Longer OR shorter lens...smaller max aperture
Moderators: rjlittlefield, ChrisR, Chris S., Pau
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Longer OR shorter lens...smaller max aperture
Can anyone explain why there is a sweet spot around 50mm or so where large apertures are possible, but for both shorter and longer lenses the apertures get smaller? In my mind the longer end makes sense because longer lenses mean larger glass for the same aperture, but on the shorter end the glass gets smaller, so diameter is not the problem. Is it thickness of the elements that is the issue on shorter lenses? I ask this because I see 35mm or 38mm macro lenses like the Canon or Olympus available at f2.8, but the 20mm from same mfr are f3.5. Same trend happens on microscope objectives of course. What is the limitation?
- enricosavazzi
- Posts: 1479
- Joined: Sat Nov 21, 2009 2:41 pm
- Location: Västerås, Sweden
- Contact:
No doubt others may provide more precise information, however, a simple answer is that at least some optical aberrations scale up much faster than focal length.
Take, for instance, a 25 mm f/1 lens. It is entirely possible to design and build one of these to provide a high resolution and low spherical aberration fully open. Now magnify the lens design twice (without changing anything but scaling up the absolute size of all components) and you end up with a 50 mm f/1. Most likely this lens will be very poor, because some of its aberrations have scaled up by much more than 2x. Magnify the lens again to 100 mm f/1, and you get a big, heavy and expensive paperweight that looks great but performs too poorly to be any use in practical photography.
There are other obvious reasons why fast lenses of long focal lengths are impractical. A 500 mm f/1 lens needs to have a front element with a diameter of at least half a meter, which would probably cost more than a luxury car to produce. The finished lens would be so large and heavy that it needs to be permanently mounted on a pickup truck for use on location. At this point, how many copies of the lens would be sold? In addition, to correct the increased aberrations discussed above, the required optical design would become so complex (e.g., a front group of 15-20 elements instead of the usual 3-4) that it may in fact be impossible to design and manufacture to the required tolerances.
At the opposite end of the range of focal lengths, other factors come into play. If you just reduce the size of our initial design to 12.5 mm f/1, it may perform great in the center of the image, but its image circle will also be halved. Increasing it requires a more complex and expensive optical scheme. Slower lenses are of course cheaper and easier to design. Also, digital camera sensors do not capture well light striking them at a low angle, so the rear lens element must be enlarged and/or moved farther away from the sensor, which in turn forces the front elements to become wider, and this in turn forces a more complex, expensive and aberration-prone design.
Take, for instance, a 25 mm f/1 lens. It is entirely possible to design and build one of these to provide a high resolution and low spherical aberration fully open. Now magnify the lens design twice (without changing anything but scaling up the absolute size of all components) and you end up with a 50 mm f/1. Most likely this lens will be very poor, because some of its aberrations have scaled up by much more than 2x. Magnify the lens again to 100 mm f/1, and you get a big, heavy and expensive paperweight that looks great but performs too poorly to be any use in practical photography.
There are other obvious reasons why fast lenses of long focal lengths are impractical. A 500 mm f/1 lens needs to have a front element with a diameter of at least half a meter, which would probably cost more than a luxury car to produce. The finished lens would be so large and heavy that it needs to be permanently mounted on a pickup truck for use on location. At this point, how many copies of the lens would be sold? In addition, to correct the increased aberrations discussed above, the required optical design would become so complex (e.g., a front group of 15-20 elements instead of the usual 3-4) that it may in fact be impossible to design and manufacture to the required tolerances.
At the opposite end of the range of focal lengths, other factors come into play. If you just reduce the size of our initial design to 12.5 mm f/1, it may perform great in the center of the image, but its image circle will also be halved. Increasing it requires a more complex and expensive optical scheme. Slower lenses are of course cheaper and easier to design. Also, digital camera sensors do not capture well light striking them at a low angle, so the rear lens element must be enlarged and/or moved farther away from the sensor, which in turn forces the front elements to become wider, and this in turn forces a more complex, expensive and aberration-prone design.
--ES
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: Longer OR shorter lens...smaller max aperture
Good question, and I don't have anything to add to Enrico's explanation.ray_parkhurst wrote:Can anyone explain why there is a sweet spot around 50mm or so where large apertures are possible, but for both shorter and longer lenses the apertures get smaller?
I disagree with this bit. It's true that microscope objectives have small effective apertures on the camera side, but that's because of the magnification. If you look at the lens itself, the equivalents are NA 0.40 = f/1.25, NA 0.60 = f/0.83, and NA 0.90 = f/0.56.Same trend happens on microscope objectives of course.
--Rik
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Re: Longer OR shorter lens...smaller max aperture
Yes, I was thinking effective rather than infinity aperture.rjlittlefield wrote:I disagree with this bit. It's true that microscope objectives have small effective apertures on the camera side, but that's because of the magnification. If you look at the lens itself, the equivalents are NA 0.40 = f/1.25, NA 0.60 = f/0.83, and NA 0.90 = f/0.56.Same trend happens on microscope objectives of course.
--Rik
A 20mm f3.5 macro lens intended for 4x-10x is a little better than a run of the mill 4x/0.1 objective at 4x, and far outclassed by a 10/0.25 objective at 10x, correct? And why in the same line of lenses is the 35mm f/2.8 while 20mm is f/3.5? And finally, why do these sell for high $ when microscope objectives with better performance can be had for far less?
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: Longer OR shorter lens...smaller max aperture
Correct. At 4X (and assuming unity pupil factor), the f/3.5 lens will be running at effective f/17.5 while the NA 0.1 microscope objective will be f/20.ray_parkhurst wrote:A 20mm f3.5 macro lens intended for 4x-10x is a little better than a run of the mill 4x/0.1 objective at 4x
I'm not sure what this means. The 10/0.25 objective at 10X will be running at effective f/20, same as the 4x/0.1 at 4X. To get a lot more sharpness from the 10X, it might be pushed down to 5X where it would be effective f/10 and then much sharper than the other two (but probably with a smaller image circle).and far outclassed by a 10/0.25 objective at 10x
There's some variation here. In the Olympus bellows macro lenses, it is 38mm f/2.8 versus 20 mm f/2.0 (newer automatics) and 38mm f/3.5 versus 20 mm f/3.5 (older manuals). I don't know why Canon decided to go the route of 35/2.8 and 20/3.5. It seems an odd choice to me too.And why in the same line of lenses is the 35mm f/2.8 while 20mm is f/3.5?
I would guess it's a matter that the objectives are simpler to manufacture and sell more units. In used prices, there's also an element of prestige and rarity.And finally, why do these sell for high $ when microscope objectives with better performance can be had for far less?
--Rik
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
Re: Longer OR shorter lens...smaller max aperture
True, but I meant that the 10x/0.25 at 10x far outclasses the 20mm f3.5 at 10x.rjlittlefield wrote:I'm not sure what this means. The 10/0.25 objective at 10X will be running at effective f/20, same as the 4x/0.1 at 4X. To get a lot more sharpness from the 10X, it might be pushed down to 5X where it would be effective f/10 and then much sharper than the other two (but probably with a smaller image circle).and far outclassed by a 10/0.25 objective at 10x
The aperture adjustment was of extreme value in the pre-digital age when achieving acceptable DOF in a single shot was important. But today these seem like relics now, like Vinyl records or VHS tapes, though you can always run them wide open...Rayrjlittlefield wrote:I would guess it's a matter that the objectives are simpler to manufacture and sell more units. In used prices, there's also an element of prestige and rarity.And finally, why do these sell for high $ when microscope objectives with better performance can be had for far less?
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
Re: Longer OR shorter lens...smaller max aperture
My bad -- I missed that interpretation. Yes, that comparison is night and day: effective f/20 versus f/38.5.ray_parkhurst wrote:True, but I meant that the 10x/0.25 at 10x far outclasses the 20mm f3.5 at 10x.
--Rik
-
- Posts: 128
- Joined: Fri Aug 15, 2008 1:53 pm
- Location: Detroit, Michigan
Simple explanation when you're talking about 35mm SLR lenses.
SLRs have a swinging mirror. Together with the shutter, this means that the rear element has to be at least 38mm from the plane of focus. That distance, the "back focus" drives normal and wide angle lens designs.
Using 1950 or 1960 lens design computations and glass formulas, relatively affordable, high performance fast designs, such as the double Gauss (a good design from f2 right down to f0.9) tended to be symmetrical, the rear elements were basically the front elements, turned around. Which made the front node and rear node "coincident", right inside the lens, at the center, along with the aperture. They als tended to be past "square", the actual optical part of the lens assembly was longer than it was wide.
So, the "perfect" fast lens was the 58mm f1.4. It's 41mm in diameter (58mm/1.4) and about 42mm long, with 21mm behind the 58mm node, which put the rear element right at the mirror. You could make f1.4 lenses longer, you just scale up the double Gauss. They just get larger and heavier. An 85mm f1.4 weighs 3x what a 58mm f1.4 does.
But making them shorter was a pain. You can't make a double Gauss shorter than 58mm, the rear element would crash into the mirror. So, you make a "retro focus" lens. You take a 58mm f1.4, and you stick a "reversed telescope" in front of it, a strong negative meniscus (bulging double curve) lens facing the subject, a positive lens facing the 58nm f1.4. For a 28mm f1.4, you're talking a 2x reversed telescope, big enough to cover the 41mm entrance of the 58mm f1.4. That's big, complex, and heavy.
Modern glasses, grinding, molding, and computation help a little, which is why there are some relatively small, affordable, fast short lenses, like the Sigma 30mm f1.4, but they're few and far between.
SLRs have a swinging mirror. Together with the shutter, this means that the rear element has to be at least 38mm from the plane of focus. That distance, the "back focus" drives normal and wide angle lens designs.
Using 1950 or 1960 lens design computations and glass formulas, relatively affordable, high performance fast designs, such as the double Gauss (a good design from f2 right down to f0.9) tended to be symmetrical, the rear elements were basically the front elements, turned around. Which made the front node and rear node "coincident", right inside the lens, at the center, along with the aperture. They als tended to be past "square", the actual optical part of the lens assembly was longer than it was wide.
So, the "perfect" fast lens was the 58mm f1.4. It's 41mm in diameter (58mm/1.4) and about 42mm long, with 21mm behind the 58mm node, which put the rear element right at the mirror. You could make f1.4 lenses longer, you just scale up the double Gauss. They just get larger and heavier. An 85mm f1.4 weighs 3x what a 58mm f1.4 does.
But making them shorter was a pain. You can't make a double Gauss shorter than 58mm, the rear element would crash into the mirror. So, you make a "retro focus" lens. You take a 58mm f1.4, and you stick a "reversed telescope" in front of it, a strong negative meniscus (bulging double curve) lens facing the subject, a positive lens facing the 58nm f1.4. For a 28mm f1.4, you're talking a 2x reversed telescope, big enough to cover the 41mm entrance of the 58mm f1.4. That's big, complex, and heavy.
Modern glasses, grinding, molding, and computation help a little, which is why there are some relatively small, affordable, fast short lenses, like the Sigma 30mm f1.4, but they're few and far between.
-
- Posts: 3439
- Joined: Sat Nov 20, 2010 10:40 am
- Location: Santa Clara, CA, USA
- Contact:
-
- Posts: 128
- Joined: Fri Aug 15, 2008 1:53 pm
- Location: Detroit, Michigan