How Lens Optical Stabilization Works
The more you understand about how stuff works, the more amazing it is. I had heard long ago that lenses with optical stabilization use “gyros”, but I never gave it much thought. I had heard of “gyroscopes”, but spinning wheels aren’t what we’re talking about. The name gyro has the Greek root gyros, meaning rotation.
A typical lens optical stabilization unit (a Canon lens)
The picture above shows a stabilization unit, which moves a compensating lens group to keep the image on the sensor (and in the viewfinder) from moving. The question to ask is how the lens knows the photographer is jiggling the lens, and what to do about it. In a DSLR, it can’t use the camera sensor to help figure out image movement, since the shutter is hiding the sensor.
Let’s start with something called the Coriolis Effect. Coriolis comes from a French mathematician named Gaspard-Gustave de Coriolis. On a big scale, it’s what causes large-scale weather patterns in the northern hemisphere moving north to have an east-word velocity, and the opposite effect in the southern hemisphere. On a smaller scale, it’s the force required to keep walking in one direction as you try to move from the center to the edge of a rotating merry-go-round.
When a photographer hand-holds a lens, the lens invariably starts to rotate a bit in various directions. This rotation results in the Coriolis Effect, which can be sensed and then compensated for.
A really smart person envisioned a gyro design that could notice rotation by jiggling a weight in one direction and sensing a force that was perpendicular to the direction of that jiggle. As shown above, when the weight was moving up, the force would push the weight to the left. The same weight would get pushed to the right if it was moving down. The forces would all reverse when the rotation switched from clockwise to counter-clockwise. This design is known as a vibratory rate-measuring gyro. It’s common to jiggle these weights at about 10,000 cycles per second (or 10 kHz). The “rate-measuring” here is the rotation rate, usually expressed as degrees per second.
Using MEMS technology (micro-electrical-mechanical-systems), the miniature weight, the springs, a drive motor, and position sensors could all be built on a microscopic scale. Power requirements scale down with the size of the parts being used, so a camera battery could drive this system easily. Even with low-power requirements, cameras will typically turn off the stabilization when you take your finger off of the focus button. This type of gyro design was introduced in 1991. The typical name for these units is “Coriolis vibratory gyroscope”. It finally found its way into lenses in 1995.
The gyro concept shown above can only sense rotation along one axis, so two of them would be needed in a lens to handle the yaw (left-right) and pitch (up-down) axis of potential rotation. A photographer typically wouldn’t be rotating the lens along its optical axis (roll axis), so that motion isn’t compensated for. Typical hand-held rotation rates being counteracted are around ½ degree per second to 20 degrees per second. The center of rotation is roughly the rear of the camera (or the photographer’s eye).
The Analog Devices, Inc. description of their gyro design
The picture above shows a little bit more detail. The “Coriolis Sense Fingers” in the drawing are little capacitors that sense the gap distance between their little parallel fingers as the “resonating mass” shifts left or right, according to the direction of rotation of the device. The tiny signal from these capacitors can be converted into a voltage that varies according to the rotation.
What the “sense capacitors” look like in the silicon design
The Sense Capacitors (via a scanning electron microscope)
The whole silicon design of the gyro
Getting into even more detail, the picture above shows “Comb-Drives”. These little guys are given an alternating positive and negative voltage, to force them to have a net positive or negative charge. The moving “Active Mass” has little fingers that fit in-between these Comb-Drive fingers. The active mass is alternately pushed away or pulled toward the stationary comb drive fingers, since its electric charge is either attracted to or repelled by the comb drive fingers as their voltage is switched between positive and negative.
Comb-drive actual silicon, via scanning electron microscope
The shot above is a close-up of the little silicon comb-drive fingers. They push and pull the “active mass” to keep it vibrating. The “active mass” is suspended on silicon beam “springs” to greatly increase the magnitude of its motion when it vibrates, which enhances the signals produced by the gyro.
The whole gyro needs to keep the “active mass” vibrating back and forth in its “drive direction”. The mass will get a sideways vibration (the sense direction) when the device (lens) is rotated, due to that Coriolis Effect. When a sideways vibration happens, that’s when those “sense capacitors” mentioned earlier produce a signal to indicate the device is rotating, and in which direction it’s rotating. The signal coming from the gyro is typically a low current, which is converted into a digital count that is proportional to the rotation rate.
These little gyros are so small that even air becomes a problem. When they’re manufactured, they have to install them into a package that maintains a vacuum.
How small is that gyro, anyway? Small.
The lens optical stabilization unit needs a separate little gyro to sense rotation about each axis being controlled (e.g. yaw and pitch). Once the stabilization unit gets the gyro signals to indicate that the lens is rotating, then its microcomputer commands little actuators to move the compensating lens group in the stabilization unit to counteract that rotation.
There are other kinds of gyro designs that are much more complicated than the “simple” one I have described. In fact, they can get mind-numbingly complex. Better-quality vibratory gyros are actually able to detect rotation rates of less than 10 degrees per hour. If that isn’t enough to bring tears to your eyes, I don’t know what will. You also probably have MEMS gyros in your smartphone.
The next time you complain about having to spend extra money for a lens that has optical stabilization, just think about the technology that goes into it. And try to imagine the brilliance of the people that invented it.
By the way, in-lens stabilization is generally preferred over in-camera stabilization. With a DSLR, lens-based stabilization lets you see a steadier viewfinder image and it makes it easier for your camera to focus on a non-moving target. One downside, however, is that the moving stabilization optics can make for slightly worse bokeh.
A big thanks to Canon, Analog Devices, Inc. et al. for the visuals used in this article. I don’t yet have my own scanning electron microscope.