List of 'Dandelion' metering chipped Nikkors

[rodeo_joe1]

However, surely telepathy has a bandwidth like WiFi and Bluetooth etc??...

If it exists at all, it probably involves something very esoteric like quantum entanglement... so how many angels can dance on the head of your pin?

 

No.... don't tell me... I'm getting a number... a big number... a very big number.

Was it 42?

And was this (produces Jack of clubs) your card?

[chulster]

A set of needle files and a junior hacksaw are much more controllable. The 'off switch' is instant and the tool doesn't run away and cut randomly on its own!

 

Joe, i’ve been thinking about this lately because i want to chip my 105mm f/2.5 AiS, another lens that requires some material removal before chipping. I’m trying to figure out how to use these tools (which i don’t own yet) to do the job. I can picture using a fine saw to make two vertical cuts where the ends of the chip will be, but how do i remove the material in the middle? Do i use the files to remove all of it a little at a time? Or is there a way to make a transverse cut that bridges the two vertical cuts to remove all of the material at once? I have no experience in metalwork.

[rodeo_joe1]

but how do i remove the material in the middle?

The aluminium files away a lot quicker than you might think. So it's not a very laborious task.

 

I used a 3/8" square coarse file to remove most of the material between the saw cuts, then finished off the last 1/2 mm or so with a selection of needle files. The most useful profiles being a 3.5 mm wide flat file with safe-edge, and a 3mm triangular file to get the corners sharply defined.

 

Be patient. Take it slow and easy with the final cuts. You can always take a bit more material off, but you can't put it back!

 

Oh, and hold the job in a padded clamp or vise. That leaves both hands free to control the tools.

[chulster]

The aluminium files away a lot quicker than you might think. So it's not a very laborious task.

 

I used a 3/8" square coarse file to remove most of the material between the saw cuts, then finished off the last 1/2 mm or so with a selection of needle files. The most useful profiles being a 3.5 mm wide flat file with safe-edge, and a 3mm triangular file to get the corners sharply defined.

 

Be patient. Take it slow and easy with the final cuts. You can always take a bit more material off, but you can't put it back!

 

Oh, and hold the job in a padded clamp or vise. That leaves both hands free to control the tools.

 

Thanks, Joe! That does sound much more manageable than the Dremel was. I'm going to try it.

 

Please have a happy new year.

[rodeo_joe1]

Thanks, Joe! That does sound much more manageable than the Dremel was. I'm going to try it.

 

Please have a happy new year.

Cheers Chulster. A happy new year to you too.

 

One more tip: There's a risk of scratching the opposite side of the ring you're filing, so it's wise to pack or cover it with cardboard. Not too thick though, otherwise the 'stroke' of the file is reduced.

[chulster]

Cheers Chulster. A happy new year to you too.

 

One more tip: There's a risk of scratching the opposite side of the ring you're filing, so it's wise to pack or cover it with cardboard. Not too thick though, otherwise the 'stroke' of the file is reduced.

 

Gotcha. :)

[richard_driscoll]

Yes, but user reg_eggen stated that he was using the aperture ring. I suspect he read and understood the warning about AI non-linearity. What then explains the exposure discrepancy? Not a rhetorical question; this is something that has mystified me since I read that warning myself. I wonder if the camera, "knowing" that the lens is a P (and thus also AI-S) lens, therefore does not fully depress the aperture control lever during an exposure, but only moves it to where it should be if the lens were AI-S.

It's my understanding that the matrix meter (and probably the centre-weighted meter to some extent) needs to know the position of the exit pupil to work properly. This information is provided by the Nikon lenses with a "cpu." The implication is that the exit pupil position provided by the Dandelion chip should match that of the lens which is being modified.

 

When you put a non-cpu lens on a digital body and provide the lens focal length (and max. aperture) I've always imagined that the body uses this information to estimate the exit pupil pupil position. It could do this by lookup table except that the combination of maximum aperture and focal length does not uniquely identify a particular lens design so I imagine that most lenses of similar aperture and focal length tend to have the exit pupil in a similar position and that is good enough. I must read up more about this Dandelion chip.....

[chulster]

It's my understanding that the matrix meter (and probably the centre-weighted meter to some extent) needs to know the position of the exit pupil to work properly. This information is provided by the Nikon lenses with a "cpu." The implication is that the exit pupil position provided by the Dandelion chip should match that of the lens which is being modified.

 

When you put a non-cpu lens on a digital body and provide the lens focal length (and max. aperture) I've always imagined that the body uses this information to estimate the exit pupil pupil position. It could do this by lookup table except that the combination of maximum aperture and focal length does not uniquely identify a particular lens design so I imagine that most lenses of similar aperture and focal length tend to have the exit pupil in a similar position and that is good enough. I must read up more about this Dandelion chip.....

 

It's certainly food for thought, and a good surmise.

[rodeo_joe1]

I've always imagined that the body uses this information to estimate the exit pupil pupil position.

I think you're grossly overestimating the complexity and sophistication of Nikon's metering system.

 

Why would the lens exit pupil distance be of any use to the metering sensor, when it's positioned above the focussing screen? In such a position that any direct rays from the lens have already been 'scrambled' by the ground glass (or etched plastic) surface before hitting the sensor.

 

And incidentally, in such a position that light entering through an unshielded eyepiece can grossly affect the meter reading. An effect that anyone using a Nikon SLR or DSLR tripod-mounted and in bright light ignores at their peril.

 

Nikon's metering has been prism-based for decades, and those old centre-weighted TTL systems had absolutely no way of 'knowing' the lens focal length, let alone where the exit pupil was.

Edited December 30, 2019 by rodeo_joe|1

[richard_driscoll]

I think you're grossly overestimating the complexity and sophistication of Nikon's metering system.

 

Why would the lens exit pupil distance be of any use to the metering sensor, when it's positioned above the focussing screen? In such a position that any direct rays from the lens have already been 'scrambled' by the ground glass (or etched plastic) surface before hitting the sensor.

 

And incidentally, in such a position that light entering through an unshielded eyepiece can grossly affect the meter reading. An effect that anyone using a Nikon SLR or DSLR tripod-mounted and in bright light ignores at their peril.

 

Nikon's metering has been prism-based for decades, and those old centre-weighted TTL systems had absolutely no way of 'knowing' the lens focal length, let alone where the exit pupil was.

Sorry Joe but that argument can't possibly hold water. If it were true there would be no difference between metering systems with one element, several hundred elements or several thousand. It can only be that the screen has a partial but not total diffusing effect. In the case of the old Nikon cameras they were all centre weighted and I'll bet the degree of centre weighting varied with the exit pupil position. There were articles in the BJP back in the sixties discussing the effect of exit pupil position on TTL metering.

BTW I can't find any statement about what the Dandelion chip does with the the exit pupil position but my guess is that it sets it to be the same as the programmed focal length. Can you confirm that with any of yours? The illumination will fall off as cos**4 and so will be most marked with lenses where the exit pupil is close to the plane of the sensor.

[rodeo_joe1]

Sorry Joe but that argument can't possibly hold water.

Well, just have a look at the cross-sectional view of almost any Nikon (D)SLR. The meter is clearly in the prism housing, and well above the GG screen.

 

With later cameras the sensor has an auxillary lens in front of it, which presumably projects an image of the focussing screen onto the sensor and allows multi-segment analysis thereof. The screen is just as diffusing as earlier type B screens, with no clear centre-spot or other apparent means for the meter sensor to receive direct rays from the lens.

The illumination will fall off as cos**4...

That's only true for a simple meniscus lens, pinhole or theoretical thin lens model. Real-life complex lenses 'distort' the aperture shape when viewed off-axis; such that the Cos^4 'rule' doesn't apply. Otherwise no fisheye lens would be possible, and a modest 24mm wideangle would fall off by over 2 stops in the corners, regardless of aperture used. This is clearly not the case with most lens designs.

[richard_driscoll]

Well, just have a look at the cross-sectional view of almost any Nikon (D)SLR. The meter is clearly in the prism housing, and well above the GG screen.

 

With later cameras the sensor has an auxillary lens in front of it, which presumably projects an image of the focussing screen onto the sensor and allows multi-segment analysis thereof. The screen is just as diffusing as earlier type B screens, with no clear centre-spot or other apparent means for the meter sensor to receive direct rays from the lens.

 

That's only true for a simple meniscus lens, pinhole or theoretical thin lens model. Real-life complex lenses 'distort' the aperture shape when viewed off-axis; such that the Cos^4 'rule' doesn't apply. Otherwise no fisheye lens would be possible, and a modest 24mm wideangle would fall off by over 2 stops in the corners, regardless of aperture used. This is clearly not the case with most lens designs.

I'd imagined that the lens in front of the metering sensor on the digital cameras is there to form an image of the scene on it's surface. It does this by using the real image in the plane of the screen as its object. The prism system, if well designed, shouldn't have much effect on this.

 

As well as the imaging system we now have to consider the illumination system which is where the exit pupil of the main lens comes in. The cone of light from it is incident on the screen and the condenser, fresnel lens or whatever they are using now ensures that (hopefully) most of the light is not lost, either by the magnifying eyepiece or the metering system. Some of the light goes straight through the screen, some is scattered, it depends on the design of the screen, and I suspect there is a tradeoff between brightness and ease of focussing.

 

In any event my suppositions are all (retrospectively) pretty well confirmed by bjørn rørslett on this site back in 2007:-

Nikon DSLR Firmware Updates - Wish List

 

"A final issue is that the Nikon light metering system really needs information on the position of the exit pupil of the lens in order to work reliably, and my guess is that most users haven't a clue what that is not to speak of its position (which, for a zoom lens, does vary along the zooming range as well). All Nikkor CPUs do report the exit pupil position to the camera and I have verified that giving the camera false values of the pupil's position *will* lead to severe errors in the calculated exposure. Hence even inputting the max./min. f-numbers and the focal length of the lens won't guarantee good metering unless the camera finds an exact match for a Nikkor lens in its database and can extract the missing data from there."

 

Of course for the matrix meter to work properly it needs to measure the absolute scene brightness anywhere on the meter sensor surface and any undefined falloff will make this impossible. It needs to do this so that it can classify scenes and also ignore extremely bright areas which are considered to be the light source or bright refection of it.

 

My understanding with wide angle lenses is that the light loss is much reduced when the lens is of retrofocus design since the exit pupil is then further from the focal plane than the focal length. For example, my D7200 reports the exit pupil position of the 35 mm 1.8G DX as something over 100mm IIRC, I don't have any fixed focus wide angle's with ICs in them but at 10mm my new 10-20 zoom has the exit pupil at 70.6 mm (I just looked) so in these cases the focal length doesn't really tell you anything about the illumination falloff.

[rodeo_joe1]

So how does the metering apparently cope perfectly well with manual focus Ai-S lenses that have no way of communicating with the camera at all?

 

I'd also like to know what hardware and software Bjorn Rorslett used to get access to and emulate the lens data and camera manipulation of it.

My understanding with wide angle lenses is that the light loss is much reduced when the lens is of retrofocus design since the exit pupil is then further from the focal plane than the focal length.

That has very little to do with it. The Cos^4 'law' is based on the mistaken assumption that the aperture shape becomes linearly and increasingly oblate when viewed at angles away from the lens axis. This is not the case, as can easily be verified by looking into the back of any real lens and rotating it. It doesn't matter if your eye is placed at the exit pupil position or otherwise, and is an effect also visible in non-retrofocus lenses.

Edited January 2, 2020 by rodeo_joe|1

[richard_driscoll]

So how does the metering apparently cope perfectly well with manual focus Ai-S lenses that have no way of communicating with the camera at all?

 

I'd also like to know what hardware and software Bjorn Rorslett used to get access to and emulate the lens data and camera manipulation of it.

 

Because all the discussion about the absolute illumination of a multi-element metering sensor is concerned with matrix metering isn't it? In the case of the FA and F4 there is some mechanical communication 'twixt AI or AIS lens and camera I believe but I don't know the details. We all knew that the digital cameras can do C/W metering without extra data, though Nikon say in my manual that it improves if given lens data.

 

IIRC Bjorn used some samples of Nikon's own chips and installed them in some lenses he had but I don't know for sure.

[rodeo_joe1]

Because all the discussion about the absolute illumination of a multi-element metering sensor is concerned with matrix metering isn't it?

On those DSLRs that support use of non-CPU lenses, color matrix metering is available with any AI coupled lens; including ancient AI modified lenses.

 

3D matrix metering is disabled, but that's unsurprising since the camera has no subject-distance information.

 

I suspect that any metering dependence on exit-pupil position is purely due to the interaction between it and the Fresnel condenser focal length beneath the GG screen. This will naturally affect the efficiency of the viewfinder/metering system and introduce a slight error factor. Non-linearity of screen brightness with aperture number will have a far greater effect.

Edited January 2, 2020 by rodeo_joe|1

[chulster]

It's my understanding that the matrix meter (and probably the centre-weighted meter to some extent) needs to know the position of the exit pupil to work properly. This information is provided by the Nikon lenses with a "cpu." The implication is that the exit pupil position provided by the Dandelion chip should match that of the lens which is being modified.

 

When you put a non-cpu lens on a digital body and provide the lens focal length (and max. aperture) I've always imagined that the body uses this information to estimate the exit pupil pupil position. It could do this by lookup table except that the combination of maximum aperture and focal length does not uniquely identify a particular lens design so I imagine that most lenses of similar aperture and focal length tend to have the exit pupil in a similar position and that is good enough. I must read up more about this Dandelion chip.....

 

Using exiftool just now I discovered that a Dandelion-equipped lens reports its Exit Pupil Position as 0.0 mm, or else it doesn't report any value for it at all. At any rate, 0.0 mm is what's recorded in EXIF. Nikkor CPU-equipped lenses, by contrast, report meaningful values such as 75.9 mm for the 50mm f/1.8G.

 

Now I'm curious exactly how exit pupil position affects metering. I should start by learning what the exit pupil is!

[rodeo_joe1]

Now I'm curious exactly how exit pupil position affects metering. I should start by learning what the exit pupil is!

It's simply the virtual (apparent) size and position of the aperture opening, as seen from the rear of the lens. Specifically, at the focal plane.

 

You can get an approximate idea of its position using parallax - i.e wiggling your head from side to side and comparing the apparent position with your finger or the point of a pencil.

 

The exit-pupil position obviously has an effect on screen brightness with those cameras having a Fresnel condenser underneath the screen, due to light-gathering efficiency. That's nearly all modern DSLRs. However, it's difficult to see how it could much affect those cameras like the F2, F3 and F4 where the condenser is above the screen. So early mithering about TTL metering and exit-pupil position was purely a hair-splitting exercise IMO.

 

A much more puzzling phenomenon is that lens apertures above f/1.8 make absolutely no difference to the visual or measured screen brightness of any Nikon (D)SLR, of any vintage or screen design. So how the above-screen metering system copes with this is quite mysterious to me.

 

Also, FWIW, modern lens designs for digital sensors are (usually) made as telecentric as possible. Telecentricity, by definition, places the exit pupil at infinity, or at least a great apparent distance. Therefore modern digital lens designs should only show a relatively small variation in exit pupil position as far as any metering system is concerned. That said, true telecentricity requires a rear lens group with a diameter equal to, or greater than the sensor diagonal, and that's obviously not the case with many lower-priced lenses.

Edited January 3, 2020 by rodeo_joe|1

[rodeo_joe1]

It's simply the virtual (apparent) size and position of the aperture opening, as seen from the rear of the lens. Specifically, at the focal plane.

I should have added 'shape' to that list of virtual aperture parameters as well.

[chulster]

It's simply the virtual (apparent) size and position of the aperture opening, as seen from the rear of the lens. Specifically, at the focal plane.

 

Thanks, Joe. Now, pardon my ignorance, because the next questions are founded in it.

 

At the focal plane is either the main sensor or the focusing screen, depending on whether the main mirror is up or down. Assuming you've mounted a lens that projects an image circle large enough for the format, how does the position of the exit pupil influence the image falling on either surface? Does the exit pupil position directly influence light falloff away from the center of the image? Does it change how well the focusing screen mimics the image sensor, due to the Fresnel condenser you mentioned?

[rodeo_joe1]

Does the exit pupil position directly influence light falloff away from the center of the image?

Errrm, yes and no.

A short exit pupil distance may show increased physical (catseye) vignetting, caused by rays from the aperture being blocked by element edges or by the lens barrel. This sort of vignetting rapidly diminishes as the lens is stopped down.

 

Cos^4 type vignetting depends only on the relative angle that the aperture circle makes to any particular position on the image plane, and theoretically is therefore independent of pupil position and aperture number. But as discussed above, the apparent shape of the aperture in a real lens rarely follows the Cos^4 law by becoming an oblate circle.

Does it change how well the focusing screen mimics the image sensor, due to the Fresnel condenser you mentioned?

Yes. Most definitely if the Fresnel is placed below the GG surface. Because ideally the Fresnel should focus all the light from the aperture onto the screen. This doesn't happen if the exit pupil doesn't coincide with the Fresnel focus, but it's not a drastic change, just a drop in light-gathering efficiency.

[richard_driscoll]

It's over 40 years since actually designed any of this stuff and that was a flying spot slide scanner for my PhD project. My book on engineering optics is even older and was not terribly helpful.

 

It must be our lucky day because IOP Publishing have decided to make part of their "Physics of Digital Photography" available as free to read:-

https://iopscience.iop.org/book/978-0-7503-1242-4

The section available is the one on fundamental optical formulae.

 

Section 1.4.7 on natural vignetting is relevant to our discussion. It points out that the image space angle version the cos**4 law is modified by the pupil magnification factor which in turn depends on the lens design with telephoto lenses having mp less than 1 and retro-focus lenses greater than 1.

 

As well as that there is a section on focus and recompose and its relation with DOF which I still have to read. I worked out some of this a few years back so it's interesting to see if the conclusions are the same.

[richard_driscoll]

........ And if you can cope with the errors and the mirror image diagram this shows how much better the illumination of a retrofocus 21mm f/2.8 is than a rangefinder version. Can't say I'm convinced by the explanation though.....

http://www.cs.cmu.edu/~sensing-sensors/readings/vignetting.pdf

[chulster]

Errrm, yes and no.

A short exit pupil distance may show increased physical (catseye) vignetting, caused by rays from the aperture being blocked by element edges or by the lens barrel. This sort of vignetting rapidly diminishes as the lens is stopped down.

 

Cos^4 type vignetting depends only on the relative angle that the aperture circle makes to any particular position on the image plane, and theoretically is therefore independent of pupil position and aperture number. But as discussed above, the apparent shape of the aperture in a real lens rarely follows the Cos^4 law by becoming an oblate circle.

 

Yes. Most definitely if the Fresnel is placed below the GG surface. Because ideally the Fresnel should focus all the light from the aperture onto the screen. This doesn't happen if the exit pupil doesn't coincide with the Fresnel focus, but it's not a drastic change, just a drop in light-gathering efficiency.

 

Fascinating stuff. Thanks for explaining!

[richard_driscoll]

Fascinating stuff. Thanks for explaining!

That second reference I posted shows how the Biogon suffers from pretty dire cos**4 vignetting which is much the same whatever the stop. The Distagon is a retrofocus design and has the exit pupil further away making the cos**4 vignetting much less. It suffers from optical vignetting wide open but that disappears as the lens is stopped down.

 

If I worked it out right the loss of illumination on the Biogon is about 2.8 stops!! Is it really this bad? I don't know but I've seen colour slide film shot with the 21mm non-retrofocus Leica (Schneider) Super-Angulon that has really dark corners. Some used to say that you could use this creatively - perhaps. In the film days they used to say that if you had a lousy enlarger it made things better in a B&W print because the poor coverage of the enlarger reduced the vignetting due to the camera lens!

 

IIRC Zeiss offered a 16mm Hologon for the relatively recent Contax 35mm film camera. That came with a special ND filter with a dark middle to compensate for the vignetting.

 

We are getting a bit far from discussions about Dandelion chips here. My guess is that it's only with wide angle lenses that there is likely to be a problem. It's probably fortunate that if the chip reports zero (or is it nothing at all) for the exit pupil distance that Nikon probably sets it to be the same as the focal length or otherwise make no compensation for falloff. In principle all bets are off guessing what they might do.....

[rodeo_joe1]

It's probably fortunate that if the chip reports zero (or is it nothing at all) for the exit pupil distance that Nikon probably sets it to be the same as the focal length or otherwise make no compensation for falloff. In principle all bets are off guessing what they might do.....

Exactly!

 

A pity that the publishers of that vignetting.pdf didn't bother to check the orientation of their illustrations. I'd also take issue with their involvement of the inverse square law in a focussed spot's brightness. The inverse square law only properly applies to light radiating away from a point source with no lens involvement. Whereas in the case of focussed light from a real lens, there can be apparent magnification of the exit pupil diameter away from the axis. A situation where simple and linear theory falls down.

 

The only way to truly model a complex lens's behaviour is to ray-trace the design in its entirety, and that just can't be done using one or two simple formulae.