinfobase.org/ao/abstract.cfm?id=131885
http://www.sciencedirect.com/science/article/pii/S0045782599003989
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=823279&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D823279
http://onlinelibrary.wiley.com/doi/10.1002/mop.20577/abstract
http://spiedigitallibrary.org/proceedings/resource/2/psisdg/3482/1/110_1?isAuthorized=no
http://www.springerlink.com/content/rvcx1uaewuqnlqb7/


In addition to having the possibility of finding the exact original spacing when using these techniques, there might even be chance of finding something even better than what the manufacturer did :)

Leigh
18-Apr-2012, 18:50
...empirically calculated. ... Monte Carlo methods are very successful in just this sort of thing.
Empirically calculated from what? What parameters do you propose to use?

As to MC analysis, how would you set the goal and evaluate the results? What dataset would you analyze?

Buy a copy of Zemax, plug your numbers in, and let us know what you find.

starting point...
What values will you use for the optical characteristics of the glass, for the radii of the lens surfaces, and lens thickness?

- Leigh

genotypewriter
18-Apr-2012, 19:11
Empirically calculated from what? What numbers do you propose to use?

As to MC analysis, how would you set the goal and evaluate the results? What dataset would you analyze?


Again with the give an inch, obliged to complete it with a mile thing. I suggest you find out how focus can be evaluated. It's not that hard to come up with.

Also there is no "data set" here in a finite sense. We're optimising focus as a function of two variables:
http://en.wikipedia.org/wiki/Mathematical_optimization#Minimum_and_maximum_value_of_a_function




Buy a copy of Zemax, plug your numbers in, and let us know what you find.

- Leigh


The method I suggested doesn't need Zemax :D Remember how I said one only needs to know only two parameters? If you were to use Zemax you'd need more information on the optical element properties of the lens cells too. Good luck measuring those.

Leigh
18-Apr-2012, 19:22
If you were to use Zemax you'd need more information on the optical element properties of the lens cells too. Good luck measuring those.
Exactly, as I said in post #5 above and again in post #19.

- Leigh

Mark J
19-Apr-2012, 13:08
I come onto this forum with trepidation, knowing that I might be making a rod for my own back here .

Anyway . The difficulties here are not about technical writing .
The issue is that genotypewriter is assuming thin-lens theory to explain the situation, whereas we are dealing with a real lens design, where the elements are thick .
As Leigh says, we need to know more info about the real ( thick-lens ) design if we do not have the existing mechanical data regarding the spacing between the reference faces of the two cells . It seems here we have that data, so the cells can ultimately be placed in the correct relationship .

If we do not know this mechanical data , then we can make a stab at the spacing by testing the lens focal length while trying different shims to space the components apart by different amounts . This is only approximate, as the actual focal length of the lens example may vary from stated : modern lenses by maybe +/-1% , older lenses by 2% or more from the stated focal length . However, we should get 'OK' performance at least , with a fair bit of effort .
Hope this helps

Mark J

Mark Jeffs ( Senior Lens designer, Qioptiq UK Ltd, Pilkington PE Ltd , Zeiss Oberkochen 1997-1998 )

E. von Hoegh
20-Apr-2012, 07:14
I come onto this forum with trepidation, knowing that I might be making a rod for my own back here .

Anyway . The difficulties here are not about technical writing .
The issue is that genotypewriter is assuming thin-lens theory to explain the situation, whereas we are dealing with a real lens design, where the elements are thick .
As Leigh says, we need to know more info about the real ( thick-lens ) design if we do not have the existing mechanical data regarding the spacing between the reference faces of the two cells . It seems here we have that data, so the cells can ultimately be placed in the correct relationship .

If we do not know this mechanical data , then we can make a stab at the spacing by testing the lens focal length while trying different shims to space the components apart by different amounts . This is only approximate, as the actual focal length of the lens example may vary from stated : modern lenses by maybe +/-1% , older lenses by 2% or more from the stated focal length . However, we should get 'OK' performance at least , with a fair bit of effort .
Hope this helps

Mark J

Mark Jeffs ( Senior Lens designer, Qioptiq UK Ltd, Pilkington PE Ltd , Zeiss Oberkochen 1997-1998 )


Thank you.

Mark J
20-Apr-2012, 08:08
Yes, I think the trepidation was well-founded !


Speaking about assumptions, Mark, could you point out where I have asked anyone to use any thin lens theories?

So you clearly didn't read my first post:
http://www.largeformatphotography.info/forum/showthread.php?89770-Space-Between-Lenses&p=876823&viewfull=1#post876823

... in which I say "after finding a rough (1mm precision) configuration, fine-tuning can be done by hand easily to whatever the possible precision is."

The function here is focus. It has two parameters. There exists a configuration or a set of values for these two parameters that will bring the function to a maximum. I.e sharp focus. Simple as that.

I don't know if it's a generational gap thing or something to do with engineers but young and learned people don't take part in pissing contests any more.

-GTW

Ok, from your description of the process of adjusting this one air gap and the using some target for back-focus, it seemed like you were simplifying the problem too much, I may have misinterpreted the ideas you were using, thinking you were applying the lens equation to the focal lengths of the two parts or some such calc . Apologies for that .
Certainly you're right that if you know the design back-focus for this lens ( not the focal length, but the distance from the rear vertex to the image ) , you can get a good estimate of the correct spacing , by adjusting to get axial back-focus about right . Unfortunately this doesn't take account of the likelyhood that this lens example will have some error from design that might be 2 or 3 % or more if it's an older lens ... so it may not have optimum (field) performance when you set what you think is the 'correct' internal spacing . Best axial image focus spot may also occur at a very different spacing from best overall field perforance, as many lenses have residual spherical aberration on-axis which is balanced up against field aberrations to get best overall field coverage and also best stopped-down performance .
Now, if you set the internal gap roughly as above, then the best thing to do is use a set of different shims to alter the gap one way or the other, while checking the performance over the field , either with film 'shots' or some sort of careful assessment of the image on the ground glass , be sure to check 70% field height and full corner field .
This because the field astigmatism ( mainly the T-field ) will move quite sensitively as a function of the internal air gap . I would expect the finesse required to be about 0.1 to 0.2mm .

On the final point, I did deliberate about whether to put this information about my background in, as some people take this the wrong way, but I thought it would be helpful if people knew I had some pedigree in this area .

Mark

Frank Petronio
20-Apr-2012, 08:43
Wanna bet the Zeiss lens designer is sufficiently scared of participating in this forum now?

Hermes07
20-Apr-2012, 09:30
My input as a humble photographer who has had cause to try and re-assemble a couple of lenses in the past is that it is not a simple matter of rough adjusting and then fine-tuning.

As has been pointed out, with nothing to go on other than the declared focal length of the lens, you are left with plenty of margin for error. Fine-tuning by eye (or even by measurement) once you have reached the assumed focal length is full of pitfalls. Do you adjust for the on-axis image or the periphery? - my understanding is that at the design stage, the on-axis image is generally sacrificed to an extent to improve the correction of the overall field so you would have to strike a balance between the two, without knowing exactly what field the lens is intended to cover.

Going through this balancing act, measuring and adjusting for every aberration and finally settling on a compromise that looks good or makes sense to you will probably get you near the original spacing but I don't see how you could reasonably hope to get it exact without something finite to aim for. You are essentially creating your own lens and while it may or may not be good enough for your purposes, you couldn't in any meaningful way judge it as a representative specimen of the original design.

If you have the elements anyway then by all means, spend a bit of time spacing them as best you can and try the resulting lens to see if it works for you, but I'd abandon thoughts of being able to exactly resurrect the original through trial and error without a lot of background knowledge, experience and precision equipment.

Leigh
20-Apr-2012, 21:14
I just wanted to add some stimulation to this thread based on Leigh's comment in which he said "calculating that is a complex task that requires many parameters that you don't have and can't get"
That's right. I said "calculating", not "guesstimating", which is what you've suggested.

My original statement is correct.

You can play with a set of cells and come up with a result that meets your requirements,
but it may not meet the factory specs.

- Leigh

genotypewriter
21-Apr-2012, 18:13
That's right. I said "calculating", not "guesstimating", which is what you've suggested.

Yet you conveniently choose to ignore my clarification on what I meant by that:
http://www.largeformatphotography.info/forum/showthread.php?89770-Space-Between-Lenses&p=876949&viewfull=1#post876949





You can play with a set of cells and come up with a result that meets your requirements,
but it may not meet the factory specs.

Who's talking about factory specs? Did the OP ask for factory specs? The method I suggested also doesn't mean worse focus than the factor spec. Can you prove that it'll be worse than the factory spec? No.

Dan Fromm
21-Apr-2012, 18:58
Leigh, do you remember your Uncle Remus? I see an effigy of an infant in front of us, made of a dense, sticky petroleum product.

Mark Sawyer
22-Apr-2012, 00:36
I've been staying out of this for the reason noted by Dan. But I'll offer a few observations...

The calculations aren't that simple. Back in the day, so to speak, when Petzval was calculating the Petzval and Dialyte formulas, he had ten artillery men (the military math experts of the day) working for months on the calculations. And he only had crown and flint glasses to worry about...

Today we have so many different glasses with so may different refractive indices...

Correct spacing of elements and the aperture are each "freedom of design" areas, and each spacing affects sperical and chromatic abberation, field curvature, coma, pincushion and barrel distortion, on- and off-axis resolution, oblique astigmatism...

There isn't a perfect position that corrects for all. The designer trades one against another, swapping curvature, spacing, thicknesses, refractive indices of all kinds of glass...

And of course things change at different apertures and different focusing differences...

Refer to the original manufacturer/designer specs. It's not a simple calculation.

Ken Lee
22-Apr-2012, 07:58
Warning: those who can't participate in a discussion of technical matters without name-calling and rudeness, are going to get banned for a while.

Moderators have better things to do than pore over threads like this, trying to determine which posts are outright rude, which use insinuation, and which are merely replies.

Keep it civil, or... Adiós amigo!

On the other hand, friendly and helpful discourse is always welcome, especially among our international collection of engineers and experts :cool:

http://www.kenleegallery.com/images/forum/scientists.jpg

Mark J
23-Apr-2012, 10:08
Back in the day, so to speak, when Petzval was calculating the Petzval and Dialyte formulas, he had ten artillery men (the military math experts of the day) working for months on the calculations. Today we have so many different glasses with so may different refractive indices...

Correct spacing of elements and the aperture are each "freedom of design" areas, and each spacing affects sperical and chromatic abberation, field curvature, coma, pincushion and barrel distortion, on- and off-axis resolution, oblique astigmatism...

There isn't a perfect position that corrects for all. The designer trades one against another, swapping curvature, spacing, thicknesses, refractive indices of all kinds of glass...



Kingslake tells us (in his excellent paper on the History of photographic lenses ) that it was 'a detachment of eight Austrian bombadiers' ;o)

You're right, Mark, that there are a lot of design parameters available, even in something like a Petzval or a Dagor ; however, in the original subject of the thread, we are lucky in that the manufacturer has helpfully 'frozen' almost all of the available degrees of freedom of the lens, by making it, hopefully close to design spec . What we have left is only one parameter - the spacing of these two groups .
We just have to be careful to realise that we need to evaluate the image quality over the extent of the design image circle , not just a check on focal length or the back-focus .

Now, looking back at the original post, I see that it may be even easier, given that the lens type and possible the shutter are known . There ought to be drawings or information out there ( someone can measure ) that would get the spacing pretty close. Maybe the older manufacturers would also use a variable shim in there , I'm not sure .

These days ( since 60's I guess ) everyone knows what the spacing of the reference faces of a Copal 0, Copal 1, Compur 1 etc. should be, and the manufacturers expect to screw in the lens cells without shimming . The tolerance is probably held to something pretty tight, like 0.025mm or so .

Mark Sawyer
23-Apr-2012, 11:06
Well, there are two different takes on the question from the OP. My response was to the first:


How does one compute the space between lenses is it hard or just a simple formula
See attached : Thanks a lot :

...and I meant to point out that there isn't a simple formula leading to a defined perfect answer.

In post 16, the OP refers to a set of specific cells. Here your suggestion, "There ought to be drawings or information out there ( someone can measure ) that would get the spacing pretty close", would work. Even more likely, post the information on the lens and ask if anyone has a similar lens and what shutter is it in or what is the overall length, front to back.

On Petzval's mathematics resources, Kingslake wrote in A History of the Photographic Lens, "For help in his calculations he (Petzval) approached the Archduke Ludwig, Director General of Artillery in the Austrian Army, who ordered that Corporals Loschner and Haim and eight gunners skilled in computing be placed at his disposal." That's ten. :) (And this to help out someone who had already been appointed to professorships of higher mathematics at the Universitiy of Budapest and the University of Vienna!)

E. von Hoegh
23-Apr-2012, 11:34
And Herr Schiempflug was.... an artillery officer. From Austria.

Lynn Jones
23-Apr-2012, 12:40
Lens designers consider the air space between lens elements and groups as just another lens element.

Lynn

E. von Hoegh
23-Apr-2012, 12:49
Lens designers consider the air space between lens elements and groups as just another lens element.

Lynn

Correct. Therefore the airspaces, all of them, are an important part of the overall formula. The Plasmat was an improvement on the Dagor for this reason, it gave the designer an extra two "elements", and also made production a bit less costly.

Mark J
23-Apr-2012, 13:09
On Petzval's mathematics resources, Kingslake wrote in A History of the Photographic Lens, "For help in his calculations he (Petzval) approached the Archduke Ludwig, Director General of Artillery in the Austrian Army, who ordered that Corporals Loschner and Haim and eight gunners skilled in computing be placed at his disposal." That's ten. :)

You've got me there, Mark. I hadn't checked who else was in the room !

There are various ways of looking at the air-gap / lens splitting option . One way to look at the various degrees of design freedom, is that each lens element has 'power' and 'bending' available ... unless it is cemented to something else. Releasing the inner positive elements from the others in a Dagor gives you the freedom to bend them independantly .
In the case the Dagor-to-Orthometer ( Plasmat ) development , it's actually more about higher-order aberrations - the ray angles in the small air gaps allow you to improve the correction of 5th-order spherical, which allows you to increase the speed retaining the same level of field correction .

The plasmat was a little ahead of its time, though, because it needed coatings to be developed before we could take advantage of the extra performance .

E. von Hoegh
23-Apr-2012, 13:18
You've got me there, Mark. I hadn't checked who else was in the room !

There are various ways of looking at the air-gap / lens splitting option . One way to look at the various degrees of design freedom, is that each lens element has 'power' and 'bending' available ... unless it is cemented to something else. Releasing the inner positive elements from the others in a Dagor gives you the freedom to bend them independantly .
In the case the Dagor-to-Orthometer ( Plasmat ) development , it's actually more about higher-order aberrations - the ray angles in the small air gaps allow you to improve the correction of 5th-order spherical, which allows you to increase the speed retaining the same level of field correction .

The plasmat was a little ahead of its time, though, because it needed coatings to be developed before we could take advantage of the extra performance .

Actually the Plasmat is a big improvement over the Dagor in that the Dagor has quite a bit of spherical abberation wide-open, also shifts focus as you stop it down. The Plasmat commits neither of these sins, and is a bit faster. The Dagor scores in half the number of air to glass surfaces, an uncoated Dagor comparing very favorably to a single coated convertible Symmar.