420 replies to this topic

[freestar8n]

Well in a lot of these threads there are missing contexts that make the topic seem unimportant. On the topic of aperture measurement, there is I think a somewhat serious issue that affects consumers - and that is potential overstatement of aperture in products. Delivering a product with lower aperture that advertised means the light collection is less than expected, and it relaxes the tolerances for aberration correction. If the issue is due to a stop "near the front" that is easy to measure, then it is easy to tell if it is a problem - but if it is inside, it becomes harder.

For inexpensive items a few millimeters might not matter much - but for small and expensive binoculars, even 0.5mm might matter - so the accuracy can be important. Having things be too big is probably ok - but I think the tendency is the other way.

The other thing that comes up is the loss of aperture when things are added to the back of an sct. This is also where the pupil shadow is thrown farther and it is harder to measure. People giving advice on the results of their measurements should make sure they are measuring the right thing.

If in fact a complex optical system is advertised as having a given aperture and f/ratio, and inside there is a non-obvious restriction that is intentionally clipping that aperture - then it is not a good thing. It makes it a lot easier to control spherical aberration - but it isn't as fast as expected. It doesn't take many millimeters in reduction to improve the aberration - so a reliable measurement is needed. And when the stop is deep inside - it becomes hard to do.

Finally - I think at least part of Glenn's motivation is to convey some of the concepts of pupils in the first place - but for that I strongly recommend sticking to the definitions, and using the concepts of image and object space that are in almost any text.

The location of the pupils has importance in many areas, and a popular one these days is panoramic photography. Amateur photographers are well aware of the role of the entrance pupil and the nodal point as the desired pivot point for panorama photography - since it is the pupil location in object space. Here is an entrance pupil database where you can find the entrance pupil location for many dslr lenses. Here is a write up that then talks about subtleties beyond simple concepts of entrance pupil in some detail. Some people find these details inherently interesting and important, and others don't.

Frank

[Asbytec]

Mike, thank you for the reply and diagram.

Nils, "If you find such an aperture stop, you would identify and remedy it, and go on observing." I think that's exactly right, if one is interested. Who knows how many are.

Recently, I used your pin hole technique to good affect, in conjunction with discussion with Glenn. It was telling.

One way you can determine vignetting, as I understand it, is to defocus a star and slew it around the field looking for it to flatten on one side. This can give a rough idea of where the fully illuminated FOV is. Of course, if you're already at reduced aperture, it will only find the next vignetting obstruction.

I think Frank makes a good point, though, on the fussiness of the test especially with a flashlight. It has to be very close to the eyepiece to project an image bright enough to work with. The laser proved much more effective, IME, and less fuzzy. But, Glenn discusses the distance behind the eyepiece required to give the flashlight a roughly parallel beam and how to compensate for the fussiness when measuring.

So, fussiness seems to be a valid point in terms of precision but possibly not so much in simply identifying a reduction in working aperture. However, if you can interpret the fuzz correctly - maybe the amount of fuzz can give some indication of where the offending aperture stop might be. If it's sharp, no worries...the entrance pupil and aperture stop are on the objective (Right?) If it's fuzzy, check closer to the visual back (in a CAT.)

If I may, cuz I'm no expert, but one of the tings that clued me in was how the moon - dead center FOV - illuminated the OUTSIDE of my secondary baffle. Slewing the moon, one can easily see when the primary baffle became a source of vignetting. But, on axis, that light not making it to the secondary was a clue. It was confirmed by running a laser along the marginal ray, as well.

Hope this is still on topic, but the point being, all these tests (including the shadow intrusion test) were all consistent with the laser version of the flashlight test. IIRC, there might have been even a little fuzziness (and fussiness.)

[freestar8n]

As soon as I said the words "Gaussian beam" the concept of a "collimated" beam go out the window.

But let's step back a bit. Apparently the 10 fl thing is gone, and I guess we are talking about a collimated laser system. The goal is to make a beam expander - and make sure the output is "perfectly collimated". Then you feed that into an sct, and adjust the sct focus of the output until it is "perfectly collimated."

Here is an introduction to laser beam expanders and it briefly describes the propagation of a Gaussian beam through a compound optical system. The input beam cannot be "collimated" - all you can do is specify the waist location. You then feed that into the eyepiece and it gets clipped, and is turned into a semi-Gaussian beam. If it were a normal Gaussian beam, it would not propagate by linear optics, but instead would keep having its input waist turned into a new output waist. Finally it comes out the end - and you adjust focus until it is "collimated" - which I guess means the waist is near the output - not sure.

I don't consider that easy to model at all - especially when you don't know what is inside the 'scope or where the obstruction is. What is the beam doing when it hits the obstruction? How should the beam waist even be prepared before entering the eyepiece - since it can't be "collimated?"

You keep characterizing the error expected based on very idealized interpretations of optics that just don't apply to reality. Very small errors in assumptions result in big errors in the shadow behavior - when cast a long way from a distant pupil. Light beams are never perfectly collimated and parallel - especially when they come out of lasers.

Frank

[GlennLeDrew]

Frank,
A laser's brightness profile is Gaussian, but the beam divergence is very small, usually a few arcminutes. For our purposes, that's as close to perfect collimation as one could ask for. And again, the visuals provide the confidence; the sheer sharpness of the shadows, with diffraction effects becoming visible. THAT is assurance of far more than adequate tininess of the image at focus; any other concerns about the light's characteristics are moot.

And again about the errors. They warrant no concern when emerging beam *parallelism* and *shadow sharpness* are obtained, for the errors have perforce been minimized. Nothing else matters; the result is good.

[freestar8n]

If the beam is Gaussian after it is focused to a spot on output from the eyepiece, then it is extremely divergent as it hits the first rear element of the optical system - and from there on it does not obey linear optics as it is affected by each element. Assuming that a tightly focused and expanding Gaussian beam will obey linear optics is a big no-no. By intentionally adjusting focus so it comes out "collimated" - you have de-focused the system away from infinity to something that makes the output seem collimated - but has no bearing on capturing the shadow of the obstruction accurately. It is a complicated mess and I don't know how it would even be modeled - but it is not as simple as you imagine, and it is not a small effect.

A lot of this depends on just how Gaussian the beam is, but in general it is dangerous to assume a prepared laser beam will obey the linear optics for which an optical system was designed. Gaussian optics is totally different.

As for "a few arc minutes" - let's say 1/10 degree, or 1/573 radian. Times 1 meter is a spread of 1000/573 or almost 2mm error due to taper if the pupil is only one meter away. In fact the pupil could be many, many meters away - and the taper of only a few minutes would generate a very large error in size - and you would have no idea the error was there because - as far as you measured, the beam is "collimated." And that is only one aspect of the overall systematic error.

So even without the complication of Gaussian beam propagation, the fact that you regard only a few minutes as acceptable means I think you still don't realize how challenging 1mm error is in this method, and how important the role of pupil distance plays in the final error.

Frank

[GlennLeDrew]

Frank,
The beam has a Gaussian profile, but all the photons are travelling in essentially the same direction. Different parts of the beam's cross section are refracted (or reflected) just like any other beam of collimated light of any imaginable energy distribution in cross section. In this respect there is nothing at all out of the ordinary with laser light. A varying intensity profile by itself does not impart some 'weird' optical behaviour.

But again, why even invent the unnecessary worry in the first place? All that matters is that a well focused spot of light be brought to a focus. It is, as amply evidenced by the supreme sharpness of shadows.

You keep introducing artificial and unnecessary worries. It's really much less fraught with the hidden dangers you're prone to imagining.

Collimated light out? Check!

Sharply defined edge? Check!

These two aspects tell us *all* we need to know regarding the variables of import; that the light source is tiny and that it's at the focus. What else matters? We now know the light path through the full system is the same as that from a star on axis. This fact does not change, no matter how hard we wring our hands or beetle our brows fussing and obsessing over imagined sources of error. We can *see* that the setup is sufficiently accurately configured.

[MKV]

For inexpensive items a few millimeters might not matter much - but for small and expensive binoculars, even 0.5mm might matter - so the accuracy can be important

Frank, could you explain this a little more? I mean, if the OG is 20 mm f/10, a ±0.5 mm on the aperture would make it f/10±0.53, i.e. f/9.5 to f/10.5? Is that really significant and why? Thanks.

Mladen

[Pinbout]

well according to his numbers as a worse case scenarios there's a +/- 1mm per 1M tolerance in measuring effective aperature.

so with those tolerances I'm still good at measuring my 8inf6 with binos to see if I'm really only getting a 7in aperature cause of my too small secondary.

According for Newt for Web if I change my secondary to a next size up I'll have a nice 100% imagae circle. So its nice to have another method to test the optical system to see if its designed well.

[MKV]

Frank,
A laser's brightness profile is Gaussian, but the beam divergence is very small, usually a few arcminutes. For our purposes, that's as close to perfect collimation as one could ask for. And again, the visuals provide the confidence; the sheer sharpness of the shadows, with diffraction effects becoming visible. THAT is assurance of far more than adequate tininess of the image at focus; any other concerns about the light's characteristics are moot.

And again about the errors. They warrant no concern when emerging beam *parallelism* and *shadow sharpness* are obtained, for the errors have perforce been minimized. Nothing else matters; the result is good.

Most inexpensive 1 to 5 mW laser diodes available on eBay have a divergence of 10 mm over 15 meters (15,000 mm). That's a tangent of 1/1500. My green laser diode has no visible divergence over 8 feet, as shown below (the white circle is a 3-ring binder page saver and the inner circle is 6 mm across).



and here is the same beam only a couple of inches from the Bath interferometer (which explains why there are two beams).



Clearly, the beam remains fairly parallel over a distance needed for most ATM telescopes to measure their aperture via Glenn's method.

For the more exacting experimenter, a focusable laser diode is always an option, so its beam can be tweaked to the required degree of parallelism.

The amount of tolerable divergence can be established very precisely by ray tracing.

Regards,
Mladen

[freestar8n]

Frank, could you explain this a little more? I mean, if the OG is 20 mm f/10, a ±0.5 mm on the aperture would make it f/10±0.53, i.e. f/9.5 to f/10.5? Is that really significant and why? Thanks.

Well - in the binocular community of CN it is considered important enough to try to measure and document. You can get very expensive 20mm binocs - and if they are actually 19.5, and a competitor has 20mm, then that is a non-trivial advantage.

This touches on a broader issue of even Newtonian mirror sizes and if they include the chamfer. Basically I think it is reasonable for the product to have the true aperture clearly stated - and it should be on the *low* side of what the consumer receives. Is that asking too much?

I am more concerned about the possible undersizing of Maksutov's, which are hard to make "fast" with spheres, but much easier if the aperture is reduced and the secondary obstruction is enlarged. In that case high order spherical is a concern, and anything high order is helped by small reductions of aperture. Also for maks, it's hard to measure the aperture accurately.

Frank

[freestar8n]

The beam has a Gaussian profile, but all the photons are travelling in essentially the same direction. Different parts of the beam's cross section are refracted (or reflected) just like any other beam of collimated light of any imaginable energy distribution in cross section. In this respect there is nothing at all out of the ordinary with laser light. A varying intensity profile by itself does not impart some 'weird' optical behaviour.

Yikes... Read up on Gaussian beam optics... It is different... They do not follow the rules of linear optics... Lasers are useful - but they aren't magic collimated beams of light...

If you only considered light as little particles flying through space, then you wouldn't be able to explain diffraction - and diffraction plays a key role in the behavior of Gaussian beams.

Frank

[GlennLeDrew]

Frank,
But again, why worry about the Gaussian 'behaviour'? The shadow sharpness *directly observed by one'e own eye* tells us the focused spot of light is very tiny. The very tightness of that light source makes it a perfectly reliable tracer of the path of light through the system. It serves as our surrogate image of a star, which happens to shine back out the way it entered, constrained to the same degree by the same limiting aperture.

We are functionally 'ray tracing' with light. For the determination of the real working aperture, It gets no simpler than this. Including the "difficult to measure" Mak.

It takes but a few minutes of trial, including practice on focusing until near and far measurements indicate parallelism of light output, to realize how potent the beam expanded laser test is. Just seeing how well defined the aperture is replicated will quell all doubts.

[GlennLeDrew]

It just occurred to me. If the Gaussian beam profile poses such a problem for "linear" optics, how can a simple 2-lens beam expander work? Such devices are typically reverse Galilean designs, using ordinary spherical lenses. They output a still-collimated beam, which proves that a Gaussian beam profile is not different than a non-Gaussian beam profile for any other collimated source.

While on the subject of finderscope beam expanders...
Because the output beam is Gaussian in profile, it has no visually discernible edge due to the tapering off in intensity. If you wish to verify that your unit is collimating the beam, place a mask over the objective just smaller than the beam diameter for a readily visible edge to be seen. (A variable polarizer or red filter [for green laser] can be used to cut down the intensity.) Measure the circle diameter at as large a distance as is convenient, and adjust focus until it's equal to the mask diameter.

[MKV]

A variable polarizer or red filter [for green laser

Potentiometers give you an infinite control over how bright your laser will be.

regards,
Mladen

[Asbytec]

I am more concerned about the possible undersizing of Maksutov's, which are hard to make "fast" with spheres, but much easier if the aperture is reduced and the secondary obstruction is enlarged. In that case high order spherical is a concern, and anything high order is helped by small reductions of aperture. Also for maks, it's hard to measure the aperture accurately.

Frank

Frank, for real world application that's one aspect that is very interesting. It was brought to light in a couple of threads, including stumbling onto it in ED Holland's monster thread a few years ago. The flashlight test was debated pretty well and effective aperture became an issue. Trick was, how to find, measure, and correct it and go on observing as Nils said. (And does correcting it reduce performance?)

I used several methods, each of which had some residual error in being able to measure accurately: shadow intrusion on a de-focused star, a collimated laser trace of the marginal ray, and the laser flashlight test variant. And later observation using a pin hole. All produced consistent results, but the pin hole method offered little means to actually measure anything. It was just observational confirmation. With a laser squared to the OG and passed along the edge (and moved inward until it cleared the secondary baffle) also showed the beam to strike the meniscus consistent with the other tests. (Admittedly, the latter test was hard to keep perfectly square, but with care and persistence could show consistent results and the potential source of the restriction.)

The laser flashlight method also produced a fairly dim image, but IIRC, it was pretty sharp projected on the wall and even the illuminated section of the meniscus. If it was fuzzy along the edge, the brightness appeared to fall off very rapidly. The projected cylinder was dimmed slightly (but not nearly as much as an LED flashlight method.) It was both bright enough and sharp enough to lay a ruler across it or sketch an outline on paper.

But, again IME, the test was consistent with all other tests that provided measurements, even if it /may/ have differed slightly or been slightly fuzzy. It was good enough test to tell something was obstructing the light path and give an estimate of effective aperture - but not where the obstruction would be. I guess if you're looking for small reductions in effective aperture on the order of a millimeter, then one might have to be much more accurate.

It is a very interesting topic on why such reduced aperture exists in some Mak designs (particularly, one might assume, those without an ashpere.) They seem great candidates for such a test and the validity of such a test is certainly pertinent.) The question arose, why would some models use undersized primaries while others had extremely tight baffling (the latter in my case and evident through inspection. Very tight, too tight.)

Was the restricted aperture designed to reduce the peak aberration from the edge? Maybe, that's a great unanswered question. I have been studying it, as you know, and still don't have any hard answers. Just speculation based on difficult star testing, studying the design, and your own input.

I hope this is not veering off topic while discussing a real world application of such a flashlight test that may speak to it's validity. I do want to offer my own testimony that the flashlight test certainly did work to show a constriction existed and that it was consistent with other tests despite any error it may have.

Post modification to eliminate the restriction, the pin hole test, inspection of the meniscus edge (thanks Glenn) and the equation you gave above both indicate full aperture is in play and the (primary) baffling remains very tight. IIRC, the flashlight test - the laser's beam - now illuminates the entire meniscus. Applying the laser test to a smaller, longer focal ratio 90mm MCT also showed a slight reduction in aperture of a few millimeters (to about 88mm, IIRC) with a very sharp projected cylinder. But, one might wonder if higher degree of accuracy is needed in this case. I was simply looking to see if it also employed a reduced aperture as other designs.

The Meade 7" MCT uses an aspheric primary, I wonder if it will pass the laser test showing full aperture. Intes designs, both basic and deluxe? Or does a more loosely baffled SCT offer reduced aperture? (I think not, and would that be due to the 5th order term of the corrector?)

Veering back on topic, I found the flashlight test most informative and consistent in studying the phenomenon of effective aperture. One can imagine applying it to all amateur optics to detect if such a condition exists.

A Newt or Mak Newt (famous for their small obstruction) is more confusing, though. Won't a laser coming off the diagonal show the fully illuminated FOV or will show that field for a reduced aperture when the diagonal is too small? It seems to show the fully illuminated field at reduced aperture, which should mean reduced resolution as well since the marginal rays are not in play?

[jhayes_tucson]

Frank,
You are over-thinking this. First, Gaussian beams are easy to propagate through an afocal system--and, they are not "non-linear." The diffraction characteristics are indeed different than an un-apodized beam but the results will not be significantly different enough for you to even measure using Glenn's method. In fact, a planar beam waist at the exit pupil will be transmitted (backwards) to become a planar waist at the entrance pupil. That means that the rays at those two locations will be parallel. The divergence of the rays beyond a waist is determined by the diameter of the beam, which means that in this case, the output beam will not measurably diverge by very much at all. This is why you start with a very large diameter beam if you want to deliver a large amount of energy into a small area on a distant target. This is a simple geometric test that will work well enough to measure the diameter of the entrance pupil to within a couple of millimeters, which should be close enough for checking that a system meets the manufacturer's aperture spec.
John

[freestar8n]

Again - this stuff is described in detail in introductory texts. Gaussian beams do not obey normal linear optics - e.g. the lensmaker's equation. Beam expanders work perfectly well - but if you set one up using the lensmaker's equation, the output would not be "collimated." That's why Oslo has special treatment for Gaussian beams. Also - in linear optics there is something called a "collimated" beam - even though diffraction prevents such a thing from existing in reality. In Gaussian beam optics, there is no such thing as a collimated beam - only a beam with a waist at a certain location and a given divergence angle. When such a beam passes through a lens, the waist on one side is imaged to a new waist on the other side. In normal optics, as you move the source toward a lens, the image of it will move away; in Gaussian optics, the image of the waist may move toward the lens - exactly opposite.

You have invented a measurement technique and you describe it as simple, accurate, and reliable - but it incorporates a ton of baggage through the introduction of unnecessary and complicating systematic errors. Unless you have a good handle on all the principles involved, and the corresponding impact they have on the resulting error, you really can't say how accurate it is. A key factor multiplying the resulting error is - once again - the distance to the pupil in object space.

A primary goal in a good measurement is to eliminate sources of systematic error - and your method goes the opposite direction - creating a house of cards all balanced against each other with the hope there is no bias introduced in the output. But in fact there is bias being introduced all over the place and it will be there in the final measurement.

I still don't know the mindset of people who want to do this measurement - but at the same time don't want to learn about the principles involved - especially the definition of entrance pupil and its relation to a physical stop somewhere in the telescope. If the measurement is important to someone and needs to be accurate, then I would use the ISO travelling microscope approach, which would be much more reliable and allow the direct identification of the limiting stop. If the measurement is not important and the goal is just to play around and look at a glowing screen, then I recommend a flashlight, some red wine, and a kaleidoscope.

Frank

[freestar8n]

Hi Norme-

I appreciate the measurements you have made but I still don't know exactly how to interpret them. The unfortunate thing for me in all this aperture stuff is that no one has ever set up the travelling microscope method of measurement, which is literally an industry standard. This method has very few sources of systematic error and only relies on the optical components that are creating the entrance pupil image in the first place. It does require a decent linear traveling stage and small telescope, and there should be an ability to adjust its height, but it would allow a direct physical measurement of the pupil diameter that is consistent with its definiton: the apparent diameter seen from object space. At the same time, by guaranteeing the view on the left side of the pupil is identical to the view on the right, you know you are measuring the correct stop - and you can take pictures of those views and provide them as evidence the correct measurement was made and the stop was identified, as a form of validation.

In all these different optical measurements of a system, there is a need for a ground truth reference to indicate just how accurate it is. With an unknown system and an unknown aperture stop - the ground truth is unknown, and it is hard to tell just from the measurement if it is accurate or not - unless it explicitly avoids systematic errors and has a direct, self-validating behavior - as does the ISO method.

As I keep saying - in many cases the flashlight method should work fine - but in some cases it won't. If you have no idea where the pupil is, then you have no idea if it is working or not.

Frank

[Asbytec]

Hi Frank,

Let me say, first, I really appreciate Glenn discussing this topic. As you know, it drives straight at the heart of a topic that interests me greatly. I do read your input closely and appreciate what you have to say. And don't disagree. In a round about kind of way just saying thank you for contributing. Also Nils, Mike, and others.

The conversation seems to be focused between you and Glenn (its a fussy topic, understandably), and there is some information to be picked up in the process. But, it leaves some questions unanswered. I WOULD like to know the difference between the entrance pupil and the aperture stop and what affects they have when they are not in the same location - both observational and as a result of testing. It seems a bit of improved aberration at the expense of a fully illuminated field would be the trade off. The converse being true when the aperture stop is placed at the objective...and in a CAT that seems like a complex issue.

Truthfully, I am not qualified to interpret my own test observations, just kind of in a position to make them. Effective aperture could well be very misleading from one test to another. I wish I had the knowledge and the equipment to make precise measurements. Instead, and thanks to what you've said, it seems like the sharpness of the projected cylinder might be a measure of accuracy in itself. Or it might be off all together. But the flashlight test does seem to show aperture is being restricted (or not.) That was my case, at least, and all tests showed it to be reduced by about the same amount - to within some unknown measure of accuracy.

If you have no idea where the pupil is, then you have no idea if it is working or not.

That makes intuitive sense. Even if I knew where it was, and I do, I still would not know how to deal with it other than just eliminating it all together. I did toy with the math of how far to trim it back, based on what data was available, but that proved difficult and involved further trade offs, as well.

I am still struggling to understand why the fuzziness of the projected shadow could be misleading - or dramatically so. Is it due to an expanded light source offering off axis illumination as one graphic above shows? Much like the soft shadow cast by the sun relative to the object's distance casting that shadow. Or is it diffraction passing through the system backwards that does strange things? It seems a collimated beam would eliminate or reduce errors to an acceptable level. But, that's only intuitive knowing little about the behavior of Gaussian image space.

In any case, this is just a learning experience for some of us...maybe only me because everyone else seems well opinionated on the subject. {LOL} My own reduced aperture is fixed and I've moved on to the observing stage, as Nils said. But, I still want to know what's going on inside the optic and whether adding some aberration is worth the use of full aperture. Observation seems to suggest it's just fine despite the additional marginal ray aberration being unleashed. Such topics seem pertinent to folks who want to explore effective aperture - even too small an iris located at or beyond the exit pupil.

Sorry for being verbose...but it's an interesting topic if not a hotly debated one. Thanks for humoring me.

[freestar8n]

Hi Norme-

No one has provided an error analysis or answered my questions in my first post, but as a concrete example - if I have a 20" f/15 system with a small obstruction 750mm, or 1/10 efl, from the focus, then the entrance pupil will pushed back 9xf away, or a distance of 67.5m. If there is a taper of only one minute of arc in the beam, projected over that football field distance, the error in the measurement will be 19.6mm - and that does not include the difficulty in interpreting the fuzzy edge. I think that in most of these scenarios, 1 minute of arc is optimistic - and for a general system the pupil could be even farther away. Plus - there are other sources of systematic error that are not considered here. Claiming that 1mm is a reachable accuracy in a very general way, and that the pupil distance does not play a role - seems crazy to me.

Frank

[freestar8n]

Hi Norme-

The separate thing about your mak is relevant here because a good way to know exactly what the limiting stop and entrance pupil are - is to take the system apart, measure accurately, and do a ray trace. The problem with the data for your mak was - I could not find a system that worked well with spherical surfaces, and also fit the values you provided. This touches on my personal interest in spot maks with and without aspheric primaries - and in this case it means unless I know the actual conics involved and can make an accurate ray trace, I can't really be confident in the layout. The first order stuff should be ok - but I like to see that the optical performance and the layout of the ray trace agrees with the real thing.

Since then I have taken my own mak apart and confirmed it has an aspheric primary, and that the performance looks good - so I can look at the role baffles play in it. I haven't finished this but I will probably report on it later - and it is relevant if I do find an internal aperture stop.

Frank

[GlennLeDrew]

I have provided a schematic of the method.

I have provided a clear description of the technique.

I have provided a means of verifying utility.

I have provided some basis in theory.

I have demonstrated experimentally the validity of the method.

What more must I do?

[Jon Isaacs]

You can get very expensive 20mm binocs - and if they are actually 19.5, and a competitor has 20mm, then that is a non-trivial advantage.

In my book that is a trivial advantage and certainly the difference between a binocular operating at the full 20 mm and that same binocular masked to 19.5mm could not be detected visually by even the most discerning observer.

Jon

[freestar8n]

Well - this is your test and you said it was accurate to one millimeter - and the accuracy was independent of the pupil distance. Since these are your claims about the test, and they are at odds with the counter-examples I have provided, it seems like those claims need to be modified and there is a need for more caveats on the usage of the test, the care needed, and when it is not a good idea to use it.

I think it is not good to encourage the use of such a test in a vacuum from pre-existing methods, and from the practical limitations that would make it very error prone even in good hands. It serves no pedagogical purpose since it intentionally avoids the concept of pupils and image space in the interest of not scaring newbies off from doing the test - and keeping it "simple."

This is the second "flashlight test" that I know of - and the other one is where a newbie gets a brand new sct and shines a flashlight into it - and sees horrible swirls and dust on the corrector. They are told not to do that test because it makes things look bad even in a very good and clean optical system. I would say this flashlight test is also something to be avoided. If the answer matches the expected value then nothing has been learned, and if the answer suggests it is undersized, without knowing the limiting aperture and the care taken in performing the test, it is too prone to systematic errors to be interpreted with confidence.

Although you have provided some examples of it working, in order to estimate the error in the simplest form, you would need to provide expected tolerances on the taper of the beam when prepared by a newbie - after whatever method of ensuring it is "collimated" - along with the expected distance to the pupil. Since this is your test, you would need to provide good estimates for these parameters in order to make a statement of the accuracy. This is not my test at all - and when I look at it I just see a morass of completely unknown systematic errors injected by a user with unknown skills. I myself have no idea how to estimate them - but it is not my test so I don't have to.

The ISO method does not have such user-dependent sources of systematic error because there is no carefully prepared beam involved, and there is no shadow thrown a long distance with fuzzy edges that has to be "collimated" and then measured.

If you really want to encourage usage of this test somehow, I recommend using my list provided earlier, which indicates the special cases where it should work robustly. IF the user knows, or suspects, the offending stop, and IF the pupil is near the front of the 'scope, then the test would be fairly reliable. IF the user knows the stop is near focus and there are no lenses back there - the geometrical approach would work well. Anything else is a special case that may not be suitable.

No matter what - the travelling microscope test should be more robust - and still "fun."

Frank

[freestar8n]

Yes I agree. It isn't my concern if the 19.5 is a problem or not - I am only talking about the accuracy of the test. In fact - I am also one of the people asking - Who cares? and Why are people being encouraged to measure aperture in the first place?

I do think that aperture should be stated accurately and not "skimmed" from the consumer - but that is not my primary interest in this test. I think the test is prone to systematic errors, it is based on an over-simplified view of a complex situation, and there is a pre-existing better way to do it.

Frank