Front focusing lens - formula?

[RichC]

I've got a lens that front focuses by a few inches and thus needs the internal shim ground down.

Is there a mathematical formula I can use to calculate how much I need to reduce the shim thickness by, to cut down the trial-and-error assembly/disassembly of the lens a bit?

(The camera has a Leica M mount, and its rangefinder is adjusted accurately.)

 

[jlw]

Hello, Rich --

I've got a spreadsheet I made that might help you. It's based on formulas in Rudolf Kingslake's book Lenses in Photography. I see that .zip and text formats are no longer allowable as attachments, but you can download it from this link:

http://homepage.mac.com/jlw/temp/lens_extension_wksht.zip

The part that ties in to what you want to do is the table at the bottom. It's based on a test photo made at an intended focusing distance of 1000 mm. Common lens focal lengths are listed down the left side, and actual best-focus distances are in columns across the top. The idea is that you'd photograph an angled ruler with its midpoint 1000 mm from the image plane, then look at the photo to determine the actual best-focus distance (remember to compensate for the angle of the ruler.)

For example, suppose you're using a nominal "50mm" lens (actual focal length according to Leica lore of 51.6mm) and you find your actual best-focus distance is 992 mm instead of your intended 1000 mm.

This means your lens is "front-focusing" (focusing in front of the intended target) by 8 mm, which in turn means that the optical section is extended too far relative to the rangefinder cam (since closer focusing distances require more extension.) The table tells you that to correct this condition, you need to move the optical section backward by 0.03 mm, which you'd do by thinning the shim by this amount.

(Actually, if you download the table, you'll notice that the same correction is given for 990 mm and 992 mm, which might make you wonder how careful I was in setting up the table. Keep in mind, though, that commonly-available measuring instruments aren't going to read out any finer than the nearest 1/100 of a mm, so I figured that carrying everything out to four or five decimal places would be "wasted precision" and just rounded the numbers off to the nearest 0.01 mm.)​

WARNING! Because of this rounding, and also because I used a shorter formula that Kingslake described as good approximation, rather than an extremely long formula that's more precise, there's no guarantee that these values are exactly correct! USE AT YOUR OWN RISK!

Since it's a lot easier to remove material from a shim than it is to add it back, I recommend thinning the shim gradually until it starts to get close to the value in the table, then check frequently and "sneak up" on the thickness that gives the best results -- rather than removing the full amount all at once!

The table has given good results for me in zeroing in some of my Canon screwmount lenses for use on my R-D 1 camera body, but again, there's no guarantee on this data, so you're on your own!

 

[RichC]

Thanks jlw - exactly what I need! Conveniently, the lens I'll be adjusting is a Canon 50/1.2 on an R-D1...

(I don't know why this lens often gets panned: I'm extremely impressed - I was expecting it to be mediocre, but was stunned by its performance wide open. Especially considering that its design was calculated without computers.)

(Tip for those collimating lenses: I've always had difficulty finding shim material in the UK, but when I took my friend's son to a local model show recently, I found a stall (www.westbourne-model.co.uk) that sold metal for model-making (brass tubes, sheet, etc.), and they had packets of assorted brass sheets for shims. So, anyone needing brass shims, just google something like "model craft brass shim".)

 

[Kim Coxon]

Many model shops in the UK only stock a limited range. Mainly Trains in Devon have a good selection and a fantastic mail order service. 😉

http://www.mainlytrains.co.uk/acatalog/gen-metals.html

Kim

(Tip for those collimating lenses: I've always had difficulty finding shim material in the UK, but when I took my friend's son to a local model show recently, I found a stall (www.westbourne-model.co.uk) that sold metal for model-making (brass tubes, sheet, etc.), and they had packets of assorted brass sheets for shims. So, anyone needing brass shims, just google something like "model craft brass shim".)

 

[RichC]

Lens now sorted (unlike my finger, sadly - edges of ground-down shims are sharp!), and focus is spot on.

Thanks once again to jlw for the chart, which made the whole operation quicker by lessening trial and error, and to Brian (decided to grind down my own shim).

Suprised by how much the lens was off (10 cm front focus at 1 m and f/1.2 with an accurately calibrated Epson R-D1), as it looks as if it's never been dismantled since leaving the factory, and all the components were correctly located and nice and tight - had to grind down the shim by nearly a quarter of a millimeter: perhaps that explains why this Canon 50/1.2 looks almost unused despite being about 40 years old!

 

[jlw]

Lens now sorted (unlike my finger, sadly - edges of ground-down shims are sharp!), and focus is spot on.

Yeah, my fingers are the same way. Try working on the shim from a 50/0.95 -- it's bigger, so it makes bigger slashes!

We've all got to make sure we never commit crimes now, or we'll be too easy to identify: "Captain, we didn't get any matches on the fingerprints, but we know from these distinctive marks that we're looking for someone who owns a Canon rangefinder lens!"

Suprised by how much the lens was off (10 cm front focus at 1 m and f/1.2 with an accurately calibrated Epson R-D1)

Recently while flipping through the repair manual for a Canon 7s I came across something on the specifications page that may partially explain this. It seems that Canon used both a slightly different measurement and a slightly different tolerance than Leica did for flange-to-film-plane distance.

(This doesn't mean that either Canon or Leica was "wrong" -- instead it reflects different assumptions about such factors as film thickness and curl. Film never lies completely flat within its channel, which is somewhat thicker than the film itself, so a manufacturer has to make some assumptions about where the "average" film will lie at the moment of exposure. Dante Stella has a long explanation of this in his writeup about Hexar RF focusing.)

Epson had to make a different set of assumptions in setting up the internal dimensions of the R-D 1, since it forms its images on a sensor instead of a piece of film. (In another thread it's been claimed that they standardized their setup using the Cosina 35mm f/2.5 lens, presumably because it's popular and widely available.) So if Epson's choices lay at one end of the tolerance band, while Canon's lay at the other, it would be no surprise that many Canon lenses are a bit "off" when used on an R-D 1.

Optical characteristics such as focus shift and field curvature pose another potential judgment call: most lenses of the '50s (especially high-speed ones) would shift focus slightly as they were stopped down, so in collimating them the manufacturer had to guess whether the user were more likely to shoot them at full aperture or a more moderate aperture. And many lenses then (and now) formed their image on a slightly spherical plane rather than a flat one, so the manufacturer had to guess whether the user would be more likely to position the main subject smack in the middle or farther out toward the edge (which would have a slightly different focus point.)

With all these variables, it always surprises me that photography works at all! And it certainly makes it explicable that many parts of the camera system require a bit of "fine tuning" for optimum performance, especially when you're mixing and matching pieces made by different manufacturers many, many years apart!