Focus Calibrations for (Pentax) (D)SLR Bodies and Lenses

From time to time, there exist DSLR users report about different AF issues, regardless of brand. Nevertheless, the fact is that some particular DSLR or lens models are being reported by more users than the others. It is just because either that those particular models are indeed less accurate or simply that they are more popular (so that there are more users to report, by proportion) or both.

For example, for those reports on AF issues about the K10D, you can refer to my previous Blog entry:-

http://ricehigh.blogspot.com/2006/12/af-errorsproblems-of-k10d.html

In this article, I shall talk more in-depth about the essentials of focus calibration for (D)SLR bodies and lenses, by first discussing into the working principles and design of those equipment, in general sense, and then particularly about Pentax gear on the possibilities of how to get those stuff correctly adjusted, totally or maybe just partially.

Actually, a camera is a precision tool which means that alignments and all those distances within it (which means both the body plus the lens for what is meant for a camera) must be *calibrated* to a high precision with adequate accuracy. In order to achieve this requirement, first of all, the design of the focusing system by itself, no matter it is an AF (Auto Focusing) or MF (Manual Focusing) one, must be accurate *enough* by itself. And then, good calibration procedures are required to ensure the accuracy by making proper precise alignments, for all those adjustable variables in a design. Well, the (process of) correct adjustment of those variable physical parameters is the "calibration".

So, one will ask how much precision is required for a focusing system? Or, what is an “adequate” AF system? Last time, I have some talks about the possible precision requirements about the design of an AF system (and the importance as well), you can have a look on this first:-

http://ricehigh.blogspot.com/2006/12/how-af-errors-can-decrease-effective.html.

By the way, I notice that many people are often confused about different phenomenons of the focusing problems seen and the respective causes underneath.

Indeed, different focusing problems are often caused by a combination of an inferior design (with inadequate accuracy – just because manufacturers are cutting corners, but they are just doing that too much, far beyond the limits), poor alignment(s) and calibration (which is also often the case). Or even worse, the problems are caused by loose and poor QC at the factory or maybe just the alignment department in the production line are doing a bad job in the first place (well this is somehow closely related to the proper design of the manufacturing process *plus* the “quality” of the factory workers on how they are faithfully and accurately doing their jobs as required).

Well, in the following, I shall summarise the key major technical aspects of the (AF) SLR focusing system, their significancy, and the possible adjustements allowed for Pentax DSLRs:-

I. Accuracy in the Back-focus Register Distance of the Body

Note that the term “back-focus” used here is by no means related to the Back Focusing (BF) problems reported by the DSLR users daily. That is actually about the distance measured from lens mount to the focal plane, which this distance is the basis of how the lens system should be designed as it is the fundamental parameter for all the optical calculations involved in any optical formula thereafter.

Do note the back-focus distance is generally referred to the distance from the front of the lens mount to the focal film/sensor plane where the sharp in-focus image should ultimately arrive. However, the exact point of reference for the measurement does not really matter as long as the thickness of the body mount is known and then everything is relative (in displacement).

Here is an interesting article for a case study about the back-focus register distance of the Hexar Range-Finders (RFs) and the assoicated implications:-

http://www.dantestella.com/technical/flange.html

It would be interesting to note that the choice of the back-focus distance is dependent on the film/sensor size as well as the optimal focal length chosen as “standard” (for the “standard” lens, with a standard field of view). For example, the back focus distance of 35mm cameras are optimised for standard lenses in the range of around 40mm to 60mm. Beyond that, lenses are more difficult to be designed and built as optical formula will become more complicated. Of course, it’s normal that a longer focal length lens will be more bulky and it’s the Physics.

Afterall, if the lens is designed to assume for a certain back-focus register distance but unfortunately a body is made to be having a wrong one, either shorter or longer, then irregularity will result. Just say if the back-focus distance is lengthen, the camera may not be able to focus sharply at infinity (remember human short-sighting owing to long eye balls?).

So, what can be adjusted for Pentax DSLR Body?

I’m afraid that the answer to this question is “virtually NONE”, from my experience! My Pentax *ist DS was having a slightly longer back focus register distance so that many of my Pentax lenses (mostly primes), which are correctly collimated for the infinity, could not achieve focus at infinity. My local Pentax service centre had adjusted my *ist DS body to make it FF a little bit so as to “resolve” the problem. But after that I discovered my *ist DS tends too much to FF, so my unit was finally sent back to Pentax Japan for repair, but then Pentax Japan opted to adjust the AF mirror angle to “rectify” the error, but then the whole AF sensitivity area was shifted out of the marked area (despite that the original problem was solved). For more details, you can (always) read this:-

http://www.geocities.com/ricehigh/A_Letter_to_the_CEO_of_Pentax_Corporation.html

BTW, this physically wrong back-focus distance can cause troubles to the AF system if the this is one of the parameters involved in the *cor-relation* between the (sub-optical path of the) AF system and the actual optical path for the exposure, which I would explain more in the next part below.

II. Accuracy of an AF system (for its design, manufacturing precision and alignment)

1. "Range-finders"

Starting to elaborate more about the case, one can think about the focusing mechanism used in those RFs in the good old days (when mis-focusing cases rarely happened). The rangefinders have separate optical focusing system in the viewfinder, so precise alignment for an accurate “cor-relation” is crucial. Same happens for AF SLRs where the AF system is a separate system optically (putting aside the electronics sensing element) and it has its own optical path (and sub-paths as well, that will explain more below).

Strictly speaking, as long as the focusing department of a camera is concerned, AF SLRs can be considered as range-finders but not true SLRs. The only difference is only that the human eyes for doing the focusing job, ultimately, are replaced by the AF sensors (CCD or CMOS, which doesn’t matter anyway).

2. Why MF SLRs are usually more accurate?

Actually, the manual focusing principal of MF SLRs (also AF SLRs in MF mode) by utilising the focusing screen can be easier to achieve a higher level of accuracy. It is not because of the technology is proven, but just because of the design is much simpler, and involves only optical device instead of optical devices plus lectro-optical electronics, and, most importantly, the optical path is simpler, too.

The critical alignment of MF system rely mostly on the up/down position and parallelity of the focusing screen (plane) , which should precisely lay on the reflected focus point, by the main mirror of the SLR.

For more read on how the focusing screen can focus accurately with the aid of split prisms and/or microprisms and how these work, see this super clear technical article (one of the best technical articles I have ever seen on the net about cameras):-

http://doug.kerr.home.att.net/pumpkin/Split_Prism.pdf

3. Why AF SLRs can be even more inaccurate?

As you can see from the above technical article also that SLR AF system is rather complicated in its optical structure and its components are made in a very tiny way. So, a very high precision is required for making each of all these.

Besides the difficulty in making proper alignments for an AF system owing to its complexity, there are even more things to affect its accuracy. These include the optics for projecting the image to the AF sensors (which are not perfect, optically), the sensitivity and limitations of those sensors, analog and digital signal processng of the AF siganals including the software algorithms used in the phase-matching/detection and etc.

BTW, here is a detailed bibliography showing the technical details of a typical (Pentax) SLR AF system:-
http://v3.espacenet.com/textdoc?DB=EPODOC&IDX=JP2004157212&F=0&QPN=JP2004157212

By looking at the side cut view of the AF module, one can have an idea how the physical path of projected image is, starting from the lens and then the light ultimately reaches the AF sensing module.

4. The Latest Pentax Design (of the SAFOX VIII)

Here is the very recent filed patent for the Pentax AF system (which probably is the central 9 (cross sensor) points of the SAFOX VIII):-

http://v3.espacenet.com/textdoc?DB=EPODOC&IDX=US2006120711&F=0&QPN=US2006120711

It can be seen from the drawings and the descriptions that the design of the (SAFOX VIII) AF system is very complicated. However, technically it is basically of not much different from other systems and it is exactly the same of the Canon 20D’s one, for its design andworking principle, from what are described the last technical article I have pointed to.

As briefly mentioned just before, for all those numerous but tiny AF optical elements involved, e.g. those micro lenses, micro-prisms and/or mirrors. Each of those elements will undoubtedly introduces certain optical and alignment errors by themselves as nothing is perfect in this world. So, they can add up to a much larger error, especially those errors can be greatly magnified owing to the small size in the detection system and hence the components in the AF module.

All in all, the simplest focusing system will be the best in terms of accuracy. For the design of an AF system, I would rather ask for quality than quantity. That says I prefer a single high grade and high precision and well calibrated/made linear AF area in the AF detection module rather than numerous low grade sensors placing around which I would seldom use (except I give my DSLR as a P&S to some laymen to hold my camera and shoot). All those saved material, manufacturing and labour cost for such a greatly simplified design can be re-used for making a much better AF system, frankly speaking, e.g., a more sensitive AF sensor, an AF sensor with higher pixel density (so that the AF calculation will be more precise and the minimum f-stop for AF to be workable can be made smaller – just look back to the last focusing system technical article above if you don’t get exactly what I mean here), more precise micro-optics with more corrections in optical aberrations (which are all unlikely to be calibrated, but are only dependent on the precision of the design and the “mould”) as well as spending more labour costs in ensuring the “outside” alignments, which are adjustable, and etc.

Nonetheless, I have verified in controlled environment that my MZ cameras (MZ-S and MZ-30) or even the 1987 SFX are more consistent in AF results (i.e. the final AF points are usually close enough no matter how the user does the AF, say, AF starts from infinity mark or nearest of the AF lens) and that different lighting conditions such as the color temperature of the light source, lighting level, etc would cause minimal errors. In fact, the texture and pattern, e.g. color and spatial frequency of the object affects the reliability of the AF too. Here, I know that the best AF system is still the central sensor used in the MZ-S, although it is not a crossed one!

The SAFOX VIII used in my 3 models of Pentax DSLRs are not something that I can really rely on as I used to with my MZ-S. Actually, I do have seen improvement / debugging of the SAFOX VIII with my K100D than my old *ist Dx. However, the AF system of my K100D still behaves somehow inconsistently and the yellowish tungsten light can fool it easily and in a large amount.

To do simple experiments to verify, any Pentaxian can mount their own DSLR with a fast prime on a tripod, does AF on an object, focusing from infinity and then from closest and repeat for a few times. To compare, repeat the test with different camera models (if available) and/or under different light sources or even for different objects (with different colors, spatial frequencies and contrast levels, etc.). Then, you’ll see the AF system of those MZ cameras are undoubtedly more reliable and accurate, although marketing wise they are not as “sophisticated” as the latest SAFOX VIII, at least on published papers!

The yellow light symptom of Pentax DSLRs is a great annoyance which means that the effective resolution of the DSLRs are greatly reduced owing to the significant (front) focusing errors under tungsten. Again, here is my old quick experiment to confirm the problem:-

http://www.geocities.com/ricehigh/DS_AF_Dependency_on_KEV.html

5. So, what can be adjusted for the Pentax DSLR bodies?

Someone told that the plane of AF module is seated on a plane that is adjustable, see the post with diagrams/photos showing the “secrets”, below:-

http://forums.dpreview.com/forums/read.asp?forum=1036&message=19960398

After we see about the working principle of any focusing system in the above, I think now that we know about this is exactly to adjust the plane of “perfect focus” for the AF module. Well, I have checked against the relevant Pentax patent to further verify this::-

http://v3.espacenet.com/textclam?DB=EPODOC&IDX=JP2004157212&F=0&QPN=JP2004157212

Furthermore, I know that there is a "secret" port underneath the bottom plates of Pentax DSLRs. A physical connection can be established via the port and then a Pentax servicing utility can be loaded. This utility can do some additional calibrations software-wise including the input of the FF or BF offset amount.

III. Lens Collimation

“Collimation is the adjustment of a given lens to a camera body, so when on that particular camera body the lens will provide correct focus to infinity as well as to the lens scale settings”:-

http://lavender.fortunecity.com/lavender/569/lenscollimation.html

Well, again it is NOT something to do with those FF and BF issues, which are all about the inadequacy, inaccuracy and off-calibration of an AF system (discussed in the last part).

In fact, the most significant part of the lens collimation is at/for the infinity position, of which the collimation calibration should get the position exactly right. It is the most important because the near distance and middle distance can allow for some deviations which will not prevent the user from getting the right focus for the shooting, it’s just that the distance scale does not tell the exactly right distance for what the scale shows. But at the infinity, there might be no more room for the body/lens combo such that the lens can never focus/reach infinity. To cater for this problem, most current cheaper lenses will allow more room beyond the infinity but yet the camera makers do calibrate at the infinity mark at the focus scale. As such, a looser calibration & QC or simply the manufacturing tolerance can be allowed. In the past, they did this only for super telephoto lenses, which the collimation position of the lens is sensitive to change, e.g., with temperature.

So, what can be adjusted for Pentax *Lenses*?

As most of other lenses on this planet, I believe all Pentax lenses can be adjusted for the collimation. Here is a set of procedures extracted from the service manual of the Pentax FA 43/1.9 Limited lenses:-

http://forums.dpreview.com/forums/read.asp?forum=1036&message=15072863

The missing diagrams are exact scans from the Pentax maintenance manual, which has already been deleted by the poster, who knows very very well about the technical insides of Pentax gear. Indeed, I have learnt so many things from this knowledgeable guy in the past.

Here is the quote from the post for the procedures:-

“First you need a piece of rubber to unscrew the large retaining ring in front, then 3 screws underneath.

Peeling inside and you will see 3 black colour screws with washers (see pic below). Loosen them a little (don't remove them completely) then you may adjust the focus. When done, tighten them again, and apply light duty threadlock on all screws. It is important to remember never overtighten them because the aluminium alloy is rather soft.”

Just say if you have a body with wrong back-focus distance, physically, one can indeed do the collimation for all the lenses he had that will be mounted on the body. However, the focus collimation of such adjusted lenses will not be exactly correct for another normal body which has the correct collimation (back focus distance), so it is not wise to do so afterall.