What the Steam Engine, AI and Phase-Modulated Light have in common

There is one concept guaranteed to add intellectual voltage to almost any moderately ambitious presentation — rather like charging an electric car at a Supercharger instead of plugging it into a domestic socket: the Kondratiev wave.

These long economic cycles are supposedly triggered by technological breakthroughs that reshape not only individual industries, but entire economies and societies. The steam engine, electrification and information technology are classic examples.

The concept’s only slight weakness is that its genius tends to reveal itself retrospectively. Until recently, the sixth Kondratiev wave was commonly associated with sustainability, healthcare and the transition to a green economy. Then artificial intelligence arrived with considerable force, and the theory was promptly expanded to accommodate it.

But while we debate the technologies destined to transform civilisation, smaller revolutions are taking place below the radar. They may not change the whole world, but they can overturn almost everything within a particular industry.

Phase light modulation - PLM - may be one of them.

The physics behind it is not new. The ability to manipulate the phase of light has occupied scientists and optical engineers for a very long time. What is new is that high-resolution, high-speed phase modulation is becoming practically usable, industrially accessible and programmable.

And once an optical function becomes programmable, some surprisingly solid assumptions begin to move.

From moving glass to moving information.

The dream of many optical engineers is to control the phase of light freely and in real time. Why? Because phase determines how light propagates, interferes and forms an image. Control the phase, and you can influence where light travels, where it comes into focus and how optical energy is distributed in space.

For centuries, we achieved such effects by arranging physical optical elements: lenses, mirrors, prisms, apertures, gratings and mechanical scanning systems. If we wanted to change the optical function, we usually had to move an element, exchange it or redesign the system.

There is an interesting parallel with artificial intelligence. The theoretical foundations of neural networks existed long before today’s AI boom. Their breakthrough required sufficient computing power, data and suitable hardware. In the case of PLM, one of the decisive hardware enablers is Texas Instruments.

Since the 1990s, TI has produced Digital Micromirror Devices, better known as DMDs. Millions of microscopic mirrors arranged on a semiconductor chip generate the images projected in cinemas, meeting rooms and countless industrial systems.

In a conventional DMD, each mirror tilts between two positions. Light is directed either into the projection path or away from it. Millions of tiny mirrors switching at high speed create an image with millions of individually addressable pixels.

A Phase Light Modulator works differently. Instead of tilting, its micromirrors move vertically like microscopic pistons. By changing the optical path length, each mirror alters the phase of the reflected light. With several discrete positions available for every pixel, the device can create a programmable phase pattern across its surface.

This may sound like a relatively modest mechanical change. Optically, it is anything but modest.

The chip no longer controls only whether light reaches the image. It controls how waves of light interact after leaving the chip.

A conventional DMD essentially asks: “Light or no light?”

A PLM can ask: “Where should the light go, where should it converge, and what optical wavefront should it form?”

This adds a spatial dimension to the projected image. Using computer-generated holograms, the system can generate holographic images, shift the focal plane without moving the projection lens, steer beams and redistribute optical energy within a three-dimensional volume. In other words, we are beginning to replace some optical mechanics with mathematics.

What is it good for?

This is normally where enthusiasm encounters two rather reasonable questions:

What can we actually use it for? And when will it matter commercially?

The first is easier to answer.

PLM can be used to shape, concentrate and direct coherent light with a degree of flexibility that would otherwise require a combination of lenses, diffractive elements and mechanically moving components.

Consider industrial 3D printing. Instead of illuminating an entire flat layer with a largely uniform intensity distribution, a phase-modulated system could distribute optical energy more selectively. Focus, intensity and potentially the three-dimensional exposure geometry become variables controlled by software.

In maskless lithography, the same principle could enable dynamic exposure strategies, programmable focusing and more sophisticated control of the energy delivered to the photoresist. Optical functions that are fixed in a conventional projection system could become adaptable to the process, the substrate or even an individual structure.

Beam steering is another obvious field. Today, directing a laser beam often requires galvanometer mirrors, polygon scanners or other mechanical systems. A PLM can steer and split beams without moving the entire optical assembly. That is relevant to optical communication, ranging, metrology and both civilian and defence applications.

There are also less spectacular but commercially very tangible examples.

Imagine projecting a vehicle logo sharply onto the ground beside a car. With conventional optics, the image is in focus at a defined distance. The road may be lower than the pavement, however, and the kerb may sit somewhere in between. A programmable optical system could adapt the focus to the actual surface without mechanically moving the projection lens.

That may sound trivial next to holography or laser communication. In engineering, however, “trivial” applications have a habit of becoming very large markets.

What happens if we turn the idea around?

Things become even more interesting when phase modulation is incorporated into the receiving path of an optical system.To be precise, a PLM is not itself an imaging sensor. It does not measure the incoming light. But it can manipulate an incoming wavefront before that light reaches a camera or another detector.That distinction matters and so do the possibilities it creates.An adaptive optical system could compensate for wavefront distortions caused by atmospheric turbulence, imperfect optics or changing operating conditions. Instead of replacing a physical optical component, the system could calculate an appropriate correction pattern and display it on the modulator.Such technology could support high-resolution imaging through disturbed media, optical communication over long distances, microscopy and certain astronomical or remote-sensing applications.It could also become part of computational imaging systems in which optics and algorithms are no longer designed separately. The optical front end modifies the incoming light in a series of known ways; the detector records the resulting signals; and software reconstructs information about the observed object.This opens the door to new approaches to three-dimensional imaging, object recognition and ranging. It does not mean that a PLM simply replaces a lidar sensor. A lidar system still needs a detector and timing electronics to measure the returning photons and determine their time of flight. But PLM may alter how light is emitted, received, filtered and spatially encoded - potentially simplifying some functions while enabling entirely new ones.The broader idea is more important than any single application:The optical system is no longer a static arrangement of components. It becomes a programmable interaction between hardware, wavefronts and algorithms.

So when will PLM become commercially relevant?

Soon, but not everywhere at once.The hardware has reached a point at which companies can begin serious application development. The remaining challenges are less glamorous than a hologram floating in mid-air, but considerably more important to industrialisation.Computer-generated holograms must be calculated quickly enough for real-time applications. Algorithms need to translate a desired optical result into phase patterns that respect the physical behaviour and limitations of the device. Speckle, unwanted diffraction orders and optical efficiency must be addressed through a combination of algorithms, illumination design and Fourier optics.Thermal management also matters. High optical power, fast electronics and tightly integrated optomechanical assemblies do not become easier simply because the optical function is programmable.Then there is the question of system architecture. A PLM chip alone is no more a complete product than an AI processor is a complete autonomous vehicle. It requires illumination, electronics, high-speed data processing, mechanics, thermal design, optics, control software and - above all - an application that benefits sufficiently from the additional freedom.This is why PLM is unlikely to replace conventional optics in one dramatic sweep. It will enter through applications in which programmability solves a problem that cannot be solved economically, compactly or quickly enough with conventional components.Initially, these will probably be specialised systems. But specialised systems are where many platform technologies begin.

Why am I writing about this?

Because at IN-VISION, we are already working with this technology. As an optical systems company and design-house partner of Texas Instruments, we had the opportunity to experiment with PLM at a very early stage. We develop driver electronics and optical subsystems that make these devices accessible beyond a semiconductor laboratory.

That experience has made us enthusiastic, but not uncritical.

We have seen the impressive results: freely programmable focus, software-controlled beam steering, structured distributions of optical energy and light fields that change without any mechanical adjustment. We have also seen the work still ahead: diffraction artefacts, speckle reduction, algorithm development, optical efficiency, thermal design and the integration of all these elements into a stable industrial system.

This is not magic, although some of the first demonstrations look remarkably like it.
It is engineering.

And engineering is the process through which a fascinating physical effect gradually becomes a reliable product.

PLM may never receive its own Kondratiev wave. It probably will not be mentioned alongside the steam engine or artificial intelligence in economic history books.

But within optics, it has the potential to change a fundamental question. For centuries, optical engineers have asked: “What combination of physical components will create the function we need?”
In the future, they may increasingly ask: “What optical function should we programme?”
That is a smaller revolution than AI. But for optics, it could be quite a large one.