Tolerances, alignment, enlargement

One of my favourite YouTube rabbit hole in the recent years has been machinists' videos, people who chip away at steel on lathes and mills, measuring things by the thousandths and ten-thousandths of an inch. It made me develop a whole new appreciation for precision, and helped me understand a thing or two about manufacturing that make sense when looking at the way cameras are made.

It's also fascinating because there are case scenarios were the thickness of a few human hairs is enough to cause problems. It's not the insane levels of precision of high-quality optics (like those used to make microprocessors), but it seems that cutting metal is much more, demanding in terms of alignment than enlarging.

I look at a Bridgeport mill, and I see the most perfect enlarger alignment system; I look at a granite surface plate and tell myself it would make an amazing baseboard...

But in actuality, we're dealing with quite soft sheet metal materials, screwed in wood or plywood, that can bend in and out of alignment, without major issues when we're printing at most an 11x14 in the home darkroom.

Who has actually measured the effective tolerances of a home enlarging setup? How many thousandths of an inch (or microns) can a light path suffer before an 11x14 is fuzzy instead of sharp? And before you say "Ctein", I did check Post Exposure, and he mentions being able to spot misalignments of 0.25 mm with his Zig-align system, but it tells me little of what is an acceptable misalignment from an unacceptable one.

I certainly have. Whether I needed to or not is another story. I just like making things right, and enjoy working with precise equipment. But I also happen to make very precise 30X40 inch high gloss color prints from large format originals, so every little bit helps. I was an equipment dealer for many years, so had access to things on an affordable basis which might be prohibitive for most others. No Zig-Align. I used a far more expensive industrial laser, along with true machinist's levels. On my biggest enlarger, the controls are micrometer-gear driven.

Call it dumb luck or whatever; but the yaw correction supporting the whole support rod on my biggest 8X10 color enlarger (14 ft tall) was a free military surplus Naval big gun sight mount - solid machined bronze, 3-way leveling, micrometer driven - more solid than any gear driven tripod head in existence.

The vacuum easel is a pin-registered machined unit so solid that I can stand on it without deflecting it. It was cannibalized from a 22 foot long Japanese process camera, along with a full set of Apo Nikkor repro lenses, superior to any official enlarging lenses.

No flimsy plywood. The support column is a massive phenolic strand column, picked with marine penetrating epoxy, then black laminated and edge. Completely dimensionally stable, and designed to be earthquake proof. All kinds of deluxe details like that, all through the machine. Pin registered carriers. Some guys like working on cars in the shop, some motorcycles, others speedboats. I liked fooling around with enlargers, etc.

But thinking in 10/000th or even 1,000th? That might apply to lens manufacture; but it's nothing you need to worry about in a darkroom. No purchased enlarger is that tight on its specs. And I have several true commercial enlargers - 5X7 and 8X10 Dursts, along with my personally built big one. Basic flatness and leveling are the most important, along with rigidity to the whole system, including wall bracing.

An exception would be punch and register carriers and easels. The spacing of punch holes and pins does have to be very accurate, along with the repeatability of carrier position itself. A single component for a system like that would have cost quite a bit more than any complete amateur enlarger, and would have involved true machine shop workmanship - no extruded aluminum in that kind of thing!

One can use the focusing equation of Hansma to predict the toleranced needed for a sharp print, given a goal circle of confusion size on the print. This equation takes into account both the deteriioration of the image by diffraction and the diminished tolerances required as f-stop number increases:

N = (20/(1+M)) * square root of 'dv'

N = Aperture number
20 = empiric constant for a circle of confusion about 0.15mm on the print
M = magnification
'dv' = millimeters of focal depth on the baseboard.

So, with some real-life numbers, at f8, a five power enlargement has an error of focus at the baseboard of 5.76mm and still yield a sharp print.

At the negative state the tolerance would be about 1mm.

I prefer not to be confused, and avoid that social circle entirely. Not everyone has the same definition of "sharp". But what I really mean is that seemingly insignificant errors add up. You might get away with one or two misdemeanors in focus or alignment; but people who tend to be careless end up with a felony at some point, and it shows. And math that makes sense on an optical bench gets meaningless pretty fast with a misaimed funky enlarging lens - those aren't all created equal either.

But I'd disagree even with your math. A 5X enlargement from 4X5 would be approximately a 20X24 print, and the neg stage focus being off 6mm (1/4inch) at f/8 would indeed be a visible felony, and not a misdemeanor. But where does enlarging lens focal length itself fit into all that? - that's a factor too. I've done 4x5 enlargements using lenses all the way from 135mm to 240mm.

NEGATIVE STAGE tolerance in the example is ONE mm.

The constant (20 in the example) is an EMPIRIC constant. So, for example, in your work if only 0.5mm tolerance at the negative stage is acceptable for a 5x print, the constant you would use would be "10."

Lens focal length cancels out on both sides of the equation when magnification is used in the equation.

All it takes is a slight deflection at the film stage, due to a glassless carrier, to cause a visible anomaly in the print. All these seemingly little things potentially add up to a bigger problem. That's why my mantra is, You're only as good as your weakest link.