Live and learn. On my recent trip, I shot extensively on the Leica M Typ 240 using 90/100/180/280mm Leica APO R lenses with the Leica R-adapter M, the camera + lens being mounted using the tripod foot.
How about a week’s hard work irreparably damaged by shutter vibration?
For example, sharp at 1/250 @ f/2.8, 1/125 @ f/4, visibly blurred at 1/60 @ f/5.6 (peak performance aperture). And similar, depending on focal length and camera angle and so on. The danger range seems to be 1/8 to 1/125 second, varying depending on focal length and how the camera + lens are angled, which tripod is used, etc.
My working theory is that the closing/cocking of the shutter in the Leica M Typ 240 induces vibration that persists into the exposure (in conjunction with using the R-adapter M tripod foot and 90+ mm lenses). I do not yet consider it a proof, but I have no other explanation as to the results in numerous aperture series that I shot with the M240, including some that are a total loss at 280mm over a wide range of shutter speeds.
Since the weight of the lens really needs to be supported to avoid stress to the lens mount*, the 100mm f/2.8 APO, 180mm f/2.8 APO and 280mm f/4 APO lenses must use the tripod foot (or a bean bag, or something to support the lens). One can get away with mounting the camera itself for the 90mm f/2 APO (much more stable), but I did not do so; I used the tripod foot of the Leica R-adapter M. Bad idea.
It is not a question of build quality (the Leica R-adapter M is the best adapter I have ever used) and the super-tele Nikon and Canon lenses vibrate (oscillate) like crazy which one can see easily at 10X Live View. It doesn’t matter how heavy or massive the tripod is, though too-light tripods exacerbate the issue. The problem is the tripod foot and the length of the lens; vibration oscillates through the lens and this is true of most every design, the old Nikon 50-300mm f/4.5 ED being one notable well-built almost-exception. I covered and proved this years ago in Making Sharp Images at trivial megapixels, but it is far more of a concern in the 24-36 megapixel range. Serious long lens shooters take elaborate precautions (a tedious nuisance, but essential). However, Canon’s “Mode II” Live View feature is the ideal solution: zero vibration, making the image with the already-open shutter. But Leica does not offer this behavior in the M240, and it’s not clear that the sensor can even support it.
I’ve long known that the shutter itself can be a problem with long telephoto lenses on DSLRs (even with mirror lockup), but I had assumed that a rangefinder would be better off. Assumptions are never a good idea.
* Leica documents the advisory weight limits in the manual for the R-Adapter M.
Leica M Typ 240 Behavior
Behavior can be observed just by looking: set the exposure time to 1 second or so, remove the lens (but not the R-Adapter M), so that the shutter/sensor become visible, and observe:
1. The Leica M240 is in Live View mode.
2. Press release (2 second self timer). With regards to the shutter, nothing happens for the two-second duration.
3. 2 seconds have now elapsed. The open shutter slams shut, then reopens for the exposure.
4. Shutter closes at end of exposure, then reopens, restoring Live View.
The shutter cycling in step 3 is easily felt as a solid vibration into the hand if one grasps the camera. It appears that those ~6 micron pixels “feel” it too, resulting in a marked loss of sharpness at a shutter speed centered on 1/30 second (just about the same as the worst-case for DSLRs). The danger zone varies depending on the focal length and camera angle, the tripod mounting foot being the apparent weak point, just as with Nikon and Canon super-teles; the lens + camera length oscillates. It’s not that the tripod foot is weak so much as it is thin/narrow, and it allows the entire 'rig' to oscillate, amplifying even very small vibrations (easily seen by observing any long telephoto lens). With 6 micron pixels, even a small oscillation is a big deal, an amount that the eye cannot even see as camera movement.
The solution has to come from Leica: that shutter must NOT be cycled in this manner; a Canon “Mode II” solution is essential. Perhaps the shutter can be closed at the beginning of the self timer (half the cycle), or even closed and opened at the start of the self-timer. But why does it have to close/open at all? I don’t know.
[Aside: why is the self timer delay either 2 seconds or 12 seconds? 3/5/8 seconds would be useful, why such an arbitrary too-short or too-long?].
I was not aware of this subtle behavioral difference when shooting last week. It was foolish to assume that an exposure in Live View mode would be vibration-free.
In short: does exposure with the shutter closed/cocked to start with avoid the issue, e.g., if the shutter has only to open does that avoid the vibration blur? For the shutter to be closed/cocked and ready to open, both the EVF and LCD needs to be dark (off), e.g., classic rangefinder mode. But this is not an obvious thing.
Testing the M240 behavior, I found that with menu Light Metering Mode = Advanced the shutter is *always* open, even if both the rear LCD or EVF are dark/off. That prompts the close/recock cycle. To avoid this, set the M240 to Light Metering Mode = Classic; in that case the camera operates in a shutter-closed rangefinder mode provided that both the LCD and EVF are dark/off. This seems to be related to a sort of dual shutter internally, one a true shutter, the other a black/gray/light gray metering curtain(this can be seen by switching modes and observing the camera with the R-adapter M mounted, but no lens on the camera).
Leica speaks to this in an obscure way (pg 181 of M240 manual), but not to the implications as to vibration and/or the self timer:
For the metering methods based on the image sensor, the shutter must be open and it is then closed and re-cocked when the function is cancelled. Of course, this is audible and may result in a slight delay in the shutter release.
I was shooting in manual exposure mode, so no metering is needed or wanted. So the bad behavior ought to be moot; i.e., the camera ought to do the thing that minimizes vibration, especially with manual exposure in self timer mode!
In shutter-closed rangefinder mode, an exposure does not need to close and cock the shutter first; it only has to open the shutter for the exposure. While opening the shutter is not vibration-free, surely opening only is better than a close/cock/open cycle. Still, holding the camera, my hand can easily feel a solid click of the shutter opening (set to a 1-sec exposure, there is no confusing opening from closing).
This possible mitigation strategy means that one has to program Light Metering Mode = Classic and in addition exit LCD/EVF viewing for each and every exposure (LCD and EVF go dark/off). But is this sufficient to avoid the vibration issue? TBD. At any rate, it is an undesirable operational requirement, demanding a lot of the operator. One wonders if the new Sony A7R will be able to take a picture without shutter-cycling prior to the exposure.
Tell me how/why I am wrong!
I have compelling evidence that my theory is correct. Meaning I have days of careful effort in shambles from vibration-induced sharpness damage, which largely trashes it in terms of showing optical performance. It’s a serious setback in my plans to show the performance of the four Leica R APO lenses, though the lemons can produce some lemonade in showing the issue itself.
If readers think I am in error, tell me what I am missing. Really, I want to be wrong, and I want a simple 'fix'! It won’t fix my damaged images, but maybe there is something simpler than the Light Metering Mode = Classic mitigation strategy discussed above.
I have sent a link to this blog post to Leica technical staff, and I hope to hear something. The most egregious algorithmic error is that the camera does not immediately close/cock the shutter as soon as the self timer is started: this can be a software change that at least would avoid the close/cock part of the exposure cycle, see the discussion above. At the least, in manual exposure mode there is no rational excuse to not do so. In short, this is an error in camera logic.
Note also that the nature of the sensor might not allow a Canon-style vibration free “Mode II” exposure to be made. But if it does, then this is the one and only true solution.
It seems that Light Metering Mode = Classic as discussed above might be of real value—to be confirmed.
Herb S writes (excerpts):
Your post about shutter vibration using longer adapted lenses on the Leica M was very interesting. Something I would never have expected myself for a rangefinder camera.
I decided to test it as well to see if I could repeat your experience. Well to go straight to the point: I got (almost) the same results.
I shot a banknote at 4 meters distance from tripod and self timer. Leica M with Novoflex LEM/NIK NT adapter and Novoflex ASTAT MFT bracket with slight modification (see picture) and Zeiss ZF.2 135 mm.
Differences due to vibration are clearly visible at 100% view with exposures from 1/60s to 1/8s. Classic mode is always superior here and not showing real vibration in itself.
At 1/250 s and 1/125 sec I would even say the Advanced mode seems to perform a fraction better. So this is something different and quite unexpected. Vibrations can come and go in unexpected ways.
At 1/4 s I see no difference but this is possibly due to lower contrast and higher diffraction at f/16.
At 1/30 s the vibration of the Advanced mode is most clearly to see, but this might be also related to the best performance at f/5.6 of the Zeiss lens.
In short I can confirm your findings that vibration can show up in Advanced mode at shutter speeds from 1/60 s to 1/8 s with longer adapted lenses on a bracket / tripod.
DIGLLOYD: Makes perfect sense to me. As I noted in the field, there is some variance, and this is to be expected from the lens used, the angle of the camera (e.g. pointed up/down/level), and the resonance frequency interacting with the rig.
The somewhat good news (which I need to independently confirm) is that Light Metering Mode = Classic as discussed above apparently is a big improvement, though it might not be a cure.
Peter W writes:
Nice piece, and I do not think you are wrong. The general phenomenon with long lens vibration explains why VRII is important in Nikon's big glass, which I had not really realized before… I thought VR was largely a marketing gimmick aimed at amateur customers.
I think you have uncovered a design issue that Leica must have missed, and certainly the beta-testers of the M240 did not have R adapters to make the comparisons you have completed now. Your point is that until Leica offers a FW solution to use the self-timer as a means to allow for the dissipation of shutter-cocking vibration, then the M240 is inherently limited in use with long lenses.
Jef M writes:
Without knowing the force of a small fast moving shutter it's hard to make a prediction on whether or not it would be an issue for the combination of a number of factors including the weight of the system, the direction of the primary force and any counter force is applied in, the balance and damping factors of the mounting system, the leverage of the weight and length of the lens. As well maybe wind and minor earth vibrations could have been contributing or cumulative factors in your results.
The root cause fault analysis would be to experiment in a controlled setting and remove as many external factors as possible.
DIGLLOYD: All true, but root cause is of academic interest beyond the mitigation strategies it might suggest, so only the practical solution in the field in general is of interest to me.