I've noticed that most SLRs have shims or adjustments around the sensor, lens mount, or both that obviously are there to keep proper alignment between the sensor and the lens.
Mirrorless cameras don't seem to have shims or adjustments at either location.
I assume, then, that either alignment is less critical in a mirrorless camera (shorter flange-sensor distance, perhaps?) but I don't understand why that would be. Or perhaps 'on sensor' focusing and metering make it less critical?
Or perhaps manufacturing tolerances can be tighter in the smaller cameras? Or perhaps something else entirely?
Anyway, I'm just curious about why if someone knows the reason.
Interesting observation!
Yes, since the focussing is done through the image sensor, focus will be accurate regardless. Plus most lenses for the mirrorless system can focus past infinity to compensate for small flange distance variation.
In an SLR, the focussing screen, the PDAF module and the image sensor must be calibrated against each other.
Yes, it is the fact that it is a CDAF system that means there is no need for shims.
A DSLR requires three optical paths to be identical - flange to sensor (for the actual image capture), flange to OVF screen via the first mirror (for MF), flange to AF sensor via second mirror (for AF). As long as all three paths are the same it doesn't matter if the path is a little shorter or longer. However, even a slight difference in path length between any of these paths and you'll either have an AF error, MF error or both. So at least two of the three need to be adjustable - often with shims.
A mirrorless system has only one optical path - flange to sensor - regardless of whether exposing, AFing or MFing. So typically no shims required anywhere.
Is it practically feasible to set these tolerances low enough that their effects are trivialized, or are DSLRs bound to have a bit of slop in the focusing accuracy relative to what can be had with mirrorless?
Dan1 wrote:
Is it practically feasible to set these tolerances low enough that their effects are trivialized, or are DSLRs bound to have a bit of slop in the focusing accuracy relative to what can be had with mirrorless?
Yes it's possible, but at a cost. Take 100% coverage viewfinders - it's more expensive to manufacture / calibrate to such tight tolerances.
Now you can calibrate a lot of things yourself - through Micro Adjust you help the auto focussing accuracy, and by shimming the focussing screen correctly you can have better manual focussing.
Note that for the stock focussing screens, accurate shimming isn't as important because its greater depth of field covers up small errors.
jotdeh wrote:
Oh, and it will never be as accurate as a mirrorless, which as pointed out doesn't have different optical paths that need to be matched.
Of course not, but there must be some point where tightening DSLR tolerances further yields no further tangible benefit. My question is whether DSLRs produced for the mass market can/do reach this level.
Two optical paths = two sets of precision manufacturing. One optical path = one precision.
Part of the extra cost @ pro grade DSLR's is the precision @ both paths ... consumer grade DSLR's aren't given such tight tolerance @ spec as pro-grade and run into more needs for shim (not all are shimmable).
There's actually three paths through the mirror box... Sensor, viewfinder and AF. Getting them to play together takes some adjustment points. There typically shims in the matte screen, mount flange and sensor board mount, and adjustment screws for the alignment of the AF-sensor module. And three separate ways to adjust the mirror / submirror angles... :/
Another adjustment is collimation, making the optical axis normal to the sensor plane. Even with mirrorless designs, collimation error will defocus one side of the image. Fix requires shimming the mount.
