Rangefinder lens disassembled: how to calibrate focus?

[brbo]

But in the meantime, I’d really appreciate any insights into what I might be overlooking here.

One reason why a lens manufacturer would calibrate the lens not to be perfectly in focus wide open is focus-shift.

Basically, all faster lenses that are not perfectly corrected will have some focus shift. That means that focus will shift further back as you close the aperture. Most of the lenses made 50 years ago reached max. resolution only stoped down to f5.6-8.0 or even smaller. User knew that and would shoot images at those apertures when they wanted the best performance from their lenses. So it made sense for manufacturers to take focus-shift into account and calibrate them for those apertures. Especially true for rangefinder lenses where you don't have TTL view. But even with SLRs, the magnification in the viewfinder is much too small and much too dim to check for optimal resolution even if you stop down to shooting aperture.

Now, your Olympus RC f2.8 lens is not particularly fast and focus-shift will not be a big problem, but I would not obsess on getting it 110% correct in "theory". At some point you will want to shoot some real shots with you camera and that will involve some real film. But since you explicitly stated in your OP you don't want to check the calibration with developing some piece of film, it might take longer than necessary...

 

[OAPOli]

@OAPOli I highly doubt the 0.05mm offset is relevant. I've recalibrated focus on old cameras plenty of times without this added distance. It always ended up perfectly fine.

The 0.05mm is 100% observable with a collimator, so it has some relevance. For example, Minolta cameras require +0.02mm -0.00mm tolerance for the flange distance, and parallelism within 0.02mm.

In practice you're probably right. I'm sure the OP could use his DIY adjusted camera with good results. I'm trying to explain his observations.

 

[mariozelaschi]

I decided to follow @OAPOli's suggestions exactly and started my collimator from scratch.

I cut two pieces of plexyglass. On the smaller one, which rests on the inner guides, I applied a layer of frosted tape on one side and marked the tape with a focusing grid. Then I used the larger piece of plexy as a pressure plate. I also applied a small cutout so that there would be no hassle with the roll transport gear.

From the picture below you can see that the larger pressure plate does not rest on the outer rails since the small one mimicking the film is thick, but this is not a problem since the focusing surface is the frosted tape on the smaller side of the plexy, which mimics the film and rests properly on the inner rails.

I used my Konica with a 50 mm lens + lens hood as a test camera and ... tadan! The unmodified camera with factory seals was perfectly in focus! The split screen was perfectly aligned and the focus was sharp!
Now my collimator is accurate!
I then proceeded to focus the camera to adjust and now the clamping screw is perfectly in the same position as the screw mark that was on the metal plate!
I did it!

Later I will do a test outside on something far away to see if the rangefinder agrees (I haven't touched it but it may be misaligned with all this handling).


I have to say that now I am still puzzled on why the exposed film + pressure plate was not working, since it made much more sense to me..

 

[OAPOli]

@mariozelaschi I was suggesting to put the thin tape on the rails only, and the focusing grid on the plexiglass, so that the reference plane is slightly offset. Maybe it will make a difference?

 

[brbo]

I have to say that now I am still puzzled on why the exposed film + pressure plate was not working, since it made much more sense to me..

Well, it does make sense since you will be recording pictures on plates and not on film, right?

Seriously, the setup with film and plexiglass acting as pressure plate with no additional material on film rails (unless you intend to keep putting scotch tape there forever) is the best simulation for real life operation and the setup you should use to calibrate you camera with. I would still place focus slightly in front* when calibrating with the lens wide open (but since you obviously can't check the results on film quickly, you might just aim at the best focus wide open). Just because a setup shows perfect focus on a random 50-year-old camera does not mean that this is the proper way.


* I recently had to calibrate Minolta 24 Rapid and film tests gave better sharpness that way. The problem is that when you stop down the lens to sharpest aperture the dof increases and it is very hard to establish whether you are at optimal sharpness even if you have a mirrorless digital camera with high magnification. I used Olympus E-M5II which has 14x viewfinder magnification, but developing film and examining it under enlarger did show difference in sharpness.

 

[mariozelaschi]

@brbo Yes, I posted a photo a few posts earlier with my film + pressure plate setup to mimic the real situation as best as possible, and it ended up giving me a slightly out of focus infinity on both Olympus (the one I adjusted by matching the screw marks from before disassembly and the one that is still factory sealed), plus even my Konica TC + Hexanon 50 (which is perfectly in focus).

So this system for some reason is not reliable, at least the way I've done it.

Instead using a plexy plate worked and the factory sealed camera was perfectly in focus without touching anything.


@OAPOli I will try and post the results later

 

[brbo]

Instead using a plexy plate worked and the factory sealed camera was perfectly in focus without touching anything.

For some reason, you seem to insist on the "factory sealed camera" being the golden benchmark. Also you insist on not taking into account reasons why not perfect calibration wide open could be and in fact IS a thing (google "zeiss c sonnar 1,5/50 factory calibration" for the most obvious example).

Still, I think you are at least on track to calibrate the camera to mimic the "factory sealed camera" which might be all you ever wanted...

 

[mariozelaschi]

I do not insist on having my factory sealed Olympus as a golden benchmark.
I'm just saying that I am puzzled because with the film + plate system, even my Hexanon 50mm lens, which I've been using for years and is perfectly razor sharp in focus, appears out of focus.

 

[mariozelaschi]

I can confirm that what @OAPOli suggested works 100%.

I created a new focusing target with a piece of plexyglass with lines engraved using a cutter and put some tape on the borders where the inner rails are, and added some frosted tape on the back so that it's easier to spot the focus lines from the viewfinder.
I tested it on 2 lenses that I am sure of their focus (I have a small collection of Konica cameras and lenses) and the lines in the split mirror are just perfect!

One of my beauties (I changed the shrinked leatherette with velvet) <3

 

[baachitraka]

Nice. Can you put everything back now, so that you can shoot some Berlin street scenes during winter with your Olympus RC?

 

[OAPOli]

Just to reiterate, the correct tool for this calibration is an autocollimator with film in the camera. Film is 0.15mm thick, the space between rails is 0.20mm. So an offset of 0.05mm simply assumes that the film is flat against the pressure plate.

But even with that tool you can start splitting 0.05mm hairs. 1) the film isn't perfectly flat and this changes from frame to frame and over time; 2) the image from the lens has a curved field which varies with f/no and focusing distance. So the optimal focal plane is an average of all those factors.

When you're shooting handheld, stopped down, it probably doesn't matter too much.

 

[koraks]

But even with that tool you can start splitting 0.05mm hairs.

I'd like to add that the kind of tolerances you're referring to also bring into question the emulsion itself; the actual image-forming layer in the film is generally beneath a relatively thick (still tiny in absolute terms) protective top coat. The order of magnitude is in the same range as the 0.05mm hair you're trying to split.
Field curvature of the lens in a 1970s consumer camera is generally waaaaay beyond this magnitude, as are operator-originating focus errors (even more so).

 

[baachitraka]

I am curious whether OP have managed to put everything back into the camera and make it work again...