Canon A-1 back focuses for close subjects only

[hartacus]

I've had to sideline my Canon A-1 due to a few rolls of film that have demonstrated a back focusing issue, especially with close subjects in low light where the FD 50mm f/1.4 lens has a paper-thin depth of field. Today I had a bit of time so I set about attempting to fix it. I suspected that the mirror angle might be incorrect and the mirror stop might need adjusting. However, using another SLR with known-good focus as a collimator showed that infinity focus was spot on, and I confirmed this using a distant (>300m away) object. I was surprised by this - checked it several times and I'm fairly confident of that result. I guess I hadn't noticed it during use because I don't come across many situations where I'm taking photos of distant objects with the lens wide open.

So I tested a much closer subject, at or near the lens minimum focus distance. Initially with the FD 50mm f/1.4 SSC lens. The subject was a piece of film, cut between two sprocket holes, placed at around 70 degrees to the film plane angle so that when focused (using the viewfinder) on the leading edge of the piece there'd be a relatively clear visual indicator of whether the focus (on the film plane) was behind the leading edge. Focus at the film plane appeared to be at around the second sprocket hole, not the leading edge. I re-tested with an FL 50mm f/1.8 and got a very similar result, which leads me to think that it's not the lens.

What could cause an SLR to have accurate focus at infinity but back focus for close subjects?

(Note that I re-seated the focusing screen not long ago, taking care to insert it in the correct orientation as it had been replaced backwards by a previous owner.)

 

[ic-racer]

https://photographylife.com/what-is-focus-shift

 

[shutterfinger]

I confirmed this using a distant (>300m away) object.

300 meters = 984 feet. A good infinity target should be at least 5000 feet / 1524 meters away.
Select a further infinity target and check again. Celestial objects make excellent infinity targets.
With bellows focusing cameras the difference between a 1000 foot infinity target and a 5000 foot infinity target is .001 inch to .002 inch of bellows extension. .001 inch = .025 millimeters.

 

[John Koehrer]

SF & I tend to disagree on this subject. He likes celestial thingies & I like you, deal with more earthbound realities.
He once sent me a link to a DOF scale to demonstrate how a lens focused on "X" would be in focus from ~35' to over a hundred(been a while)
to which I thought. Really? Doesn't that cover photographic infinity?

 

[shutterfinger]

A DoF calculator will show 57 meters to infinity to be in focus with the focused distance at 300 meters, 67 meters to infinity in focus with a focused distance of 1500 meters.
In actual application a object 1,000,000 meters away will be slightly sharper with the lens focus set to 1500 meters than it will be with the lens focus set to 300 meters. To many this minute difference does not matter.

 

[hartacus]

Thanks for your replies, all!
Fairly confident that it's not focus shift, as the lens wasn't stopped down at any point in all this. The photos that show the most back focusing are of family members indoors, usually with artificial house lighting, so the lens was as open as possible. Really fascinating read though - not something I'd been aware of.
I might have to wait a day or two to find a sufficiently distant and large-enough object to check against. Hopefully a decent view of the Sydney skyline, which is about 7-8km away. The only distant (>1km) thing I can see from my apartment appears to be a two-story building on a hill on the other side of the harbour, about 5km away, but it's just not big enough - its apparent size is only about three times the width of the split in the split image. I will report back. Though, the distant object check was only confirmation of the check with the known-good SLR, which I'd done both ways around (partly to convince myself it worked). Scratched film at the film plane of one, both lenses at infinity stops, and observing through the viewfinder of the other, then switched. As both directions showed a clear focus aligned well via the split image, I think that ought to indicate correct lens infinity focus and mirror/focus screen positioning for both cameras. Please correct me if I'm wrong.

 

[grain elevator]

I've had similar problems. IIRC it turned out that DOF at or around infinity is so large that infinity was still sharp after I adjusted the camera for correct focus at close distances. This might not be the case with long tele lenses, but if you've only tested with a 50... What do you have to lose? Try it!
One thing about your method that might introduce error is using film at the film plane. Can you be sure it is as flat as film is during exposure, with a pressure plate?

 

[Dan Fromm]

Its time for the obvious stupid question. hartacus, when you took your test shots was the camera well locked-down, as on a tripod, or did you shoot it handheld?

I ask because I used to do a lot of closeup work handheld with flash and occasionally lost shots because I moved very slightly as I was taking the shot.

 

[roblopes]

I've had to sideline my Canon A-1 due to a few rolls of film that have demonstrated a back focusing issue, especially with close subjects in low light where the FD 50mm f/1.4 lens has a paper-thin depth of field. Today I had a bit of time so I set about attempting to fix it. I suspected that the mirror angle might be incorrect and the mirror stop might need adjusting. However, using another SLR with known-good focus as a collimator showed that infinity focus was spot on, and I confirmed this using a distant (>300m away) object. I was surprised by this - checked it several times and I'm fairly confident of that result. I guess I hadn't noticed it during use because I don't come across many situations where I'm taking photos of distant objects with the lens wide open.

So I tested a much closer subject, at or near the lens minimum focus distance. Initially with the FD 50mm f/1.4 SSC lens. The subject was a piece of film, cut between two sprocket holes, placed at around 70 degrees to the film plane angle so that when focused (using the viewfinder) on the leading edge of the piece there'd be a relatively clear visual indicator of whether the focus (on the film plane) was behind the leading edge. Focus at the film plane appeared to be at around the second sprocket hole, not the leading edge. I re-tested with an FL 50mm f/1.8 and got a very similar result, which leads me to think that it's not the lens.

What could cause an SLR to have accurate focus at infinity but back focus for close subjects?

(Note that I re-seated the focusing screen not long ago, taking care to insert it in the correct orientation as it had been replaced backwards by a previous owner.)

I have to argue against it's the mirror box misalignment or anything like that. The film plane is the film plan regardless of the focal distance and the mirror angle and focusing screen doesn't change based on focusing distance either.

I think it has to do something with the glass or technique. I've taken apart a few lenses to clean the elements only to read later on that some lenses are put together a certain way that isn't so obvious to the naked eye. Recently, I tried taking apart a Canon FD 300mm f/4 to clean all of the elements only to find some manual stating that there are optical adjustments that are required when assembling the glass, as if it were being put together at the factory.

Hope you're able to resolve the issue and let others know what you find. It's fun watching the forum try their best to come up with answers.

 

[grain elevator]

I have to argue against it's the mirror box misalignment or anything like that. The film plane is the film plan regardless of the focal distance and the mirror angle and focusing screen doesn't change based on focusing distance either.

I think it has to do something with the glass or technique. I've taken apart a few lenses to clean the elements only to read later on that some lenses are put together a certain way that isn't so obvious to the naked eye. Recently, I tried taking apart a Canon FD 300mm f/4 to clean all of the elements only to find some manual stating that there are optical adjustments that are required when assembling the glass, as if it were being put together at the factory.

Hope you're able to resolve the issue and let others know what you find. It's fun watching the forum try their best to come up with answers.

Nope, any issue with the lens would still lead to the same focus on film plane and focusing screen on an accurately focusing body.

 

[shutterfinger]

I do not like split prism finder screens, too easy to get focus errors at close distances.
Read http://www.normankoren.com/Tutorials/MTF5.html , download the test chart, print it on your printer at the printer's highest resolution, mount the chart(s) on a flat board (foam core). Set the camera up on a tripod, place the chart(s) parallel to the film plane measuring from the film plane mark to the target using the closest distance marked on the lens. Focus the target using the area around the split prism. Place a ground glass at the film plane with the back open and the shutter locked open on B.
If the focus is off refocus on the groundglass then check the viewfinder. A piece of regular glass with transparent tape next to each other without overlapping works well. A piece of clear acrylic plastic can be used also. Tape side faces the lens.

 

[reddesert]

Some thought about the lens extensions involved may be useful.
The focusing distance numbers can be calculated from the thin lens equation, 1/f = 1/d1 + 1/d2.

First a number to keep in mind is that for a circle of confusion of 0.03mm, and f-number of 1.4, the one-sided depth of focus is 1.4 * 0.03 = 0.042 mm. So that's the amount of focus error one can tolerate and be "just sharp." For critical work, one would like to do say a factor of 2 better than that.

For a 50 mm lens, the lens-to-film distance required to focus at 1500 meters is 50.002 mm, and to focus at 300 meters is 50.008 mm. So the difference in extension is 0.006mm. This is much smaller than the 0.04mm depth of focus at f/1.4. That means that 300 meters is "close enough" to infinity to be effectively infinity for a 50mm lens. (These numbers are several times smaller than the bellows extension posted by shutterfinger in post #3 - shutterfinger, did you use 50mm as the focal length?)

For the focus offset at closest distance, let's assume that the closest distance was 0.6 meters = 600 mm. And the OP found that the lens focused behind by about 1.5-2 sprocket holes, let's say 5mm behind, so 605mm.
The lens-to-film distance for subject at 600mm is 54.546mm.
The lens-to-film distance for subject at 605mm is 54.505mm.
So a difference of 0.041mm, which is just about the same as the depth of focus at f/1.4 - meaning it would be moderately sharp, but not maximally sharp. It's certainly larger than the nominal difference in extension to get from 300m to infinity.

But, 0.04mm extension is still really small. To give an idea, look at the focus scale of your lens. A 50mm lens extended 0.04mm past infinity is focused at 62 meters. On my 50mm lens, this is turning the focus ring by about 0.7 mm. It's doable, but you have to be quite careful if you want repeatable results. I think it might be useful to try focusing on the ground glass, not with the split image, using a focusing magnifier, and a tripod as Dan suggested.

When you use a split image focusing aid, it "looks through" the lens at a given f-number that depends on the angle of the prisms, so it is possible that one could get focus shift between that f-number and f/1.4. The aberrations of a typical lens are different at its closest focus, so you might get shift at close distances and not at infinity. If you have a macro lens you could test it also.

 

[hartacus]

Thank you so much for your replies. This is all quite fascinating, in addition to the repair challenge.
I examined whether there was a discrepancy between the split image and the microprism ring. There looks to be about 2mm difference at the lens's minimum focusing distance of 0.45m. This was considerably smaller than the focus discrepancies I'd noticed, so I was feeling quite humble regarding my photography technique at that point.
Yesterday, I took a walk over to somewhere I could see the Sydney skyline. Centrepoint Tower and a few of the other skyscrapers made much better verticals to test on. Negligible difference between split image and microprism, both of which showed focus at infinity. So then I tested at the film plane using the MLC building which is white with dark windows and very tall, so it was easy to tell when it was focused (clear plastic, frosted tape facing the lens). And it was way off - the buildings (~6km away) were in focus at just past the 10m mark on the lens. So this would suggest that both the mirror is off (as the viewfinder and film plane give different results) and either the lens focus or the mount distance (less likely) needs adjustment, I think.
Today, I went back to confirm, and got the same result. I also took my 'known good' camera and confirmed that it is, in fact, good - perfect focus through viewfinder and at film plane. I'm confused about why the "SLR collimator" test showed the focus on the other camera was fine at infinity. I must have not done it correctly in one direction.
So tomorrow I'll probably take another walk. I need to confirm whether it's the lens or the camera that's off, or both. To clarify, I didn't get this lens with the A-1, it was a later purchase, so it's not possible that the mirror angle was calibrated to any inaccuracy in this lens (though possibly to another lens). I'll take another Canon and another lens to the same place and try to determine exactly where the fault(s) lies. It seems, at this point, like it's both the camera and the lens, which just happened to create a situation that masked both problems until I examined further. I'm actually quite surprised at how closely the (assumed) inaccuracies match, given camera and lens were sourced independently.

 

[shutterfinger]

So then I tested at the film plane using the MLC building which is white with dark windows and very tall, so it was easy to tell when it was focused (clear plastic, frosted tape facing the lens). And it was way off - the buildings (~6km away) were in focus at just past the 10m mark on the lens.

 

[hartacus]

Thanks for your reply shutterfinger. Are you pointing out that anything between 70m and infinity would be acceptably sharp at the correctly set infinity focus of a 50mm lens? I think that accords with what I saw. I think the 'grains' of my frosted tape were probably considerably larger than 0.02mm so there would be considerable margin of error in the measurement. But the point was that setting infinity focus using the viewfinder coincided with the infinity mark on the lens but produced an image at the film plane that was not even close to acceptably sharp. An acceptably sharp image at the film plane was obtained when the lens was set just past the 10m mark (kind of on the final vertical of the zero).

I found what was wrong with my two-SLR collimation attempts. It was a combination of erroneous preparation and inexperience. I was using a piece of film scratched by the end of a compass, which turned out to produce a scratch that was far too wide and irregular for the technique. I repeated the experiment using overhead transparency (quite good at staying flat, as that's what it's made for) that I very finely scored using a craft scalpel producing much, much finer marks. And I aligned the backlighting so that it would intersect with these scores at an angle (or wherever I could get maximum contrast - not heaps easy given the fineness of the scores). This made my inexperience evident - I was expecting a large misalignment between the split images, but the misalignment was in fact incredibly small - barely the width of the fine score. Nevertheless, aligning the marks exactly using this finer method coincided with the same point on the lens where the city skyline was in focus.

I switched out the FD 50mm f/1.4 for an FL 50mm f/1.8 II which I've never had a problem with, and infinity focus on the lens was was correctly in focus at the film plane. So that confirms for me that the FD lens is not set correctly. And that the viewfinder is not set correctly, since it accords with the lens. Currently waiting on a 3.5mm spanner to arrive in the post to adjust the mirror angle so that the focus in the viewfinder matches the film plane.

(Note: I also checked the flange to film plane distance - all good there.)

 

[shutterfinger]

Thanks for your reply shutterfinger. Are you pointing out that anything between 70m and infinity would be acceptably sharp at the correctly set infinity focus of a 50mm lens?

No, just showing the range one should expect. I take DoF calculators with a grain of salt. DoF calculators are math formula based while actual lens are rarely the exact marked focal length and the lens design can affect the actual results one gets.
One of my lens test targets on 2 ft x 3 ft foam core. The 1951 USAF targets are printed at 2 inch square on gloss photo paper on an Epson 2200 photo. They are taped with double sided tape in the pattern seen.

Nikon 70-200mm f2.8 G at 70mm f2.8, full frame.

 

[flavio81]

Hi,

Seems like this is a combination of different issues, right?

I own an A-1 and have used it on the past for professional assignments with 55/1.2 and 200/2.8 lenses with little focusing issues. It was/is a reliable machine in all regards, no matter what people might think.

As for your infinity focus test, I used to test infinity easily -- i lived in a tall building which had a great view of a good part of the city skiline (objects kilometers away). I placed a ground glass (taken away from another SLR camera, like for example a Nikon F3 matte screen) in the film rails and checked that the focus at infinity agreed with the viewfinder.

Mirror angle can cause a LOT of problems so should be checked! I'm sure there are more professional methods to check, but for me an easy check would be to focus to infinity using the top edge of the screen and confirming that the bottom edge also is in focus. Any mirror angle problem will make you have different focus points at the top and bottom of the screen.

As for focusing iself, i hate the split-image rangefinder! A worthless piece of crap in any camera IMO. I like matte screens above all, and then microprisms. The standard screen in a Spotmatic is a good example of a well executed focusing screen.

I'd advise you to always focus using the matte area of your screen.

Now, if you are evaluating artificial light versus daylight tests, MAYBE there's a tiny bit of focus shift due to chromatic aberration, but that would be nitpicking. In any case if you want to be really fussy you would do the test using a green filter and daylight!

 

[hartacus]

I think I've reached something of a conclusion on this issue, so I thought I'd update the post with where I've landed.
I checked focus again with the FD 50mm f/1.4 using another known-good camera as a collimator, but this time I did it with a background of bright daylight. It's hard to give a moment to do that with kids, so I had been doing it at night. Viewing through the other camera, with both lenses at infinity stops and a scratched film at the Canon A-1 film plane, focus appeared perfect. So too in the other direction. So both lens and viewfinder on the A-1 seemed fine. Ate a slice of humble pie.
I had another thought. I had noticed the camera glitching sometimes, producing really long exposures without reason, and sometimes showing the wrong shutter speeds. Turns out it was jumping out of Av mode into Tv mode - they use the same dial with the mode selected by a switch, and it was the switch that was glitching. This was most evident with the aperture set to 1.4 on the dial, because the corresponding shutter speed on the same position of the dial is 30 seconds. But at an aperture of 4, the corresponding shutter speed is half a second. The higher the aperture, the closer the erroneous shutter speed gets to what it should have been. This fits with the issue being mostly noticeable taking photos of close subjects indoors, with less light, open apertures, thus longer shutter speeds than expected, and exacerbated effects of shifts of position (I.e. my bad technique). Ate another slice of humble pie. The switch simply bridges two contacts when in Av and opens them in Tv, but it uses two brushes mounted below a sliding mechanism which is mounted on a surface that can flex upwards away from the contact point of you put uneven pressure on it, such as pushing down on the shutter button surround while taking a photo. I cleaned the contacts with isopropyl alcohol and bent the brushes slightly to ensure a better contact, and so far the camera hasn't jumped modes again.
One other possible cause of blur is that this camera sometimes kicks like a mule. Hard to describe it, but sometimes feels like it's trying to twist its lens off when taking a photo.
A test roll didn't show any focus issues, so I'm putting it down to a combination of the shutter glitch, handholding blur, and not realising what I was actually seeing in the test photos. You live and learn I guess! Thanks all for your suggestions.