Saturday, March 29, 2014

The Pitfall of Ultra Precise Lens Adapter

Back in the days when lens adapters were uncommon, they were well made and precise, but expensive. Today, most cheap adapters you can buy from eBay are mass produced in China; quality is acceptable, but precision is usually not a huge consideration.  In fact, nearly all of these adapters for mirrorless cameras allow the lens to focus past infinity, and this is intentional.  In the beginning, many users found that the adapters they bought could not focus their lenses to infinity, and they returned the adapters. Manufacturers then started to make the adapters slightly thinner than the official mount measurement; this would allow the majority of lenses to achieve infinity focus, but usually slightly pass it.

To me, this is acceptable.  After decades of use, most old lenses would deviate from the original specs and they may focus closer or further than they were designed to do. If you make an adapter that’s perfect, some lenses will be blurry when the lens is focused to infinity. All Canon auto focus lenses that use UltraSonic Motor (USM), allow the lens to focus pass infinity to compensate for parameters that could affect infinity focus, such as temperature variations and manufacturing tolerance.  But, manual focus lenses do not have this luxury.

I have two relatively expensive adapters; one Canon FD to M4/3 adapter made in Poland, bought right after the G1 came out, and the other, Leica-R to Canon EOS adapter, made by Elefoto of Japan.  For whatever reason, none of my Leica-R lenses would focus to infinity with the Elefoto adapter.  I tried the 28mm f2.8 Elmarit, 35mm, 50mm, and 90mm f2 Summicrons, and two 90mm f2.8 Elmarits.  I don’t know what lenses they used as a reference when the adapter was designed, or perhaps there is a defect in manufacturing.  The Polish adapter, on the other hand, worked perfectly with most of my FD lenses on the Panasonic G1, and when lens is set to infinity, everything is sharp and clear.  I started using this adapter on the E-M5 lately and so far, found that two of the lenses are just shy of infinity focus: FD 50mm f1.2 S.S.C, and FD 35mm f2 S.S.C.  If the aperture is set to f8 or smaller, sharpness becomes acceptable, but still not quite there.  A mere millimeter could mean in focus, or not.

For this reason, I usually just buy the cheap adapters, knowing that they would at least give me infinity focus.  The only thing that worries me is the flatness of the adapter, which could cause de-centring issues resulting in one side of the picture to be sharper than the other, if tolerance not kept in check.

Smoke Break - Canon FD 200mm f4 & Olympus E-M5.


  1. I would expect that the flatness of the adapter would only affect resolution on a 2D plane, I doubt you would notice it with a 3D image since the effect would be akin to the very slightest shifting of the plane of focus with a TS lens.

    1. When the picture is shot at infinity, and if the adapter is not flat, you can see the resolution drops on different parts of the picture.

    2. have you worked out which way it tilts? Do you need to stop down much to 'fix' it? Knowing that will allow estimation of how far out it is.

      Having pulled apart an FD before the base of the mount is not a very precise design ;-)

    3. It depends on adapter. The ones I made myself were the worst! The FD design is way more complicated than necessary.

  2. Hi, its much much less than a whole mm ... to prove this to yourself go get your 35mm lens (or better yet a 28) and observe how far the lens extends when you change focus from the "lock stop" to even 5 meters away. Its between tenths and hundredths of a mm. The experience you had with other lenses on the adapter is quite likely caused by even lens mount inaccuracy. When you consider that in engineering only a fully machined part has such tolerances, adding an assembly and moving surfaces will make for only less accuracy. I believe that you'll find that FD lenses have some more variance in this area than say OM lenses (due in part to the nature of the mount).

    This was not seen as a problem back in the day because 1) few people enlarged sufficiently to identify this in their images 2) fewer people understood this issue 3) film flatness provided bigger variances than we experience today with sensors and (importantly) 4) stopping down increases DoF so as to make the issue invisible.

    I know that in the micro43 area that lenses 'calibrate' when turned on, to sort out fine tuning. So its likely that there is an allowed tolerance between flange and sensor which may be enough to make a difference "camera to camera" with the same lens and adaptor.

    I happen to like the polish guys adapters because they are the only ones I've found yet which are machined from a single bit of metal. Thus they are the strongest and the most accurate. As soon as you have 2 parts screwed together you'll introduce more errors.

    As this is related to format size (by maths) its even more an issue with the winkydink tiny Nikon sensor. Your APS camera will allow slightly greater tolerances than your OM-D :-)

    But as Mike Says, the inaccuracy will can only be an error between parallel-ness of the surfaces, unless you've got outright wiggle. So its like a fixed version of a TS-E lens. The adapter I used for my OM lens in my post ( uses a rotating ball (it is this one and so by design I'll never be assured of perfect parallel alignment. When photographing things with it set straight "by eye" I really have never observed its out. But then I don't photograph the horizon with a 28mm lens at f2.8


    1. The infinity focus issue is too easy to happen if the adapter is created with exact official parameters. One really need to take into account the lens variance, especially for old lenses.

  3. PS: couldn't help myself (sorry) plugging in paramaters into my calculator of 35mm lens and assuming infinity as > 97meters an error in the adapter of making it 0.1mm too long will bring your focus point back to 10 meters and if you were using a 21mm lens 0.1mm would bring you back to focusing at 4.27 meters and even an error of 0.05mm in dimensions would mean that the lens is focused at 8meters (as of course it is more exaggerated on shorter focal lengths)

    As 14mm is a common focal length for micr43 its no wonder native lenses auto-calibrate ;-)