Rangefinder focus depends on which direction you turn the focusing ring: why?

[onasj]

Advertisement (gone after registration)

I've noticed that the direction I turn the focusing ring on some lenses can have an effect on focusing accuracy for multiple M10 models (original, -P, -M, -R).  When I turn the focusing ring towards infinity on some lenses, I get back focusing, whereas if I turn the focusing ring of those lenses towards the minimum focusing distance, the focusing is perfect.  The effect is not large, but it is noticeable and can easily mess up a large-aperture portrait, for example (e.g., 0.5" of error at 1 meter on a 75-mm lens).

The direction of the focusing error is always the same: turning the focusing ring towards infinity causes more back focus in the photo.  It's as if my fingers are too slow to stop rotating the focusing ring once the rangefinder patch is properly aligned and I turn past the proper point... however, that's not the case, because if I rotate the focusing ring towards infinity and stop it visually, for example, at exactly 1.2 meters on the lens's distance scale, then keep my fingers away from the focusing ring and bring the rangefinder patch into perfect overlap by moving my body closer or farther from the target, the effect persists.  

That is, if I stop turning the focusing ring at exactly 1.2 meters on the lens's distance scale while turning towards infinity and get back focusing every time, then repeating the experiment except turning the focusing ring in the other direction, stopping at exactly 1.2 meters on the lens's distance scale, will cause the focusing to be perfect every time.  The lens "remembers" which direction the focusing ring was last turned, presumably because the mechanism (helicoid) that converts the focusing ring's position into a change in the depth of the lens cam that engages the rangefinder roller is direction-dependent.

The effect appears to be lens-dependent, not camera-dependent.  It is not dependent on when the lens was made—for example, both the older 75 summilux and the recent 75 noctilux show this effect.  It is not specific to my particular lens copy, because when I visit my local Leica store and try their copies of the same lenses I see the same effect.  I would guess that perhaps the lens's focusing cam position change lags a bit behind when the lens's focusing ring starts actually moving, except I think that would give the *opposite* result: instead of back focusing when turning the ring towards infinity and stopping at 1.2 m, the lens should front focus because the rangefinder mechanism would not experience the full change that occurs upon increasing the subject distance on the focusing ring.  

I suspect others have noticed this effect too, because I occasionally read posts here noting that their lens focuses differently turning the ring in one direction versus in the other.  But the question remains: what causes this direction-dependent focusing difference?

Edited August 11, 2020 by onasj

[Dao De Leitz]

Hello, compare it with the behaviour or response of mechanical backlash / hysteresis. (Links to wikipedia entries.) - Thorsten

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Bo-Sixten]

3 hours ago, onasj said:

I've noticed that the direction I turn the focusing ring on some lenses can have an effect on focusing accuracy for multiple M10 models (original, -P, -M, -R).  When I turn the focusing ring towards infinity on some lenses, I get back focusing, whereas if I turn the focusing ring of those lenses towards the minimum focusing distance, the focusing is perfect.

I've noted this as well (but in the opposite direction to what you describe). It has gone so far that I have sorted my lenses into groups which I focus differently (from infinity down or from close distance and up) depending on their profile. I always thought it was impossible to get all lenses to focus perfectly and RF precision. So I have adjusted my cameras to my most important lenses for each camera (M10R, M10M) and work from there. Most important is of course that your focusing method is consistent at all times and that you look straight into the viewfinder.

Edited August 11, 2020 by Bo-Sixten

[adan]

This kind of directional focus issue can come from two sources, in my experience. I am assuming you do not have an actual lens problem, since you say this occurs across several different lenses.

1) the connection between the lens and the camera cam is moderated by spring pressure (in the outward direction only), forcing the camera roller to stay in tight contact with the lens as the lens moves in and out. If that spring is weak, the roller may not track with the lens going out from the camera, but will track correctly when it is your finger pressure moving the lens in. The result is the lens focuses better focusing from nearest focus (lens extended) to infinity (lens at it shortest physical setting).

(Some individual lens types, like the earliest version of the 75 'lux, also use a spring-loaded cam in the lens, to push it tight against the camera roller. A weight-saving design in that a 3-4cm cylindrical threaded cam is replaced by a very short cam deep in the lens, connecting to the camera by a thin lightweight bar or rail 3-4cm long, with a return-spring. That spring can also weaken, or the travelling bar can get sticky with old grease and not react to the spring pressure, and fail to move in and out correctly. But that would show up as a problem with one specific lens sample.)

2) Psychologically, you are making a determination of when the images have "come together" for correct focus. But unless you have perfect eyesight, and/or a magnifying eyepiece, there can be a "dead zone" between where the images appear to align going left to right, and where they appear to align going right to left. And one will produce correct focus, and the other will not. The things you are aligning to focus on may just fuzzy enough or just small enough to be beyond what you can react to accurately.

Especially with a 75mm Summilux/Noctilux - they are about the most demanding lenses to focus in the M system, due to focal length and aperture.

In any event, either of those can result in better and more consistent focus results going out-to-in, or in-to-out. You can just experiment a bit to see what you can discover.

[Mikep996]

"In any event, either of those can result in better and more consistent focus results going out-to-in, or in-to-out. You can just experiment a bit to see what you can discover."

I had never thought about this but I have noticed that sometimes a few pics (same lens) are slightly soft.  I assumed I messed up the focus (finger wave and all) but maybe I didn't!  

 In thinking about it, it's no different than tuning a guitar. For stability you always tighten (raise) the string to pitch so the load on the gears is the same/constant.  If you go back/forth to tune, the tuning is not stable due to the play in the tuning machines/string tension.  Of course, camera lenses have more precise gearing than guitar tuning machines but some "play" still exists.

So, to avoid the issue when using the rangefinder, after reading Adan's post, it makes sense to me that it's more consistent accuracy-wise to focus in one direction if possible.  IOW, if you are set at infinity and need to focus on something that is, say, 3 meters, you would rack the focus to some point closer than 3 metres and then focus "out" to the subject.  OR, do the opposite if that's what the lens "prefers."

As Adan suggested, an experiment is in order!

[onasj]

Some more testing today revealed:

- The effect I reported in my original post is present in just about any Leica M lens I tested that has a narrow enough depth of focus to be able to detect the differences.  Affected lenses include the 50/0.95, the 75/2 APO, the 90/2 APO, the 75/1.4 (late model, Germany), and the 75/1.25.  So it's not a bug unique to any one lens model or lens copy.

- The direction of the error for me is always the same: the actual focus point in the photo is farther away if I reach a given distance (say 1.2 meters) by rotating the focusing ring towards the infinity stop, than if I reach the 1.2 meter mark by rotating the focusing ring towards the minimum distance stop.  The effect is very noticeable if I mount the camera on a tripod so I know there's no variation in camera-to-subject distance between tests.

- For all of my lenses, the focusing is pretty much spot-on when I focus by turning the ring towards the minimum distance stop.  For several of my lenses, if I focus turning the lens towards infinity, the resulting image can rarely be in sharp focus.

Assuming this hysteresis is normal and a fundamental aspect of the Leica M rangefinder mechanism, I suppose I am happy that all of these narrow-depth-of-focus lenses focus accurately in one direction, but it's still a source of missed photos, because sometimes you don't have the time to ensure that you reach the desired focusing point by approaching in only one direction.

[Good To Be Retired]

Advertisement (gone after registration)

8 hours ago, adan said:

This kind of directional focus issue can come from two sources, in my experience. I am assuming you do not have an actual lens problem, since you say this occurs across several different lenses.

1) the connection between the lens and the camera cam is moderated by spring pressure (in the outward direction only), forcing the camera roller to stay in tight contact with the lens as the lens moves in and out. If that spring is weak, the roller may not track with the lens going out from the camera, but will track correctly when it is your finger pressure moving the lens in. The result is the lens focuses better focusing from nearest focus (lens extended) to infinity (lens at it shortest physical setting).

(Some individual lens types, like the earliest version of the 75 'lux, also use a spring-loaded cam in the lens, to push it tight against the camera roller. A weight-saving design in that a 3-4cm cylindrical threaded cam is replaced by a very short cam deep in the lens, connecting to the camera by a thin lightweight bar or rail 3-4cm long, with a return-spring. That spring can also weaken, or the travelling bar can get sticky with old grease and not react to the spring pressure, and fail to move in and out correctly. But that would show up as a problem with one specific lens sample.)

2) Psychologically, you are making a determination of when the images have "come together" for correct focus. But unless you have perfect eyesight, and/or a magnifying eyepiece, there can be a "dead zone" between where the images appear to align going left to right, and where they appear to align going right to left. And one will produce correct focus, and the other will not. The things you are aligning to focus on may just fuzzy enough or just small enough to be beyond what you can react to accurately.

Especially with a 75mm Summilux/Noctilux - they are about the most demanding lenses to focus in the M system, due to focal length and aperture.

In any event, either of those can result in better and more consistent focus results going out-to-in, or in-to-out. You can just experiment a bit to see what you can discover.

This was a well written and informative post.  Thanx.

[adan]

1 hour ago, onasj said:

Assuming this hysteresis is normal and a fundamental aspect of the Leica M rangefinder mechanism, I suppose I am happy that all of these narrow-depth-of-focus lenses focus accurately in one direction, but it's still a source of missed photos, because sometimes you don't have the time to ensure that you reach the desired focusing point by approaching in only one direction.

Well, true.

By the mid-1960s, many (most?) photojournalists (LIFE, LOOK, AP) had quit using their Leica Ms (or other rangefinders) for anything longer than 50mm, and carried a Nikon F or Pentax SLR for longer lenses that had less depth of field and needed more critical "fast moment" focusing. 21/28/35/50 on the Leicas, and 85/105/135/200+ on the SLRs.

See vintage "at work" pix of the photographers, and enlarge where possible:

https://www.npr.org/sections/pictureshow/2013/01/27/170276058/an-iconic-life-image-you-must-see

https://alchetron.com/David-Douglas-Duncan#david-douglas-duncan-0cf55e76-1152-4d05-bf1c-e0f308e7e93-resize-750.jpeg

https://lavidaleica.com/content/jim-marshall-gets-grammy-posthumously

BTW - Jim Marshall, in that last link, famously said of his Leicas that once he found a lens and body that focused perfectly together, he never separated them again. Which accounts for needing 4 cameras for 4 lenses.

Now, as someone who would prefer not to carry a lot of cameras most of the time, I've just trained myself to account for the problem. By using smaller apertures where possible (or enforced, e.g. 135 f/4.0 Tele-Elmar, which is my most reliably-focusing longer lens.) Or the 90mm f/2.8 TE, which is also pretty reliable. But longer than 50mm at an aperture larger than f/2.8, I plan for possible trouble. It is not a rangefinder's strong point.

Others just carry their lenses preset at minimum focus - or infinity - so that focusing is always in the same direction that works best. Or use magnifiers or diopter corrective lenses. Or "bracket-focus" a little over 3-5 quick snaps. Or focus in the close range by setting the lens roughly, and then leaning in and out (instead of turning the focus ring) until the images merge. Or sink to the level of an EVF ).

[Good To Be Retired]

3 minutes ago, adan said:

Or sink to the level of an EVF ).

I like your attitude

[mmradman]

The EVF debate will not go away.

Recent Leica promotion of noctilux 75 and summilux 90 states best focusing result with SL and EVF.
Peter Karbe in his recent video states fast M lenses best used at max F stop.  Perhaps he knows something he is not telling.  Considering ageing eyes of typical M user either better OVF in development or M with built in EVF is on the drawing board.

[mmradman]

28 minutes ago, adan said:

Well, true.

By the mid-1960s, many (most?) photojournalists (LIFE, LOOK, AP) had quit using their Leica Ms (or other rangefinders) for anything longer than 50mm, and carried a Nikon F or Pentax SLR for longer lenses that had less depth of field and needed more critical "fast moment" focusing. 21/28/35/50 on the Leicas, and 85/105/135/200+ on the SLRs.

See vintage "at work" pix of the photographers, and enlarge where possible:

https://www.npr.org/sections/pictureshow/2013/01/27/170276058/an-iconic-life-image-you-must-see

https://alchetron.com/David-Douglas-Duncan#david-douglas-duncan-0cf55e76-1152-4d05-bf1c-e0f308e7e93-resize-750.jpeg

https://lavidaleica.com/content/jim-marshall-gets-grammy-posthumously

BTW - Jim Marshall, in that last link, famously said of his Leicas that once he found a lens and body that focused perfectly together, he never separated them again. Which accounts for needing 4 cameras for 4 lenses.

Now, as someone who would prefer not to carry a lot of cameras most of the time, I've just trained myself to account for the problem. By using smaller apertures where possible (or enforced, e.g. 135 f/4.0 Tele-Elmar, which is my most reliably-focusing longer lens.) Or the 90mm f/2.8 TE, which is also pretty reliable. But longer than 50mm at an aperture larger than f/2.8, I plan for possible trouble. It is not a rangefinder's strong point.

Others just carry their lenses preset at minimum focus - or infinity - so that focusing is always in the same direction that works best. Or use magnifiers or diopter corrective lenses. Or "bracket-focus" a little over 3-5 quick snaps. Or focus in the close range by setting the lens roughly, and then leaning in and out (instead of turning the focus ring) until the images merge. Or sink to the level of an EVF ).

No need to sink.  You made excellent case for M camera with EVF.

[onasj]

What's odd is that Leica's machining tolerances seem to be sufficient to support highly accurate and reproducible focusing.  If I turn the focusing ring in the favorable direction (towards the close distance stop), then even the 75/1.25 wide open, which has the shallowest depth of focus of any of my lenses, focuses reliably (90%+, comparable to state-of-the-art autofocus hit rates from Sony, Nikon, etc.), and the errors are not the fault of the camera or rangefinder mechanism, but instead are caused by a moving subject, moving photographer, etc.  

So it really seems to be a fundamental—and under-appreciated!—aspect of the rangefinder mechanism itself that there is noticeable directional hysteresis in focusing Leica M lenses.

It would be helpful if others could post their experiences.  The test can be done in minutes: turn your M lens and camera to a specific focusing distance on the lens (for example, the decimal dot in "1.2" meters).  Note the direction you rotated the focusing ring to get to 1.2 meters.  Place a target at about that distance, and adjust your camera's position as needed without changing the focusing ring until it is perfectly in focus by rangefinder.  Take a few shots.  Then rotate the focusing ring to achieve the same 1.2 meter distance, but from the opposite direction, so if you approached the 1.2 m position from infinity before, now approach 1.2 m from the close-distance end.  Move your body position without changing the lens's focus until the rangefinder is in focus and take a few shots.  Compare the two sets of shots and see if the actual focusing position has changed.

When I repeated the above experiment on a tripod, so the camera-to-subject distance was fixed, the result is interesting—it appears that the issue arises mostly from rangefinder hysteresis, and not from hysteresis of the lens's focusing cell position.  That is, the viewfinder patch is noticeably different when the lens is at 1.2 m from infinity versus 1.2 m from close distance, such that if I were to align the patch perfectly, I end up focusing at a farther distance when turning the focusing ring towards the infinity stop than when turning it in the opposite direction.  This result implies that the origin of the hysteresis is the transmission of the focusing ring position through the lens cam to the rangefinder roller to the rangefinder—something in that process is dependent on the focusing ring rotation direction.

[FlashGordonPhotography]

I remember this being discussed at length on the M9 forum, when that camera was current. I had exactly the same issue with a 135APO, Noctilux and a 90 Summicron (and others to a lesser extent) except I found I needed to focus them in opposite directions to get things right. There was even a proper name for it... *gear lash*?? rings a bell. Basically all the tolerances in the system (RF, focus ring, focus cam etc) added up and combined with our own tolerances led to the focus being ever so slightly behind or ahead of the RF patch in one direction. Like tuning a guitar it was advised to always focus from the same direction and return the focus to that end after each shot. It was also supposed to help with focus speed due to muscle memory and all that. I still tend to focus from the infinity side.

With the M240 I noticed it was less of an issue. And less again with the M10. My focusing with a Noctilux was much better on the M10. But I sold my Noct soon after getting the M10 and my 135APO hasn't been used in months. The lenses I use most with the RF. 50 Summilux, 28 'cron, 90 macro, 50 CV1.2. CV75mm helliar, don't seem to suffer as much and I've moved away from the larger lenses on the M system on a day to day basis. That's what the SL is for, now.

After I move, I'll have to pull out my 135 and 90 Summicron APO and see if I still get the same behavior.

Gordon

[onasj]

2 hours ago, FlashGordonPhotography said:

I remember this being discussed at length on the M9 forum, when that camera was current. I had exactly the same issue with a 135APO, Noctilux and a 90 Summicron (and others to a lesser extent) except I found I needed to focus them in opposite directions to get things right. There was even a proper name for it... *gear lash*?? rings a bell. Basically all the tolerances in the system (RF, focus ring, focus cam etc) added up and combined with our own tolerances led to the focus being ever so slightly behind or ahead of the RF patch in one direction. Like tuning a guitar it was advised to always focus from the same direction and return the focus to that end after each shot. It was also supposed to help with focus speed due to muscle memory and all that. I still tend to focus from the infinity side.

With the M240 I noticed it was less of an issue. And less again with the M10. My focusing with a Noctilux was much better on the M10. But I sold my Noct soon after getting the M10 and my 135APO hasn't been used in months. The lenses I use most with the RF. 50 Summilux, 28 'cron, 90 macro, 50 CV1.2. CV75mm helliar, don't seem to suffer as much and I've moved away from the larger lenses on the M system on a day to day basis. That's what the SL is for, now.

After I move, I'll have to pull out my 135 and 90 Summicron APO and see if I still get the same behavior.

Gordon

Thank you Gordon—very helpful.  I think our experiences point to the same directional preference: focusing is accurate for my lenses if I focus them FROM infinity TOWARDS close distance and stop when the viewfinder patch is aligned.  I'm tempted to get an SL2 to help with focusing, but I like my M bodies too much

[250swb]

As a student many years ago I was taught that after a shot or series of shots always to return the lens to infinity. Now this was using Nikons and the advice was so you always knew which way you needed to turn the focus ring when bringing the camera up to the eye. I've always used that advice whichever camera I'm using, but perhaps it has some additional significance with this thread?

 

[jdlaing]

Same here. Always parked on infinity.

[Jeff S]

Likewise.

Jeff

[Pyrogallol]

Tricky if you use both Leica and Nikon, they focus in opposite directions !

[oka]

This is something I discovered just recently.

For me the most problematic lens is Noctilux 75. For critical focus I need to focus turning from infinity.

I noticed that with Noctilux 50, I’m able to focus quite accurately in both ways BUT the results are always more consistent when focused from the infinity.

With summilux I can focus both ways accurately.

Then this issue combined with the RF drifting, I do think Leica has gone too far with the RF. RF is just not accurate enough for the Noct75 like lenses. Personally Noctilux 50 is the max I can comfortably focus.

In my opinion Leica will bring proper EVF to the M(11).

[Advertisement]

Hallo,

Look here: Leica M

[paulsydaus]

On 8/12/2020 at 7:41 AM, jdlaing said:

Same here. Always parked on infinity.

I’m always a tad off infinity which usually works well for the 5m to infinity range.

Either that or zone focused further off infinity if I know I’m not going to open the aperture too much.

At most apertures you don’t need to be sitting on the hard stop to have infinity in focus plus it gives you less distance to move.