View Full Version : Checking calibration of aperture scales
there seems to be a lot of people, myself included, swapping lens cells between shutters and then worrying about how accurate the aperture scales are. Or not worrying and perhaps getting caught out with bad exposures? Or having no scales at all and wanting to construct new ones.
I think I've got an easy method for checking the scale calibration [and perhaps constructing new scales] that just uses the gear most of us have to hand... but can those who know more please tell me if there's a basic flaw in what I've come up with ???
here goes:
Start with any lens that's know to be stock standard and assume that the aperture scale is correct because it's come from the factory. This becomes the reference lens.
Mount the reference lens on the camera, set it to any chosen f-stop and focus infinity. Then point the lens at a bright uniform light source like an area of clear sky and get under a darkcloth to reduce extraneous light. Get a light meter with the dome withdrawn, hold the light meter directly on the ground glass and measure the brightness in EV.
Then get the suspect lens and do exactly the same thing ie focus it to infinity, set to the same f-stop, point it at the same light source/ same area of clear sky and measure the EV of exactly the same "object" while under the darkcloth to reduce extraneous light
The EV should be exactly the same.... if there's a difference it could be the light has actually changed between measurements [!], you aren't measuring the same point or the aperture scales are inaccurate. If you're happy the light hasn't changed and the difference is due to the scales, the difference in EV gives you the magnitude of the error.
I've tried this with a few unmolested lenses from 125mm to 305mm and the EV readings are pretty consistent so I think it works. There's been some variations in EV readings that could be explained by measuring different areas of sky and the light changing between measurements but the variation was small.
If that's all valid and you find an error in the aperture scale you're testing, you could just open/close the aperture till the reference EV reading is reproduced and that'll give you the point which corresponds to the selected f-stop you started with on the reference lens.
feedback/ comments please!
.
Bill Burk
30-Aug-2013, 18:19
Seems like you should determine one aperture by actual exposure tests to get the transmission value... Then use your technique to find all the other stops.
ic-racer
30-Aug-2013, 18:22
Or just use a film-plane meter all the time and ignore lens aperture scales (Horseman meter for example).
Hi Andrew,
Sounds like that method should work well.
Remember that shutter aperture assemblies have tolerances, just like every other manufactured item.
I'm sure they're not worried about a 1/10th stop error, so your measurements may vary a bit.
Modern meters will easily catch changes of 1/10th stop.
- Leigh
BetterSense
30-Aug-2013, 19:17
Why focus on infinity? If you are just comparing brightnesses, you could place the lens radically OOF, and it would help to blend together any variations in scene brightness.
Why focus on infinity?
You focus at infinity because that's the position at which the aperture scales are correctly calibrated.
The scales are in error at any closer focus.
The error varies with the shift from infinity focus, increasing as the focus distance decreases.
As you focus closer, the lens moves farther from the film, and the area of the circle of illumination increases.
- Leigh
Leigh, thanks for commenting.... I was hoping you would.
frankly, I'm not even too worried by 1/3 f-stop errors and if I can be that accurate reliably then I'm happy
Bob Salomon
31-Aug-2013, 07:28
The ground glass absorbs some light. So a direct reading through a ground glass will not be correct unless you factor in the absorbtion due to the gg. That means that you have to first measure a grey card with the light source you want to use through the gg. Then measure the same gg directly. The difference is the factor and you apply it to your meter accordingly. Otherwise your scales that you make will be incorrect by the amount of the factor.
This technique also works when there is any bellows factor in play due to distances less then infinity.
The ground glass absorbs some light.
Hi Bob,
That's true, but irrelevant.
He's comparing the readings of a known lens with that of an unknown lens at the same aperture.
The known and unknown readings should be identical regardless of the characteristics of the GG.
The absolute magnitude of the reading is irrelevant to the comparison.
- Leigh
Bob Salomon
31-Aug-2013, 11:10
Until he puts it on another camera. Or changes the gg/Fresnel. Or go to sell it or loan it to someone else.
Bill Burk
31-Aug-2013, 11:23
I assume there is a standard procedure to calibrate apertures. Is this discussion covering them?
And once done, do you have f/stops or have you just gotten T/stops?
Peter De Smidt
31-Aug-2013, 12:01
Andrew, I've done what you're suggesting for a couple of lenses. It's worked just fine.
A few weeks ago there was a small brass device for calculating apertures on Ebay (UK), it was around 100+ years old. It was a calliper to measure the aperture diameter with a brass plate with focal lenght and f stop scales. Quite a useful tool.
Ian
Until he puts it on another camera. Or changes the gg/Fresnel. Or go to sell it or loan it to someone else.
Bob,
You totally misunderstand the concept of calibration.
You determine a value presented by a standard, then you compare the value from an unknown against it.
If the two values agree, then the markings on the standard and the markings on the unknown are equivalent.
This is the basis for calibration of all measuring devices of all types everywhere in the world.
Any lens set to f/11 puts exactly the same amount of light on the image plane* as any other lens set to f/11 when both are focused at infinity.
This is the entire basis of photographic exposure calculations.
And yes, I am well aware of your credentials.
That's why I'm surprised you would make such a fundamental error.
- Leigh
*Note: This does ignore some basic differences in efficiency, like transmissivity of the optical system.
That's what T/stops are for, when you need greater precision than afforded by the f/stop system.
Bob, I don't quite agree.... but you have flagged a potential problem
Yes there will be absolutes that aren't taken into account BUT the technique should take them out of the equation if they aren't allowed to vary. The only thing allowed to change is the f-stop value on the lenses. You'd definitely have to use the same camera and light meter, use the same area on the GG etc and then it's comparing like to like. And once the f-stop is "calibrated" against a lens that's known to be correct, there's no reason to think it won't still be accurate when transferred to another camera
if you mean that it's incorrect to use two camera in tandem for the initial calibration check then you're absolutely right to say there's a problem. The GG/fresnel introduced unknowable variables between the two cameras and the initial comparison definitely won't work because of variable light transmission. The test has to be the same camera.
And you couldn't focus the reference and test lenses to different distances, use different light meters, measure off a fluctuating light source etc
It only hangs together when the use the same kit at the same time in the same way
Andrew, I've done what you're suggesting for a couple of lenses. It's worked just fine.
THANK YOU, PETER... I like to think I'm fairly pragmatic and there's nothing more pragmatic than that statement !
The GG/fresnel introduced unknowable variables between the two cameras and the initial comparison definitely won't work because of variable light transmission by the different ground glass/ Fresnel.
I found over 3 stops difference in brightness between two 5x4 cameras by measuring the brightenss and comaparatively.
Ian
I found over 3 stops difference in brightness between two 5x4 cameras by measuring the brightenss and comaparatively.
Ian
exactly !!!!!!!!!!!!!!!!!!!!
I found over 3 stops difference in brightness between two 5x4 cameras by measuring the brightenss and comaparatively.
That sentence makes no sense.
Please describe what and how you were testing, and what results you obtained.
- Leigh
Bob Salomon
31-Aug-2013, 14:55
"Any lens set to f/11 puts exactly the same amount of light* on the image plane as any other lens set to f/11 when both are focused at infinity."
And that is your error. If you do it Andrew's way without taking in to consideration the light loss from the GG and Fresnel - and for this to work you need a Frenel to even the light across the gg) and bellows factor then you have not really found f11.
What I have described are the instructions for using the Linhof Focus/Metering bellows attachment with the light meter attachment from Gossen and a Lunasix meter. So the creds go back to the Linhof factory.
"Any lens set to f/11 puts exactly the same amount of light* on the image plane as any other lens set to f/11 when both are focused at infinity."
And that is your error. If you do it Andrew's way without taking in to consideration the light loss from the GG and Fresnel - and for this to work you need a Frenel to even the light across the gg) and bellows factor then you have not really found f11.
Bob,
You're intentionally twisting my words to cover your posterior. C'mon, Bob. You're above this tactic.
I said "the same amount of light on the IMAGE PLANE".
This has nothing to do with GG or fresnels or anything else.
In the comparison test that Andrew described, the same GG/fresnel is used for both lenses, so any issues
related to its transmissivity or dispersion are identical for both lenses, and thus irrelevant to the results.
I also said specifically "when focused at infinity", which eliminates bellows factor and other issues
related to the distance between the lens rear node and the image.
Let me paraphrase the test procedure in an attempt to clarify the technique...
Set up the test environment, and adjust the reference lens to f/11. Note the luminance reading*.
Replace the reference lens with the test lens. Adjust it until you get exactly the same reading.
The test lens is now set to f/11 within a very small margin of error.
- Leigh
*Note: Maximum accuracy is obtained by taking the readings directly on the optical axis for both lenses.
Jim Jones
31-Aug-2013, 15:17
Why not just measure the apparent diameter of the diaphragm as viewed from the front of the lens, and divide that into the focal length to get the optical f/number? Comparing image brightness introduces possible instrumental errors. True, such measurements do give something more like the old T/numbers, which may be desirable with some lenses and applications. However, with modern coated lenses, f/numbers seem useful enough and certainly more universal than our individual measurements.
Comparing image brightness introduces possible instrumental errors.
Comparative tests do not introduce errors.
This is one of the fundamental concepts of metrology.
It's the basis for all calibration of all measurements in all fields in all countries.
- Leigh
Bob Salomon
31-Aug-2013, 16:32
"I said "the same amount of light on the IMAGE PLANE".
This has nothing to do with GG or fresnels or anything else."
But the meter is behind the image plane, not at it, unless the reading was done with the Sinar/Prontor/Gossen in the film plane probe. And it wasn't. So the gg/Fresnel factor has to be taken into consideration for the calibration to be correct on any camera. Not just the one it was done on.
But the meter is behind the image plane, not at it, unless the reading was done with the Sinar/Prontor/Gossen in the film plane probe. And it wasn't. So the gg/Fresnel factor has to be taken into consideration for the calibration to be correct on any camera. Not just the one it was done on.
Yes, the meter is behind the GG/fresnel/whatever. But that fact is absolutely irrelevant to this test protocol.
We're doing a comparison, not making an exposure calculation.
When the measured values from the two lenses are equal, their apertures are equal.
It makes absolutely no difference how bright or dim the test image is at the point of evaluation.
The factor that you're talking about certainly is important when making actual exposure calculations.
That's not what we're doing in this test.
- Leigh
Bob Salomon
31-Aug-2013, 16:45
Yes, the meter is behind the GG/fresnel/whatever. But that fact is absolutely irrelevant to this test protocol.
We're doing a comparison, not making an exposure calculation.
When the measured values from the two lenses are equal, their apertures are equal.
It makes absolutely no difference how bright or dim the test image is at the point of evaluation.
The factor that you're talking about certainly is important when making actual exposure calculations.
That's not what we're doing in this test.
- Leigh
Forget it Leigh. The OP asked a question and I answered him. He will not correctly calibrate a lens with his technique. And if you are going to calibrate a lens then do it correctly so the calibration is correct for any camera. Not just the one you used to do the test.
The calibration protocol described previously is correct for any lens, regardless of the camera on which it's used.
In fact you don't even need a camera.
You could do the comparison with the lenses on an oatmeal box and a piece of mylar at the back.
Too bad you don't understand metrology.
- Leigh
The OP asked a question and I answered him. He will not correctly calibrate a lens with his technique. And if you are going to calibrate a lens then do it correctly so the calibration is correct for any camera.
I threw this open to constructive feedback and criticism....
if you think the idea is flawed that's fantastic but how do you suggest it should be done ??
I'm always trying to learn and I do want to sort out some equipment and not waste my time trying to do so
addit
Bob, the initial experiment that I did to try validating the idea was to get a small number of lenses covering a reasonably wide range of focal lengths and go thru the same measurement procedure with each and the result was that the EV readings that were within small fractions of an f-stop for each lens [except one where I'd swapped cells between shutters!]
Can you suggest an experiment that would be more convincing OR something that would invalidate the initial observation ??
Flauvius
31-Aug-2013, 18:42
It might be helpful to view the original poster's statement in terms of: How are f stops defined?
If defined in terms of the traditional formula: [F]ocal length divided by aperture diameter, your know whether your aperture scale's numbers are correctly positioned if the variable diameters of your lens' front element - when divided by the lens' focal length - equal the nominal exposure values shown on your aperture scale.
As such it is irrefutable that Leigh's analysis is correct: Given that f stops are geometric functions that are totally independent of a lens' light transmission characteristics, Bob Solmon's claims are not applicable to the op's comments.
Flauvius
Bill Burk
31-Aug-2013, 18:48
Let's take a step back for Andrew's sake...
At stake here is a set of lens cells in a foreign shutter without suitable markings.
Seems the original plan is fine... compare to some known lens/shutters and take it from there.
Bill, thanks but don't worry about me too much... I like to DIY but if it comes to it, I can access a technician who can make new scales for me [I'd just prefer not to]
the comment to my email enquiry was: "Best I can do is to calibrate the lens & diaphragm. Diaphragm scales are made especially for each focal length & each lens configuration. We rely on calculations then test transmission before completion of the scale" and "Can make up a correct one but will need the lens & shutter to be able to calculate & measure the transmission in order top make up an accurate scale".
he keeps coming back to measurement of light transmission being critical
To a large extent I'm trying to think thru how he could be measuring light transmission directly and rework it around the gear I already own.
from the comments so far, it sounds like most people agree the idea has validity.
worst that can happen if I'm wrong is that I flaff up a few sheets of test exposures.... but if it is as simple as I'm thinking, the method would be an easy fix for a other people who are in similar situation too.
Bill Burk
31-Aug-2013, 20:09
Transmission stops, as used in cinema, are critical... But I worry if I calibrated to transmission, would it throw off my normal calibration ... For example, I expose TMY2 at EI 250 based on experience and sensitometry. If calibration to T-stops gives me a more accurate measure of my aperture... does that mean I can shoot at EI 400 now?
For example, I expose TMY2 at EI 250 based on experience and sensitometry. If calibration to T-stops gives me a more accurate measure of my aperture... does that mean I can shoot at EI 400 now?
For still photography, T-stops are not necessary.
They're designed to produce extremely accurate levels of illumination at the film plane.
The are used so movie scenes shot with multiple cameras and lenses will appear uniform when combined.
The light lost through a lens of the type used on view cameras should be a very few percent.
This will not introduce a noticeable error in exposure. The error in true film speed will exceed the aperture error.
Movie camera lenses have many more elements, and thus transmission loss is increased.
- Leigh
Let's take a step back for Andrew's sake...
At stake here is a set of lens cells in a foreign shutter without suitable markings.
Seems the original plan is fine... compare to some known lens/shutters and take it from there.
I'm actually in this position, I bought an f4.5 14" telephoto lens at a flea market in July it's marked f4.5 but has no scale (or shutter) - it appears to have been a military lens. In fact I'll draw up a scale by measuring the diameter of the aperture and then double check it comparatively before testing. I've a front mounted Thornton Pickard shutter that'll fit it unfortunately my Speed Graphic was modified for Wide angle use or I'd use the lens with it.
Ian
Hi Ian,
In this case I suggest reversing the "roles" of the two lenses as follows:
Mount your f/4.5 lens and take a luminance reading. Record the value.
Mount the reference lens and adjust its aperture to achieve the same reading.
The reading shown on the reference lens scale should equal the true aperture of the f/4.5 lens.
- Leigh
Thanks Leigh, I'll try that - I figured on using the closest lens I have with similar bellows draw. I was also going to measure the aperture diameter at f4.5 and work from that to find the diameters needed for other f-stops. My Minolta spotmeter is fine for this.
Ian
I agree this is a useful test. In fact, I've gone a similar way more then once. BTW, focusing may be done not only at infinity; any distance is OK if only its the same for all the lenses. And one can also make this a projection test putting the lens on an enlarger and the meter on the enlarger base/table (but in this case the lens is to be mounted backwards as a lens's entrance and exit pupils are most often not identical). BUT...
May we assume the difference between T-stops (which are actually compared in the test discussed) and f-stops to be negligible for a modern LF lens? Well, that certainly depends in the accuracy we want to imply... Heaving measured lens transmission myself, I'm pretty sure we can't speak of 0.1 of a stop (that our meter yields) at all.
No, LF lenses don't usually have many elements but f/1.7 to f/2 50mm small format Planar derivatives don't either, and transmission in those differs quite a lot. 0.1 EV difference is even seen in comparing Zeiss T* to Pentax SMC - two really top-quality brands of coating that in fact were developed in collaboration from the start. And the pretty excellent in contrast control CZJ GDR coating lacks exactly 1/3 of a stop in transmission compared to SMC. And there are lots of brands pretty inferior to CZJ in their coating quality.
Besides, even all the rest thoroughly taken into account, our 0.1EV-accurate meter is as useless for evaluating anything close to real f-stops as a micrometer is for measuring a soft rubber band's width - unless every glass surface in each lens is perfectly clean. Think your second-hand lens that needs a shutter replacement is clean? Well, I've no doubt it is - for taking pictures. But not enough for tests. Apply a little ether with an eye-grade cotton swab and see the ugly shapes of dirt when the ether dries. No that dirt was not introduced by your swab. That's the dirt you had on your lens. Ether just made it visible to you by rearranging the dirt structure. And no the fact that you were unable to notice it does not mean it didn't influence lens performance. It did. To make a lens clean enough for tests, one needs dozens to hundreds of passes with those swabs with ether, each pass with a fresh swab, on each of the glass surfaces. And if not done in a proper laboratory-grade clean room, with any amount of efforts the lens still would not be as clean as needed for tests... not within 0.1EV of transmittance surely.
All the above is about just one problem: the light loss due to reflection. Alas, that problem is far from being the only one. As LF lenses tend to be bigger then those for other formats, for LF, light absorption in the glass itself becomes pretty prominent. Transmission of a 100mm Symmar is way greater then of a 300mm Symmar of the same formula. And an Apo-Ronar, regardless of the coating type, is transmitting noticeably less then a Sironar of the same focal length due to the much much yellower glass in the Ronar. The type of light and the meter spectral sensitivity do matter quite a bit, too (ever tried comparing an Aero-Ektar's T-stops to a Dagor's ones for a blue-sensitive film - or wetplate?).
So... as the repairman quoted above puts it, this type of test is useful as an additional verification to avoid an occasional blunder in calibration - but not as a method to do the calibration itself. And as far as I understand, the additional verification of the ready to use aperture scale was the very purpose the OP offered this method for. And as mentioned above, if in any doubt, I do occasionally perform similar tests, too.
But for actually making a new aperture scale, I'm pretty sure finding an original one to copy is the most sane way to go. Call a friend or post a question here, etc, to let someone owning the same lens measure at least one of the original manufacturer's iris diameters for you, and duplicate those. That the smartest thing one can ever do....
Except that for some old lenses, factors to multiply the iris diameters to get the entrance pupil diameters were actually published by the manufacturers. And with the factor known, there is no problem to make any new scale for any lens of the type. For example, below is the C.P. Goerz, Berlin data of 1908. For lenses specified, to get the actual entrance pupil diameter, multiply the aperture opening by:
Dagor III: (1+1/6)
Dagor IV: (1+1/7)
Pantar: (1+1/7)
Celor: (1+1/11)
Sintor: (1+1/9)
Alethar: (1+1/7)
Hypergon: (1+1/6)
Lynkeioskop C: (1+1/9)
Lynkeioskop D: (1+1/9)
Lynkeioskop E: (1+1/10)
Lynkeioskop F: (1+1/11)
Why not just measure the apparent diameter of the diaphragm as viewed from the front of the lens, and divide that into the focal length to get the optical f/number?
Because (as I've mentioned in this thread: http://www.largeformatphotography.info/forum/showthread.php?104407-How-are-the-aperture-scales-on-shutters-determined) you need a collimator, an optical bench, and a set of optical measuring instruments to get the real (not nominal) focal length and the entrance pupil diameter measured - at least if one insists on 0.1 of a stop accuracy... which I personally do not ever need in my own lenses BTW. :)
To Andrew- I use a very similar method to yours, except I use a DSLR mounted in place of the back. Set to manual, the shutter speed is adjusted until it produces a histogram to match the known value. Difference in shutter speeds will indicate difference in stops...
This does produce effective T stop differences, but I wouldn't complain about that-
For the purpose of testing transmission of lenses, perhaps it might be possible to cobble something together, perhaps even hand hold the camera under an effective dark cloth. Maybe a dedicated DSLR mount might even be overkill-
Jim Jones
1-Sep-2013, 07:29
. . . Because (as I've mentioned in this thread: http://www.largeformatphotography.info/forum/showthread.php?104407-How-are-the-aperture-scales-on-shutters-determined) you need a collimator, an optical bench, and a set of optical measuring instruments to get the real (not nominal) focal length and the entrance pupil diameter measured - at least if one insists on 0.1 of a stop accuracy... which I personally do not ever need in my own lenses BTW. :)
Quite true. Sometimes I get involved in the pursuit of perfection rather than in the practical production of photographs. Lens and shutter manufacturers have to do this. However, the functions of an optical bench and measuring instruments can be simulated well enough for most of us.
Powered by vBulletin® Version 4.2.5 Copyright © 2024 vBulletin Solutions Inc. All rights reserved.