Analog Computing, back to the future?

TLDR; The 1960s were the heydays for analog computer, but they are making a comeback driven by a broad range of users from musicians to artificial intelligence researchers. Some companies are bringing to market affordable devices that will put these back into schools, just like the Raspberry Pi did in 2012 for digital computing and IoT.

School days

In college, I remember hearing about analog computers. Sounded very ancient. It was something to the effect that "Vanevar Bush invented the first electro-mechanical analog computer". That was it. I wouldn't have had any clue as to the coolness of analog computers, if it was not for all the time I spent in the physics lab. It was a fun place with a laser, a Jacob's ladder (high voltage travelling arc), and it had the tools for me to debug my robots.

This particular lab also had a lot of older equipment on the shelves, hiding in corners. This is not surprising as the school was started in the 1920s ( Collège Jean-de-Brébeuf ). Still, finding a 1960 Heathkit EC-1 (I think that was the model - see picture below) at the top of a case a whole quarter of a century later, was quite a thrill. We were able to do a bouncing ball simulation on it. That was my first interaction with an analog computer. Unfortunately there was no option for a student to go and buy one (the few remaining manufacturers at the time were selling expensive laboratory machines, and all but disappeared for decades).

Analog vs Digital

Digital computers operate on discrete values through discrete steps. Your laptop, your desktop, your smartphone, your raspberry pi, all are digital computers. They work with zeros and ones. Put enough of them together and you can represent any discrete value. Continuous values are simulated. They also require memory for storage and need to run at high speed, as they cannot do complex tasks within one cycle. They also tend to be quite power hungry. And they are quite vulnerable to cyber threats. They do offer higher precision than analog computers, but can still suffer from instability with small variations of very large values.

Analog computers in contrast operate on continuous values through continuous processes. Their active electronic circuits are an analogy to the systems they model. Some features and uses:

• They can apply complex calculations in real time
• They are inherently parallel (do not suffer from Amdahl's law)
• use barely any power compared to digital computers
• They are much less vulnerable to cyber threats
• They can do differential equations, integrals, addition, subtraction or multiplication of continuous signals
• They are well suited for control and feedback applications (for motors, cruise control etc)
• They have been used to do simulations in physics, economics, aviation (more generally, dynamic systems) and research in mathematics.
• They have also been used with modular and semi-modular synthesizers in music, where modules are controlled by continuous voltages.
• For the past few years, there has been increasing research in using them for neuromorphic simulation (neurons, brain research, AI, neural networks).

How about a computer that is both analog and digital? These are called hybrid computers and can combine the best of both worlds. This is another reason why analog computers are making a comeback.

The Analog Thing

Earlier this year, I heard about a new analog computer from Analog Paradigm coming to market at less than 300 euros (EUR269 for teachers, students and institutions). That was quite incredible. I ordered one on the spot, even though it wouldn't ship for a few months. The few basic analog computers that had been on the market were several thousand dollars!

I felt that if they could pull this off, they would have a huge impact, and we'd soon see these units in schools and in the hands of hobbyist. Without a pipeline of people who know how to program them, analog computers wouldn't be able to make a comeback.

With parts shortages, the few months grew longer, but I finally received my unit (picture above) in August 2022. It came with patch cables, a daisy chain cable and a first steps guide (pdf).

The top section is the patch area. This is how you "write" your program, by creating the circuit that is analogous to the system you are simulating. You connect and chain the devices (integrators, summers, inverters, multipliers, comparators etc) by connecting the patch points with patch cables with banana plugs (mini 2mm type). In the above picture, I've patched it to simulate radioactive decay (dollar bill for scale). The coeff knobs are your parameters or variables, while the display can be used to view a coefficient value or a result. For any serious work there are multiple outputs in the back that can be sent to a display (an oscilloscope), an X/Y analog plotter, a soundcard, a mixer or a synthesizer module, depending on the goals. It is also possible to use external analog inputs. The back also has a master out header, a minion in header (for daisy-chaining several THAT) and a hybrid header (for hybrid computing).

What's next?

Before I jump into the data applications and hybrid computing, I'll first use it for music. I use modular and semi-modular synths, and I've been building my own synthesizers, modules and pedals for a while now, so this will be a lot of fun. I'll do a follow up article to some of the specific applications to my music (you can buy some of it on bandcamp and listen to many albums on soundcloud)

We had dismissed the analog computer too quickly, and there is no changing the past, but the future is yet unwritten.

Francois Dion

@f_dion