Interesting observation on focus accuracy

[Ed Schwartzreic]

Two days ago I decided to recheck the focus accuracy of my R-D1, since it had been on many trips, plane, car, and shoulder, since DAG optimized the rangefinder. Indeed, I was glad I had the thought because the RF was again front focusing with all the lenses I checked. Out came the screwdrivers, flashlight, trfipod, meter stick, and Nikon DG-2 eyepiece magnifier. After I had got everything correct at 1 meter, including magnified jpegs to my satisfaction, I went outside to do the trimming at infinity, using the center screw adjustment. Here I was quite surprised that various lenses did not exactly line up the same. For example, when I adjusted infinity correctly for my 1st gen 35/1.4 and 3rd gen 50 Summicron, my 75/1.4 focused slightly but obviously beyond infinity, while my 90AA came in at a hair under. I decided to leave the setting as correct for the 35 and the 50.

Mulling this over, I first thought that this might be due to slight mis-matches between the lenses' cams and the R-D1's cam-follower wheel, as has been opined both here and on other sites. Then I had the idea of tring out the focus of these same lenses on a Leica body, in this case a .85 M6TTL with the 1.25 magnifier added. Once again I could see the same differences, in the same directions.

Conclusion? At only .72 on a Leica, likely none of this would have been visible. I think I have been tapping into sample variation at the limits of Leica's (or Leitz's) quality control. Your thoughts appreciated.

Ed

 

[Sonnar Brian]

I have had to build up the cams on two 135mm lenses using thin copper tape for this reason. Damn Cams. The Nikon and Contax RF's just do not have this problem. The Nikon RF uses an RF pickup wheel much like the Leica, except it rides the built-in focus helical. Lenses with their own helical couple to the built-in helical and rotate it for the correct distance. On the Leica, each lens has a cam that pushes on that focus wheel, and the cam has to be just right to provide accurate coupling. If it wears or is off by the slightest amount, the focus error is huge.

 

[jaapv]

I would say all these lenses have been designed for film use, as were the respective viewfinders. Any of the focussing differences you found would probably have been invisible due to the difference in thickness between films, the fact that different films can react differently to being pressed more or less flat by the film-pressure plate, etc. When you start measuring in the digital style, which involves a totally flat and fixed sensor and extreme enlargements on your monitor (I presume you viewed at 100%) it think is quite normal to see the variations in tolerance of manufacture. If it affects your photography you can correct it like Brian did. It will be interesting to see if Leica will have some solution for this in their digital M.

 

[jlw]

I think Ed wasn't just talking about variations in the on-film image, but about variations in the infinity position of the rangefinder itself.

I've noticed the same thing, and not just on the R-D 1: If I look through the RF at a distant object with three different lenses, one might show the moving image a bit short of the fixed image; on another, both images will line up perfectly; and on a third, the moving image might be a bit past the fixed image.

Since there's a lot of DOF at infinity, I guess it doesn't really make much difference -- although it's annoying, since many of us use infinity lineup as a quick check to see whether or not the rangefinder is correctly adjusted.

Meanwhile, an observation (although I admit I don't have many data points to support it) -- Canon lenses seem to line up better on Canons, Voigtlander lenses on Voigtlanders, etc., etc. If that's really the case, then some of this may be because of different manufacturers' assumptions about where the various parts will line up.

Anyone concur or dispute?

 

[Ed Schwartzreic]

Thanks for the responses. I do believe that what I am looking at is the machining of the RF cams on these lenses. On a 50.3.5 Elmar or a 28-35-50/4 Tri Elmar, it would be most unlikely to have a problem, but with long fast lenses, it is another story. Note that all these lenses are by Leica / Leitz. I once successfully sanded down the cam on a collapsible SM Summicron, at both near and far ends, because it was definitely "out" in these sectors. My 75/1.4 had in the past its cam replaced by Solms due to rather egregious errors; it is much better now, but clearly more "out" than the others lenses I mentioned. Since I don't have an R-D1 to dedicate to the 75 (by DAG and myself adjusting the RF parameters), since it is most often used in the close-to-midrange, and since it will focus accurately at 1 meter, it is usable.

I find this all both fascinating, and annoying. Heartily agree with Brian that Leitz chose a difficult method of lens-to-camera focus transfer. I think jaapv raises a good point about film thickness. Erwin wrote recently about Zeiss and Leica having different ideas about where to place the actual focal plane, behind or in front of the film pressure plate. With a digital sensor, however, the question ought to be moot.

Given jlw's thoughts and if I have the time and energy, I will run this test with all my lenses. Included ther, along with many Leica lenses, are a couple of SM Nikkors, several SM Canons, a couple of Voigtlanders, and a soon-to-arrive ZI 25/2.8

Attached are 2 shots with my 75/1.4, both wide open. The firs, hand-held, is of a lamp finial where I was probably 4 feet away. The focus is sharp yet shows the typical softness of the 75 Summilux in this application (and/or camera shake). The second is actually a "corrected" image. As oriiginally viewed through the RF, the fence posts were just short of infinity, but were out of focus in the image when shot at that setting. With the lens racked to infinity, focus is fine, as you can see.

Ed