Hackaday Prize Entries: Inventing New Logic Families

One of the favorite pastimes of electronics hobbyists is clock making. Clocks are a simple enough concept with a well-defined goal, but it’s the implementation that matters. If you want to build a clock powered only by tubes and mains voltage, that’s a great skill tester. A relay-based timepiece is equally cool, and everyone should build a Nixie tube clock once in their lives.

For [Ted]’s Hackaday Prize entry, he’s building a clock. Usually, this would be little cause for celebration, but this is not like any clock you’ve ever seen. [Ted] is building this clock using only diodes, and he’s inventing new logic families to do it.

Using diodes as logic elements has been around since the first computers, but these computers had a few transistors thrown in. While it is possible to make AND and OR gates using only diodes, a universal logic gate – NANDs and NORs – are impossible. For the computers of the 1950s, that means tubes or transistors and DTL logic.

For the past few years, [Ted] has been working on a diode-only logic family, and it appears he’s solved the problem. The new logic family includes a NOR gate constructed using only diodes, resistors, and inductors. The key design feature of these gates is a single diode to switch an RF power supply on and off. It relies on an undocumented property of the diodes, but it does work.

Although [Ted] can create a NOR gate without transistors — a feat never before documented in the history of electronics — that doesn’t mean this is a useful alternative to transistor logic. The fan-out of the gates is terrible, the clock uses about 60 Watts, and the gates require an AC power supply. While it is theoretically possible to build a computer out of these gates, it’s doubtful if anyone has the patience to do so. It’s more of a curiosity, but it is a demonstration of one of the most mind-bending projects we’ve ever seen.

You can check out a video of the diode clock below.

The HackadayPrize2016 is Sponsored by:

46 thoughts on “Hackaday Prize Entries: Inventing New Logic Families

  1. Very well done, clean build. Curious to learn more about the NOR gate constructed using only diodes, resistors, and inductors. Didn’t the IBM 7030 do something similar in 1961 by alternating the “definition” of one and zero at each stage?

    Two diode anodes as inputs, tie together the cathodes and you have an OR gate. Then you define the outputs as being negative logic and you have a NOR gate. Lastly, add some amplification every few stages to compensate for voltage drops.

    Is this similar?

    1. Alternating logic levels is exlained there : https://en.wikipedia.org/wiki/Emitter-coupled_logic#History

      “ECL was invented in August 1956 at IBM by Hannon S. Yourke.[10][11] Originally called current-steering logic, it was used in the Stretch, IBM 7090, and IBM 7094 computers.[9] The logic was also called a current mode circuit.[12] It is also used to make the ASLT circuits in the IBM 360/91.[13][14][15]

      Yourke’s current switch was a differential amplifier whose input logic levels were different from the output logic levels. “In current mode operation, however, the output signal consists of voltage levels which vary about a reference level different from the input reference level.”[16] In Yourke’s design, the two logic reference levels differed by 3 volts. Consequently, two complementary versions were used: an NPN version and a PNP version. The NPN output could drive PNP inputs, and vice versa. “The disadvantages are that more different power supply voltages are needed, and both pnp and npn transistors are required.”[9]”

  2. “Creating a NOR gate without transistors”, “a feat never before documented in the history of electronics.”

    Assuming this is actually a serious claim and not somehow intended in jest, transistor free NOR logic was being done and patented back in 1962. Here is one example using tunnel diodes. Still a very neat project but I am not sure the sensational hyperbole from Brian is entirely warranted here.

    https://www.google.com/patents/US3255361

    Transformer trigger tunnel diode nor logic circuit

    Publication date Jun 7, 1966
    Filing date Nov 29, 1962

    1. Well, you need to look into the Laws of Form. All the rage for a brief time in the 1980’s with claims that people had developed gates with half the transistors of current minimal gates, etc. I had to read the very stuffy paper (reverse engineered from switching circuits into an academic sounding paper of the silly variety) for my boss at the time, Jef Raskin. Then explain the why and why not. In particular, why none of it was coming true :-)

      I’m expecting a resurgence at any time.

          1. Nope, that blob could be kilometres into the underlying rock by now, it will flow down at the rate it can melt the rock and that diminishes as it dilutes itself, you just end up with a huge column of hot rock so you could at some point drill it and use it as a geothermal heat source for power generation, LOL! Obviously in a closed loop and it is hot in more than one way.

  3. That is really cool! As I read the article I thought ‘wonder how many of those diodes are LEDs’ and then wham! Awesome first video frame hits me.
    The skill it takes to say no to transistors! Great gallery too man.

  4. It is in theory possible to create a NOR gate from a small set of conductors of varying lengths. [ inductors for the larger scale application. ]
    Using high frequency, resonance and cancellation together with propagation delay to steer the signal and implement the logic.

  5. Oh that is awesome. At Uni 2 decades ago one of the lecturers said it was possible to make amplifiers using only diodes and passives. I asked how and he summarised it was done by turning the input signal into pulses. Gain is pretty much a requirement for logic. I read a schematic for generating fast pulses just last week that turned out to be diode switched, that used two diodes feeding into a constant current sink, when one side (the gate drive) turns off the other side (the output) turns on and emits a negative pulse. Some similarity to ECL now I think about it.
    /* Also DiOHd */
    Possibilityofownsuccess–;

  6. ” While it is possible to make AND and OR gates using only diodes, a universal logic gate – NANDs and NORs – are impossible.”

    You probably means it is impossible to create an “inverter” using only diodes and resistors. After all NAND and NOR are only AND and OR with inverted output. The key component here is the “inverter”.

      1. To build a NOR or NAND gate you need first an inverter! To build logic circuit of any complexity one can opt either for AND or OR gate but the inverter is not optional.

          1. Anything stateful is obviously impossible with ROM. Let’s ignore that.

            Instead, we can consider adding two 128-bit numbers together to form a 129-bit sum. You could do this with a few hundred diode logic boards. It would fit on your desk and add a few thousand numbers per second.

            Now, let’s do it with a ROM. The inputs to a diode ROM have to be one-hots, and you’ll need 2^256 of them. The ROM also has to have all the results pre-calculated – and you need a diode for each possible 1 in the output (or for each 0, there are equal numbers of them). Half of the outputs of all sums are 1. So, you need (129/2) * 2^256 = 7.5 * 10^78 diodes. There’s enough matter in the observable universe for about 10^80 hydrogen atoms. I’ll let you have this one – say you can make a diode with matter equivalent to about 13 hydrogen atoms – then you could build such a ROM (you’ll undoubtedly find a clever solution for wiring them all together). But, you’ve just used up all the matter making diodes; neither you nor I still exist to see the thing work.

          2. Yes you have to ignore anything stateful to answer the question, because I was specific when I specified functions. :-) Obviously a ROM can’t do RAM, registers or clock generation. So assuming we use latches etc. only where they are actually needed then your 128-bit 2 input function “problem” is moot as it can be done sequentially with 128 sliced into bytes, as is actually the case in the smaller solid-state MCUs, right? Those memory units can also be done with transformers, delay-lines or other exotic methods like Williams tubes, nothing new there. Assuming tape for storage the simplest universal Turing machine I could work out only needs a handful of devices to hold single bit states, assuming you have tape or drum storage using magnetics. You either go toward hugely complex logic trees (massive ROMs) operating in parallel which are fast, or very sequential logic that is compact be needs more states tracked, which is slow, or a combination of the two which makes the most of the strengths of both with the resources available to you. The tech you use makes a big difference, e.g. an optical computer may be able to operate in the terahertz range but be very hard to implement using much if any parallelism and the ROM logic is the very opposite case, except you can encode ROM logic in holograms and that can make things very interesting….

          3. The problem with implementing the address decoder for your diode ROM as another diode ROM is that the second diode ROM still needs an address decoder. In fact, if you think about what the diode array in such an “address decoder” ROM would look like, it’s obvious that the diodes are simply passing the outputs of its own internal address decoder to the outputs of the ROM. Congratulations – you’ve managed to make an address decoder using only some diodes and a working address decoder of the same size.

          4. No, the real problem is with making latches. Any completely parallel circuit will have the same issue as a diode ROM, except the dROM uses the least amount of components. You can only get things smaller by “spending time”, by serialising part of the logic, and that requires latches to hold states.

  7. What did you mean by “undocumented property of the diodes”? The fact that when they’re biased, they’re RF shorts, and when unbiased, they’re RF opens?

    That’s not undocumented. That’s their IV curve.

    Also, I call shenanigans! He’s got transistors there: in the 74AC74 that’s on the DDL02 RF power supply board – and that’s not like the regulators/etc. that could be avoided using only diodes (e.g. a Zener).

    1. The undocumented part is that the 1N4007 functions like a PIN diode for switching RF, exhibiting power gain – a small DC signal can switch a larger RF signal. The IV curve doesn’t alone doesn’t give you that: the slow reverse recovery time of the diode comes into play, allowing the “reverse” excursion of the RF signal to pass before ceasing conduction.

      Spoiler alert: follow the wires far enough, and you’ll find a nuclear power plant as part of the supply, too :-)

        1. I’ve seen that zinc negative-resistance thing before. Pretty neat, but you’re right, extracting enough power could be a problem.

          I discussed several alternatives in the original project write-up, including GUNN or IMPATT diodes (for an all-diode solution):

          https://en.wikipedia.org/wiki/Gunn_diode
          https://en.wikipedia.org/wiki/IMPATT_diode

          IMPATTs look like they might produce some decent power. I also discussed mechanical solutions like spark gaps and Alexanderson alternators for producing the required RF:

          https://en.wikipedia.org/wiki/Spark-gap_transmitter
          https://en.wikipedia.org/wiki/Alexanderson_alternator

Leave a Reply to YGDESCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.