Microprocessors Running on Air?
There was a time, now forgotten, where the expectation was that computers would run on fluid currents rather than electric currents.
Logical gates can be made in almost any fashion. Today we build logical gates such as AND, OR, NAND, and XOR gates using combinations of transistors. However, this is not the only way to do it. Earlier computers used vacuum tubes instead of transistors. Before that relays were used. One of the first programmable computers the Z1, made by Konrad Zuse during WWII used mechanical logical gates.
Before talking about air-based computers, I want to introduce you more broadly to the history of how computer systems have been built through the ages. Because this story is in many ways about what could have been historical. Silicon-based microprocessors are here to stay. But sometimes it is fun to imagine “what if?”
Mechanical Decimal Number Based Computers
If you go far back into computer history, to the age of steam, you will find the mechanical monsters dreamt up by computer pioneers such as Charles Babbage, who made steam-powered mechanical computers performing their computations using an elaborate system of gears. However, these were decimal-based computers and not built on logical gates, commonly used by digital computers operating on binary numbers.
The Advantage of Binary Digital Computers
Realizing that computers could be more easily built if one adopted a binary number system was a revolution. Konrad Zuse essentially did like Charles Babbage, in that he created what was mostly a mechanical computer. But because he used a binary system, he could employ boolean logic utilizing logical gates such as AND, OR, NAND, etc. This profoundly simplified his computer and allowed him to build a far more capable computer than Charles Babbage. One which could fit inside Konrad Zuse’s modest living room. And it was self-funded.
Charles Babbage in contrast was building a mechanical computer that would fill factory halls and which required the funding of what was then an empire spanning one-fourth of the globe. This shows the profound impact of getting the fundamental ideas right: The radical advantage of going binary over decimal.
That means something like the Z1 could have been built with 1800s technology. If only Babbage had realized boolean logic was the way to go.
While we are talking about alternative history, why don’t we ask ourselves if something even more profound could have happened in the 1800s? Indeed a far more powerful approach to building computers was available with the technology existing in the 1800s. Due to steam power, scientists and engineers of the 1800s were very good at working with gasses and other fluids.
Fluidics Based Logical Gates
Chances are that you have never heard of this technology called Fluidics. It is based on using fluids such as liquids or gasses to perform computations. No, it is not based on moving pistons and valves. In fact Fluidics systems are not mechanical at all. Rather they are based on how a jet of fluid can modify the direction of another jet of fluid.
This principle was discovered quite late when researchers worked on airplanes. However, engineers in the 1800s had everything they needed to build logical gates on this principle.
The benefit of fluidics is much like regular electronics. Without mechanical parts subject to wear and tear they are reliable and robust. You can also build fluidics devices at fairly high density.
Maker novelchip has a project which demonstrates creating an integrated circuit using fluidics, which can be placed on the equivalent of a printed circuit board (PCB).
What Happened to Fluidics?
Logical circuits based on fluidics was in fact gaining ground in the 1960s and there were numerous companies with large catalogs of fluidics devices and circuits from which you could order parts from.
In fact, for a period this looked more promising than electronics. So how come so few have heard about it? Well, you might not know either that the whole city sends messages through pneumatic tubes? That instead of an electric network European cities had compressed air networks driving clocks, motors, and even cooling systems in factories and stores.
Yet a lot of these systems have since been forgotten because they tended to have been only been used for short periods of time. Often a competing technology comes along too quickly. E.g. the vacuum tube never got that widely used before it was replaced by transistors. Thus vacuum tube-based computers only existed for a relatively short time period.
Likewise compressed air system to deliver messages and power only existed for a short time before electricity replaced them. Fluidics suffered the same fate. It came into prominence over a relatively short time period before electronics quickly got good enough to entirely replace it. Although fluid systems have been used to control thrust vectoring systems on airplanes.
Fluidics systems have found a niche in areas where operating temperatures are much higher than what electrical components can handle. In fact, one of the reasons I have had an interest in Fluidics over several years is due to a certain fascination with colonizing the planet, Venus.
Read more: Why Colonize Venus Instead of Mars?
If you read that story you will learn that the huge problem with Venus is the crazy temperatures on the surface. It gets to 460° Celsius. At that temperature it is game over for pretty much any kind of digital electronics, but not for fluidics based devices. This led me to explore what technologies could one use to build an entirely non-electric rover to explore the Venus surface. It would need some kind of computer, communication system and motors which are not electric-based.
It turns out that, if you dig enough around the internet, you can find solutions to all these things: Making a Non-Electric Rover For the Venus Planetary Surface.
In this story, I explain how the control system, vision, propulsion, and communication can all be done with a non-electrical system.
Have Fluidics Computers Been Built?
Sadly fluidics got abandoned before it could get too far but some researchers did manage to build a prototype for a very simple fluidics-based computer called FLODAC. It was built in 1964 and was made up of 250 fluid NOR gates. . It had 4-bit word sizes and a memory of 4 words as well as 4 different instructions:
This was to demonstrate that a program made with the 4 most fundamental instructions for any computer. FLODAC ran at 10 cycles per second. It has however been theorized that clock rates of 10 to 100 kHz are possible. That sounds extremely low compared to an electronic computer which operates at around 3 GHz today (2019). However, that isn’t necessarily as limiting as it sounds. The human brain operates at a measly 30 Hz. Still, the human brain outperforms almost every computer. It has been calculated that the human brain has a processing power of 6 petaflops. That is six million billion calculations per second. Which compares favorably to the worlds fastest supercomputer:
The world’s fastest supercomputer is actually about 30 petaflops. Of course, it cost half a month of China’s GDP to build, and requires 24 megawatts to run and cool, which is about the output of a mid-sized solar power station.
The human does roughly the same with just 20 watts.
How does the human brain achieve this while operating at such low frequency? Due to massive parallelism. Fluidics systems could likewise gain processing power from parallelism. Since you can build them in 3D using simple plastics, you can build quite a lot of channels in a small area.
So as a rule of thumb we could say a fluidics system should be able to run at 300–3000 times faster clock cycle than the human brain. No, it does not imply that we are going to build fluidics systems that can outperform the human brain or supercomputers. But it does suggest that the limits are not necessarily as profound as they seem at first glance.
Further Reading for the Curious
If you think this topic is interesting, you might want to read some of my older articles where I dig into various aspects of air-powered systems. One problem which has occupied my mind is whether you could build a non-electrical industrial economy on another planet. Now, why would I be curious about that? You might want to read my Venus colonization story to understand that better.
If you are interested in Maker stuff, then surely you know about 3D printers. These are extremely versatile in letting somebody easily make almost any kind of part at home. An obvious device for colonizers of other planets to have. Even closer to home, you got the Amish people. They try to avoid electricity but are okay with compressed air-based systems. In fact, Amish kitchens and workshops typically have their tools and equipment running on compressed air.
So the question is: Could you build a 3D printer only running on air?
If you are interested in more of this space stuff, I have an overview page over the various articles I have written about rockets and space colonization.
Read more: Space Exploration and Colonization.
What Happened to Fluidics?
I felt compelled to add this last section as you are easily led to believe by reading this that fluidics is entirely dead. But that is in fact not true. As is often the case with a lot of technology that fails in one area, it finds new use in an entirely different field. The modern version of fluidics is micro-fluidics which an active field of research typically used in the medical industry. For tasks such as analyzing liquid samples, DNA, and other things, the ability to process tiny streams of liquids is an advantage.
However building actual general-purpose computers and control systems for industrial robots, the path the technology was on in the 1960s, has been mostly abandoned. Digital electronics simply do a much better job.
But it might have an interesting application on other planets or to seed the imagination of authors writing in an alternative history setting. We got our steampunk novels and movies. But where are the fluidics punk novels? 😉