Outside of writing, I have an interest in technology. Although I’m not the most skilled technician, I’ve always enjoyed hackathons and science fairs. Recently, this video by James Scholz rekindled my fascination with older technology. Specifically, display technologies. The video lists his monitor as the Dell P780, a CRT released in 1999.
The robust display box fit perfectly in the composition of Scholz’s desk, so I decided to see if I could get one myself. After searching my local Facebook Marketplace and Craigslist, I failed to find anything I could use. Although CRTs were dominant only a few decades ago, the popularization of newer display technologies like LCD led consumer production to cease globally. The unique supply chain necessary to create one of these is no longer fit to manufacture on a mass scale, and the demand for them is almost entirely from niche communities.
Explaining CRT conceptually
We start with the goal of displaying a rapid series of images on a screen. To do this, we know that an image can be created by scanning rows in sequence. Fig. 1 demonstrates this principle with an array of dots. When a dot coincides with a bright part of the desired image, illuminate the dot correspondingly. When it coincides with black, leave it dark. Through illuminating dots in sequence from left to right and top to bottom, an image is formed.
Increase the speed of Fig. 1 using the included buttons to make the image visible. (At higher speeds, the image begins to flash, which may be harmful to those with epilepsy. Proceed with caution.) As scanning gets faster and faster, the image appears static. To implement this principle in a real-world display, you might first try making an array of lightbulbs and orchestrating them to turn on and off at the correct times. This is fundamentally how jumbotrons and large-scale LED signs work, but each individual light emitter is too large to be useful on a desk.
Therefore, your next instinct would be to stop thinking of each dot as a light emitter and start thinking of them as targets of a focused beam. If you’ve read Regression Models and Spiders in Virtual Reality, you’d be familiar with this problem. If you have a directed line of light but want to trace out an entire two-dimensional area, you can simply move the end of the beam from left to right, and downwards once each row is finished.
Unfortunately, the technology to move a line of light quickly enough was not yet available in the era of the CRT. With this limitation in mind, we go with the electron instead of the photon. An electron can be curved with a magnetic field and light up a screen by exciting phosphor, so the same scanning principle will still work to make images.
Now we’ve arrived on a final conceptual design: Some way to emit a focused beam of electrons, a mechanism to move that beam in a scanning grid pattern, and a screen that lights up in the exact spots where electrons collide with it. This layout is illustrated in Fig. 2.
On the left is the illustrated electron emitter. These electrons move towards the right in a beam. In the middle lies the beam manipulator, which uses electromagnetic fields in order to bend the beam towards the desired endpoint. Finally, the screen sits at the right. When the electron beam hits the screen, the point of intersection emits visible photons. When the beam is moved quickly enough and the brightness is adjusted in sync, the effect demonstrated in Fig. 1 manifests and images can be displayed.
Implementing the concept
We first need some free electrons to emit. Unfortunately, electrons tend to stay bound inside atoms most of the time. The solution is adding thermal energy until electrons come off, a process called thermionic emission. We first run electricity through a filament, then hold rare-earth materials very close to it. These materials take relatively little heat in order for electrons to break free, and the hot filament provides enough to create a small cloud of electrons. This electron producer is called the cathode.
Now that we have a small cloud of electrons, the next step, according to the diagram, is pulling them with great force towards the right. Since electrons are particles with negative charge, we use a positively charged metal plate placed closer to the screen. This metal plate has its electrons stripped from it, making it attract electrons. The particles accelerate towards the right in our diagram, and we use a washer design such that it has a small hole where electrons are pulled into a focused beam. The attractive metal washer is called the anode.
Now with a focused electron beam, we need a way to bend it for tracing out the scan pattern. We can use copper coils with current running through them. When done in the right pattern, the beam will sweep horizontally and vertically across the bounds of the screen.
If you built this out, you’d notice very quickly that the air itself interferes with the electron beam. In seemingly empty space, there are numerous gas particles such as nitrogen and oxygen colliding with the moving electrons. To solve this, we simply encapsulate our entire setup in a glass tube with a vacuum pump. Trace gas molecules are sucked out, giving our electrons predictable paths.
After the electron beam passes the anode, you may be concerned that its attractive force now acts to pull electrons backwards. We solve this by making the front of the setup equally charged. Using a coating called Aquadag, electricity passes through, charging the front of the tube to the same voltage and also completing the circuit. This way, the electron isn’t pulled towards a single direction and simply continues along its path.
In order to operate this machine, roughly 20,000 volts are required for charging the anode properly. This kind of voltage is usually pretty inconvenient to generate by stepping up wall-socket voltage. We also need a way to pull the beam back after it finishes a row without retracing the line and muddying the image. The flyback transformer solves both of these problems.
Recall that we used a copper coil with current running through it to create a magnetic field which bent the electron beam. After the beam finishes scanning a row, cutting the current to the coil suddenly using a transistor generates a large voltage spike. We then pour this voltage spike into a capacitor that charges extremely quickly. We then discharge it into the coil again, but this time in the opposite direction. This allows the beam to horizontally snap back to the left almost instantly.
All while this is happening, a separate coil in close proximity is affected by the sudden change (electromagnetic fields don’t require direct physical contact) and has its own induced voltage spike. This voltage spike allows us to step up a much higher frequency voltage, allowing us to use a more compact transformer. Putting all of this together, we get the layout illustrated in Fig. 3.
Now enclosing everything is the glass tube which maintains the vacuum. On the left is the cathode, which is surrounded by a local cloud of electrons. This cloud is pulled into the anode, which focuses it into a single beam. Surrounding the tube are coils which deflect the beam and target a specific point on the screen.
The benefits of CRT
The electron beam can turn off entirely, unlike an LCD which works using a flat backlight and blocking it with a layer in the front. This makes CRTs have deeper blacks than LCDs, which greatly improves the contrast possible. Also, since LCDs block light by changing the state of a liquid crystal, a process that takes time, they have pixel response times. CRTs on the other hand, receive a signal and directly display it with the electron gun.
Compared to the OLED, the CRT’s blacks aren’t quite as deep, and the OLED also achieves near-instant pixel state changes. Many of today’s gamers simply use OLED displays, and they work perfectly for almost every use case. While ghosting is an existing issue, most industry fixes make it negligible for long-time OLED users.
Retro gamers appear to be the demographic that benefits most from authentic CRT displays these days, since older games were designed for them. The resolution fits natively since CRTs are analog, the scanning pattern sometimes helped with displaying dithered color gradients, and some games intentionally designed around visible horizontal scanlines.
The chassis of these displays also simply look well-designed and I can appreciate their form factor despite LCDs and OLEDs being thinner and more portable.
The downfall of CRTs
For larger televisions, CRTs simply needed more depth to project a single electron beam over a larger screen. 40 inch CRTs could weigh all the way up to 304 lbs, while LCDs and OLEDs could maintain the same slim depth regardless of how large they were.
In the early-to-mid 2000s, the economics of semiconductors made creating LCD displays more economical than CRTs, so most manufacturers were incentivized to switch completely to LCD. The production of CRTs shut down, and the creation of the required parts didn’t continue. If you want to build a CRT today as a fun science project, you’d have to acquire the vacuum tube, assemble an electron gun, and put together countless other subsystems.
The most important parts of a CRT are no longer mass produced, so even enthusiasts can’t just create new ones. This is why there exists only a secondhand market for these displays now, and each unit only gets more worn down with time.
The modern-day CRT
After looking into what it’d take to make my own CRT monitor and concluding it to be practically impossible, I stumbled upon a conceptually similar technology: the Laser Phosphor Display (LPD). A company called Prysm makes them in very large sizes, some over 200 inches, mostly intended for advertising and business use cases.
The LPD essentially goes back to the conceptual genesis of the CRT and uses modern technology that makes it possible to replace the electrons with photons. With laser emitters, we pipe light straight to the screen and create the scan line pattern using mirrors.
Again, if you’ve read Regression Models and Spiders in Virtual Reality, you’d be familiar with the problem of moving a mirror fast enough. A physical reflective metal has so much mass that moving it back and forth is a battle against its inertia, which doesn’t get us to our desired frame rates. Instead, we can use a thin light-reflecting film that barely has any mass and oscillates at high frequencies.
Building this as a project would be fairly difficult and you’d have trouble acquiring a sufficiently fast-moving mirror, but the technology is actively manufactured today. Additionally, YouTuber bitluni created a small functioning prototype in this video. Although the display is only a few inches across and is single-color, a sufficiently dedicated technician could feasibly build a 19-inch desk unit fit for daily use and implement color through additive light mixing (since we have red, green, and blue lasers).
I plan to construct something of this sort once I have sufficient resources, and perhaps this will become a popular enthusiast project not too far in the future. That’s all for this piece. Zai jian.