Apo-Telyt-M 135/3.4 and gear lash

[Lindolfi]

Advertisement (gone after registration)

Today in a critical situation I made a series of exposures with the 135/3.4 Apo-Telyt-M at f/3.4 and discovered that some images were consistently front-focussed, while others were just perfect. Knowing that the 135/3.4 is the most difficult lens to focus on the M-system, I thought that could be the reason: just human error.

 

To make sure that it was no calibration or other systematic problem, I did a test at home with a camera with LiveView looking into the viewfinder of the M9 and focussing on the LCD screen of this camera. This appeared to be extremely accurate, since the overlap window of the M9 could fill the full LiveView window of the focus camera. Tiny movemements of the focus ring of the Leica lens could be detected. In this way I put a focus test device at 1.8 meter from the M9, mounted the 135/3.4 and focussed 5 times coming from infinity and 5 times coming from 1.5 meter.

 

The results were interesting:

 

135/3.4

 

From infinity the average focus error was 27 mm front focus with a standard deviation of 3 mm

From neardist the average focus error was 2 mm back focus with a standard deviation of 4 mm

 

So, this suggests gear lash in the lens or in the rangefinder mechanism or both. To exclude gear lash in the rangefinder mechanism, I used the Summilux 50/1.4 ASPH, focussed at 1.0 meter, using the same procedure. This gave

 

50/1.4

 

From infinity the average focus error was 5 mm front focus with a standard deviation of 0 mm

From neardist the average focus error was 0 mm back focus with a standard deviation of 3.5 mm

 

So the difference here is much smaller. The 50/1.4 ASPH also has a gear mechanism, but that mechanism only drives the floating element. The cam has no gear. The gear of the 135/3.4 needs to gear the motion of the lens down by a factor of about 2.5 to get to the correct motion of the tab pushing against the roller of the rangefinder mechanism.

 

And so this experiment shows that the (or more precisely my) 135/3.4 has gear lash. This possibility has been mentioned by some members earlier on this forum. This experiment supports it and shows how much it is: using a liberal measure of DOF, it is just outside of it, so usually no problem. With a more critical measure of DOF (like for 70 cm prints), it is the difference between in or out of focus.

 

The lesson to be learned is that in my case I should focus from near to get correct focus with the 135/3.4.

In general it makes sense to check your 135/3.4 to see in which direction it focusses correctly.

 

The lens certainly is worth it, because when it is in focus at full opening it is truly fantastic!

[russell]

I'm not sure I understood that entirely. But I think you're seeing something I also saw on my (previous) Noctilux about 10 years ago.

I would focus on a point at 3 meters. If I started from infinity to focus I it showed in focus at just over 3 meters. If I started focusing from 1 meter it ended up just shy of 3 meters. I did this over and over to prove the problem. It was there.

I sent the lens into Leica two or three times before they "got" what I was trying to explain to them --- it's not an easy problem to describe. Finally it came back that I'd need a new cam -- which cost almost the same as a new lens (at the time).

 

The problem was solved as the Noctilux was stolen in the Post on the way back from Leica!

[kokoshawnuff]

Some great info, thanks for the work. I have this lens but I've only used it on film and have noticed no focus issues. Are the tolerances different for gear lash when it comes to film and digital?

[adan]

Gear lash, quite possibly. But there is also another factor that can make focusing different depending on the direction: perceptual psychology.

 

When we focus with an RF, we bring the two images together visually, and as the RF magnification gets smaller (M3: 0.93x, M2-M7: 0.72x, M8/9: 0.68x, M6ttl "wide": 0.58x), or our eyesight gets worse - what constitutes "together" becomes problematic. There arises a "dead" zone where the focus ring can be moved a bit without the eye seeing the RF images move significantly, even if there is not "gear lash" as well.

 

I guess one could call that "eye-lash" - except it already has a different meaning.

 

As a corollary to this, it may be a function of one's personal visual psyche - whether "in-focus" means: when the images first appear to merge perfectly, or, on the other side, when they first begin to separate again. Either of which is NOT the perfect center point. Therefore, one can miss the focus point preferentially, depending on from which side one approaches the focus point - as you note.

 

With a 135, imperfect (but corrected) vision, and a lower-mag viewfinder (M9) - I've learned to fiddle the lens back and forth around the focus point, finding where the images come together AND where they separate again, and then try to nudge the ring back halfway between those points.

 

It is NOT the ideal way to use a rangefinder - with shorter lenses I just "snap" the two images together joyfully and quickly and DoF protects against both "eye-lash" and gear lash. Which is why (as mentioned on another thread) I've decided to give up 135s.

 

Leica had it right when they left the 135 framelines out of the M2.

 

And wrote in the M9 instruction book:

 

"• Limited use

Despite the high precision of the view and rangefinder on the Leica M9, exact focusing cannot be guaranteed when using 135mm lenses at full stop due to the very shallow depth of field. Stopping down by at least 2 stops is therefore recommended."

[Lindolfi]

Interesting Russell, since the Noctilux has no gear mechanism either.

 

Kokoshawnuff, the same with me: with film I did not find a problem. Digital unmasks errors more easily. One reason that the traditional measures of DOF now appear to be too liberal with the M9.

 

Andy, yes the human factor ("eye lash") is of importance, that's why I used the LiveView camera as a magnifier in this experiment. Doing so, you can turn the focus ring and see that there is a dead zone while doing it: the lens moves back and forth, but the image in the overlap window does not move. From now on I can get more consistenly focussed images (just using my eye at the viewfinder of the M9) with the 135/3.4 by coming from the near side and not used the midpoint method you mentioned. I already tried it. So in contrast to you, I do not give up on the 135/3.4. As I said, it's worth it.

[russell]

I should add that I was shooting using a M3.

At the time it soured me to the ultra-fast lenses -- just too sensitive to anything.

I had a bit of confidence restored with my 50mm pre-asph Summilux E46. It pretty much always nails the focus.

 

As sensors get faster I can see myself moving to the slower (smaller, sometimes less weight) lenses. My 50mm Summicron also nails the focus. And gives even more "slop" from DoF.

[luigi bertolotti]

Advertisement (gone after registration)

Time ago (when still used it) I had the feeling of a "dead zone" on my TE 135 2,8, whilst never experienced it on the TE 135 f4... but this doesn't mean that I always get focus spot on at f4... ... and I prefer to use it imaging it's a f5,6 lens.

[Michael Geschlecht]

Hello Bert,

 

From what you said in Post # 1:

 

135 mm lens front focussing 13mm.

 

50mm lens front focussing 3mm.

 

It appears your M9 body might be front focussing.

 

Or the body is fine & the 2 lenses are both front focussing proportionately more or less the same ammount they would if the body was out of alignment.

 

Assuming the front focussing, whatever the cause, was corrected: Would both lenses now be within tolerances such that DOF would be sufficient to provide properly focussed images for the sensor?

 

What if you used a properly calibrated later M3: .91/.68 = 1.338+ times the magnification?

 

Best Regards,

 

Michael

[jankap]

Focussing with a 135mm must be very exact. Yesterday I have made some trials with a Ricoh M-mount. Gear lash would be very bad.

The lever of the rangefinder in the camera moves back and forth from infinity to 1m for say 5mm.

The optics in the lens move back and forth according to the focal length.

So there must be a gear in the construction. For a 135mm this is a kind of an under drive.

If this is not working smooth, this could lead to lash.

I have a very old Canon 135/3.4, the heavy one. The rangefinder of the camera is actuated by a "stick" in the lens. Not by a cylinder, like most lenses have. Saves weight, but I suspect it.

With life view (or a Visoflex) it is possible to focus more exact, because the rangefindersystem is on its limit with the 135mm. An M3 has the best option with its finder of almost 1:1.

Jan

[Lindolfi]

Thanks for the responses.

 

Michael, the body is not frontfocussing and well calibrated. That 5 mm at 1 meter of the 50mm at 1.4 is within even tight measures of DOF. Also infinity looks good. The obvious thing is the gear lash with the 135/3.4 that only my 75/1.4 also shows, not the rest of my set of lenses. On average the 135/3.4 is frontfocussing, you are right, but coming from near it works OK. Naturally any help in magnifying the effective rangefinder base is welcome, but working with the M9 is what I do. Perhaps it would be a great idea to bring out a digital Leica M with a longer effective rangefinder base.

[tobey bilek]

Digital has us nudging the limit of what is possible.

 

It has been long ago shown that short wide angle lenses focus better on the rangefinder and tele on the the SLR. The crossover point is 90 mm where they about equal in accuracy. My solution is the visoflex with the screen calibrated to digital (shims).

 

Alternatively is a D700 Nikon with split image Katz Eye screen. Pro Nikons and the better models have a nice little screw to raise and lower the screen . Cheaper models resort to shims. But you can get the focus perfect.

 

Auto focus. Forget it. When I fine tune it, one setting will front focus and the next will back focus and I can not ever achieve perfection, this being on 2 pro bodys and some others. The only way to focus these with complete accuracy is with live view. However to be fair, AF is sufficiently accurate for most all my Nikon work. Visual focus with the bright screens is no good at all. Bright, but I can`t focus.

 

Even view cameras have issues. Use a loupe to focus on a nice contrasty screen and you think you got it. No the film bows just a little. This assumes you use a depth micrometer to calibrate the holders to the glass.

Sinar sold some fancy holder that keep film flatter if you can afford them.

[Michael Geschlecht]

Hello Bert,

 

Even if the ammount in millimeters is within tolerances: Wouldn't it be better if the mid point was such that the ammount of allowable error was more or less equal on either side of the point of focus?

 

1/3d of the discrepancy in the front & 2/3d's in the rear being ideal @ normal focussing distances.

 

Best Regards,

 

Michael

[Lindolfi]

Thanks Michael. I agree with that principle. But the reality of this is that reading out of a focus difference of 5mm at 1 meter with a 50/1.4 is hard. So an average focus error of 2.5 mm that I measured is not only within DOF, but also within reading accuracy of the error.

 

The DOF of a 50/1.4 at 1 meter is 32 mm with a Circle of Confusion of 0.03 mm and for digital it is perhaps half that number.

 

The 135/3.4 is certainly frontfocussung on average, you are right. But in this case I'm happy with that, since it does not front focus when coming from near. So in practice, given the gear lash, it's well adjusted. If it would be good on average (coming from near and far), I would get worse focussed images.

 

Concerning the 1/3 in front and 2/3 behind rule: that is only true for a single distance and certainly not at 1 meter with a 50 mm at f/1.4*

 

In that case the rule holds at 20 meter, while at 1 meter the DOF extends 49.2% in front and 50.8% behind the point of perfect focus. So let's call that equal in practice.

[cbarber]

This is a timely, interesting topic and I wish to share my recent experience. I purchased a new 135/3.4 APO-Telyt in March. It has been sent back to Leica twice due to its severe front focus issues and at certain distances impossible to focus even stopped down. The lens was returned to me a couple of days ago along with my M9 after proper calibration. I can vouch that this "gear lash" or whatever you call it is apparent on my 135mm lens. I was never aware of "gear lash" until I came upon this discussion and I don't have any problems focusing any of my other 5 Leica lenses including my 75 Cron and 90mm Elmarit wide open. But my new 135mm 3.4 APO Telyt is a whole new story.

 

Yesterday I went out and performed repeated focus tests at various distances at wide open and then gradually stopping down to f:11.The camera was on a tripod with a cable release attached. All focus targets were easy to focus on. My focus subjects were real world things, such as a stop sign down the street, etc. When the lens was stopped down to say f:5.6 or f:8 - f:11 everything was fine and that's great. But I was interested in wide open focus sharpness.

 

So yesterday I performed repeated tests, wide open aperture focusing at various distances (6' to 1 km away) first by turning the focus ring counter clockwise to get the subject snap into focus and then after several repeated focus attempts and exposures switched to turning the focus ring clockwise to nail focus. To make a long story short, in my case there was a big difference between focusing clockwise vs counter clockwise. In my results, except for infinity focusing (still have to work on that) my 135mm lens was spot on focus over 90% of the time wide open when I focused and stopped the focus ring turning counter clockwise - that's going from minimum distance to infinity. Virtually none of my tests by focusing clockwise (infinity to minimum distance) were in focus - they all front focused and sometimes by a great distance. I still have more testing to do but I must say I was really surprised by these results and happy to figure out how my lens behaves.

 

In the past I have always turned the focus ring on any lens back and forth to get the subject snap into focus. I never paid attention to what I was doing. But now with my 135mm 3.4 lens I will indeed make sure I focus counter clockwise to achieve focus especially when I am shooting at f:3.4 or f:4. Seems a little weird but it works for me.

[k-hawinkler]

Clive,

 

I am glad you found a procedure that works for you.

I'll stick with my TE 135/4.

 

Thanks.

[russell]

I think the lesson I learned is that once I find a lens that focuses for me I'd better stick with it than get GAS or worry about bringing the "wrong" focal length.

 

How would look this effect look with focus peaking? If someone runs a test, please let us know and post here...

[Lindolfi]

Thanks CBarber, for your report.

 

Good to know you can focus properly at full opening by coming from near focus now.

 

Russel, with focus "peaking", there should be no problem, since the focus is detemined with the same sensor as the one you expose on for the final image.

[cbarber]

Well I am glad I persevered with this lens because it has a gorgeous quality.

 

I think I have always unconsciously done my final focus nudge in a clockwise direction. Remembering to give the focus ring a final nudge in a counter clockwise direction (near to far) will become habit after a while.

[k-hawinkler]

Hi Bert,

 

I wonder which camera and adapter you used for the LiveView measurements?

How does this lens behave on an M240?

Thanks.

[Advertisement]

Hallo,

Look here: Leica M

[Lindolfi]

Sorry K-H, I'm still waiting for my Leica M (typ 240) here in the Netherlands, so I can't answer your question. Would be interesting to see what happens in LiveView and if you can see the gear lash in action (comparing RangeFinder with LiveView)