The lens is a huge and critical part of the autofocus system, after all, it actually does the focusing.
Far from being a single lump of glass, they’re actually intricate and complex systems of exotic crystals, metal and electronics which are designed to work in harmony to focus light of any visible colour to a sharp image on the camera sensor.
And when you have a complex design and a lot of interacting parts, reality can sometimes be a little different to the perfect theoretical design.
About this Series
This post is part of a series looking at on-sensor phase-detect autofocus, the whole autofocus system, and why mirrorless autofocus might still need some calibration.
The full list of posts are:
- Part 1: You SHOULD Calibrate your Mirrorless Autofocus!
- Part 2: On-Sensor Phase Detect Autofocus
- Part 3: On-Sensor Phase-Detect Issues
- Part 4: The Lens
- Part 5: The Autofocus Processor
- Part 6: The Evidence
There are 3 distinct components of the lens that I’ll look at in this post:
Issues affecting the optical path can introduce errors in the results from the AF Sensor, and mechanical issues can change how accurately the lens is positioned, both having a direct effect on the overall sharpness of your final image.
First up, let’s talk about the bit you – and the camera – actually look through: the lens optics.
Camera lenses typically contain more than 10 individual elements of various types of glass, quartz and fluorite, arranged in groups that cleverly compensate for the shortcomings of the materials and manufacturing processes.
Here’s a simplified model of a lens with all the mechanics and casing removed, showing the passage of light through the middle of a lens straight out of the factory, behaving just as it should:
[Note that in these images, the blue line just illustrates the centre of the lens – it’s not showing the effect of any errors, i.e. it won’t move or change shape as I “break” the lens in various ways below.]
As part of the construction process, some of the lens groups are physically adjusted with spacers, wedges or screws and tuned to operate within tight tolerance limits. But, as is a common theme here – that’s in the factory, which may have been months, years or even decades ago.
With so many physical components in the lens controlling the path of the light, it’s not difficult to imagine that some of them might not be in quite the same position as they were when they left the factory.
Here’s a look at some of the possible ways a lens can lose its edge:
Being able to change lenses is a superb benefit of DSLR and mirrorless cameras, offering massive potential which is well-known and exploited. But that convenience introduces one possible way in which the optical path can be upset.
I’ve added a camera behind our exploded lens, and shown a seriously misaligned lens mount in red. If your camera looked like this, I’d suggest you send it off for repair! But hopefully, the exaggeration clear shows the misalignment of the mount.
This sort of misalignment can be caused by many things: the mount screws not being tight, movement in the mount base due to hanging very heavy lenses off the camera, and even grit or dirt in the flange.
With a tiny offset at the mount, the whole lens – including the full light path – will be shifted at a slight angle to the sensor. Tiny tilts can have a significant effect at the far end of the lens, even with moderate-length lenses: a 1-degree offset, which would be visibly unnoticeable at the mount will equate to almost 4mm movement at the far end of a Nikon 70-200 f/2.8 lens.
This sort of offset causes the image to be sharp only on part of the sensor, so if your focus point was on one side of the image, the other side could be significantly softer.
And this can have an affect for the autofocus sensing pixels too, as we’ll see below.
As we’re looking primarily at Nikon Z-cameras in this series, it’s important to mention the FTZ lens mount adapter that allows you to use F-mount lenses on Z-mount bodies.
It should be fairly obvious what I’m going to say now…
The same issue can occur, but instead of having just 1 mating surface, there are now 2: the camera-to-adapter, and the adapter-to-lens.
Each of those suffers from the same problem as the lens mount did on its own, so the risk of optical offset is doubled with the use of a mount adapter.
Let’s do away with the camera and mount adapter, and look at the lens itself now.
An element or group is considered decentred if it has moved sideways from the position it’s supposed to be in, like this (I’ve shown the lens that’s not in the right place in red to make it clear in these images):
The result of this misalignment varies depending on the lens type and design, how far off-centre the lens is and which lens is out of place. The effects can be anything from slight blurring at the extremities of the image to a completely soft image.
This change to the progression of light rays through the lens has an effect on the autofocus sensor too, leading to slightly incorrect information being reported to the autofocus processor.
Lens elements may also become tilted:
The effect is similar to a decentred lens and often gets categorised similarly when testing.
When you run tests with FoCal Pro, the Astigmatism Factor can give an indication of any elements that are tilted within the lens, as it generally results in images that are sharper on one axis than the other.
Incorrectly Placed Lens Element
The optical design expects the lens elements to be in specific places along the optical path for an image to be correctly focused on the image sensor, and if any of these elements are out of place with regard to the overall optical system, the projected image may be out of focus, or the lens may not be able to focus to infinity.
Without any side-to-side movement of the lens, this sort of defect will affect the image as a whole rather than only softening certain areas. It will slightly change the focus behaviour of the lens which can be an issue when applying the AF Correction Data built into the lens (explained below, and not to be confused with AF Fine-tune).
Image Stabilisation Issues
The optical stabilisation system within lenses involves a floating lens element moved within an electromagnetic actuator.
If the stabilisation system is working correctly, the floating lens will be well controlled and have minimal impact on the optical path (at least in comparison to the benefit it brings by compensating for overall camera movement).
But any issues with this system can introduce errors in focusing which can take the form of any or all of the three issues listed above: decentering, tilting or incorrect placement.
In order to focus the image, there is a group of lens elements – the focus group – which is moved back or forth along the optical axis of the lens. For a manual focus lens, this is performed by your hand turning the focusing ring, but for autofocus lenses, it’s driven by a motor.
There are 4 primary motor types used for moving the focus lens group:
A motor in the body of the camera controls the lens mechanically. This is typically only used in older lenses which have no internal motor, and the focus accuracy and precision are poor when compared to later motors.
Internal DC Motor
When the miniaturisation of motors reached a cost-effective point, they were moved inside the lens bringing the benefit of being about to tune the control of the motor to the specific optics and mechanics of the lens.
The motors used were simple brushed DC motors – the motor turned at a speed roughly proportional to the power applied. The speed control was coarse and there was no way to control the exact position of the motor.
The lens design necessitated the addition of a position sensor which, combined with the motor and electronics, meant that the focusing lens group could be placed with the degree of accuracy required to get an in-focus image, at least on the film cameras of the time… which is a far cry from the requirements of today’s high-megapixel cameras.
The ring-type ultrasonic motor was first introduced in a camera lens in 1990 and has become commonplace in mid to high-end lenses. It eradicated the irritating whine of the original internal motor design and brought a new level of speed and accuracy. These motors can move the focus group in small fixed increments with a degree of precision that wasn’t possible with previous motors.
Modern stepper motors have excellent control, are quiet and are able to accurately and precisely position the lens elements for focusing. These motors are typically used in mid to high-end lenses, including all the native Z-mount lenses for the Nikon Z mirrorless series.
There are two main issues affecting the control of the lens that are directly related to the motor:
- Inherent design limitations
- Physical wear
In-body motors and micromotors have relatively poor precision, so with all the will in the world, the AF system won’t be able to reliably achieve exact focus without several attempts, and then it’s more a matter of luck than judgement that focus is achieved. Think of it a bit like rolling a die – keep going, and eventually, you’ll throw a 6… but it may take a while.
Ultrasonic motors and stepper motors are much more precise in their control of the lens element. Rather than just moving the lens group in a certain direction at an approximate speed, they move in well-defined precise steps.
Older and lower-cost lenses tend to have less precise focus control
Over time, however, the ability to exactly control all types of motor degrades. For DC motors, wear of the brushes changes the speed and torque, and for ultrasonic motors, particles produced from use of the motor can interfere with the piezo actuator that drives the motion. Wear of the bearings affects all types of motors.
The combination of some or all of these issues won’t stop the motor from operating, but it will change the exact behaviour of the motor from how it acted in the factory, and this is important when we consider the internal calibration data within the lens, which I’ll talk about below.
Mechanical wear can affect control of all lens motors
Other Mechanical Components
The motor moves the focus group through a series of gears, rails and other mechanical components, and it’s better to think of the whole set of mechanical components as a complete focus control system.
Each time the lens is focused (or zoomed), the mechanical components rub together and wear. Combined with motor wear over time, there will be tiny changes in exactly how the whole focus control system responds to commands from the camera.
Mechanical Control System
The electronics inside a lens related to focusing do much more than send the focus motor in one direction.
The control electronics have the concept of slow- and fast-operation modes, such that a lens can be moved quickly but less accurately, or slowly and accurately depending on the focus requirement. In the slow mode, intelligent braking is applied to the lens focus elements in order to accurately stop in the required position. There’s also functionality to compensate for lens “backlash” caused by tiny gaps in the motor gearing.
This control electronics is there to mitigate the inherent defects in the design of the lens, and is tuned to operate under the premise that the motors and mechanics are in perfect shape as they were when they left the factory. As we’ve seen above, wear to the motor and mechanics can change the behaviour, and result in less-than-perfect control of the whole focus group.
Wearing of mechanical components can result in inaccurate operation by lens control electronics.
AF Correction Data
When the AF Processor has determined a new focus position and is ready to focus the lens, it must instruct the lens to move in a certain direction and a certain amount.
The AF Sensor has measured how out-of-focus our point of interest is, so we have a direction and an amount (as shown in Part 2).
The problem is – the relationship between the focus measurement and the amount the lens must be driven to adjust is very specific to the lens. And not just the lens design, but the actual lens – and all its idiosyncrasies – that’s attached to the camera.
So how does this happen?
Inside the lens is a CPU which communicates with the camera through the contacts in the lens mount, and this CPU is responsible for controlling the motor. It also contains a table of information which can take the knowledge we have – the current focus position, the desired focus position, the current zoom setting and the aperture – and convert this to the actual amount to drive the motor in order to obtain our requested focus position.
When the AF Processor has information from the AF Sensor and is ready to move the lens, it requests the amount of drive required from the processor within the lens, and then instructs the lens to move that amount.
The last step in the production of a lens at the factory is calibration (not to be confused with AF Fine-tune or AF Microadjustment). The behaviour of the lens is measured, and information is stored in the lens which describes how this real-world lens deviated from our perfect lens design – this is the AF Calibration Data which is used in the step above.
And again, this information applies to the brand-new lens at the factory – so wear and tear in the lens can make this calibration data inaccurate.
Change in mechanical behaviour can render internal lens calibrations inaccurate
Effects on the Autofocus
Let’s look at how the above issues can affect autofocus.
The above optical and mechanical problems can be categorised into two groups:
- Those that change the angle between the light path and the sensor
- Those that change the focus position and behaviour
Optical Path Offset
A basic expectation of a camera system is that the optical axis is perfectly perpendicular to the sensor, and a subset of the issues raised above can result in an effective tilt in any direction, leading to a slight change in behaviour.
From the above list, the following are potential causes of tilt:
- Lens mount offset
- Imperfect lens mount adapter alignment
- Decentering of lens elements
- Tilting of lens elements
To picture the effect, imagine trying to focus on a brick wall that is at an angle to your camera. Which part of the wall will be in focus – it can’t all be in focus.
Although we’re only talking about tiny movements of the optical path, we’re also talking about tiny sensors – remember, a typical sensor has pixels that are only a few millionths of a metre across. So those tiny deviations to the light have a real effect.
Here’s a view of what the masked pairs on the AF Sensor will see under different conditions:
(If this image makes no sense to you, take a look at Part 2: On-Sensor Phase-Detect for an explanation)
The Far, In-Focus and Near objects each have 3 pairs of images. The top of each is with no tilt at all, the second with 1 degree of tilt, and the third with an extreme 2 degrees.
The important set is the yellow in-focus images. The AF Processor will take the data from the AF Sensor and try to align the left- and right-mask images. For the top row (no tilt), this is easy – they line up perfectly.
But with any tilt, the masked pairs aren’t capturing quite the expected information, and it becomes difficult to determine the offset. The best guesses – highlighted with the green arrows – show changing out-of-focus amounts and directions depending on the amount of tilt.
Roger Cicala, over at Lens Rentals, wrote a post about The Effect of a Decentered Lens on Autofocus, in which he states (the bold highlights are mine):
“We learned a couple of things today. The most important part is that a decentered lens not only doesn’t resolve as well as a good lens, it focuses less accurately. So, the best shot possible with the decentered lens has less resolution than the good lenses, but autofocus is less likely to get you the best shot possible with that decentered lens.
It seems logical when I think about it: autofocus, whether phase detection or contrast detection, requires pixels on a sensor evaluating an image. If the lens has some spherical aberration, decreased contrast, or (especially for phase detection) difference in side-to-side light paths, focus accuracy could well be affected.”Roger Cicala, LensRentals
Focus Position & Behaviour Changes
On leaving the factory, our lens will be precisely and accurately controllable by the camera, thanks to the internal factory calibration based on the brand-new and well-behaving optical and mechanical components within the lens.
As time goes on, various things we’ve seen above can upset how the control of the lens focus may deviate from our expectations:
- Lens element placement errors (along the optical axis)
- Motor wear, leading to changes in drive amounts
- Gear wear, leading to changes in backlash
- Focus element mechanics wear
The end result is that when the AF Processor instructs the lens to move to a certain focus position, the final focus position isn’t quite where it should be.
You might think that the AF Processor will check the position and continue to adjust the lens focus if this happens. And you’d be right for some conditions, although this will result in slower autofocus and potentially a less accurate final focus position.
There are common conditions, however, where the lens is told to move to a specific position and the result is not checked… which means these issues can end up with a significant fixed deviation from the expected focus point.
This is covered in more detail in the next part: The AF Processor.