Vacuum Tube Calculator Schematic

Find a Copy of "Pulse and Digital Circuits" by Millman and Taub, circa

You note you've built multi bit adders with basic digital ICs. Note that these IC's probably contain dozens to hundreds of transistors, so to understand how to implement such a thing, you'll need to try it with transistors and not ICs.

One thing to note: Much logic in the vacuum tube era was done without tubes - diode arrays, transformer circuits, etc. So, you might want to "cheat" and use semiconductor diodes where you can.

Another thing of note: vacuum tube computers were often serial logic. So, you'd create say a full 1-bit adder with tubes, and then shift your operands into it one bit at a time while shifting the result out. Storage, other than registers, was not tubes - more like mercury delay lines, CRT storage, cores, etc.

The gotcha is that if you do it using the same way that TTL or solid- state calculators are designed in textbooks (parallel operation) then it'll use an obscene number of tubes and passives and power for only limited functionality.

A classic science fair project from my childhood involved building a calculator from Strowger switches and rotary telelphone dials. Google if you don't recognize them. The guy who invented them was actually a mortician, amazing that the most complex and scalable switching network in the world for more than half a century was entirely based on his competition with local morticians.

It's unlikely that you can find the Strowger switches now, but if you want to use tubes, use a rotary phone dial as the "input" to your calculator and as the "accumulator and display" use decade counters built from tubes, with neons and/or nixies as the display elements. In the 40's, 50's, and 60's many manufacturers (HP and Berkley Nucleonics are the units I'm most familiar with) built these from the same principles, with flip-flops chained with feedback to give a maximum count of 10, and a very clever count sequence that allowed neon bulbs to serve as the decoding elements. Each decade uses 4 dual triode tubes. Most use a industrial (cathode-poisoning free) version of the

Brent Hilpert has a site that explains the AC-4A internals in terms of logic symbols similar to what you'd see in a 70's era TTL handbook, if you want to compare/contrast.

To feed these from a rotary telephone dial, you probably want to square up (Schmitt trigger) the output of the dial. One decade would probably be enough, but if you have enough parts two decades and two phone dials is quite impressive. None of the clunk-clunk of the Strowger switch but the sequential operation really is good eye-candy.

Tim.

Ok, you have done what is the hardest step, asking for help. That is a big start.

One, what grade are you in? Two, do you have some one to help you? Three, do you have access to a oscilloscope?

A timeline impresses the first judging team, which is usually a parent and a teacher. The second judging team you hit at a local is a pair of professionals, around here its a university professor and a professional in another field, sometimes a nurse or computer programmer who volunteered. They will see the difference between a copy of past work and something original. You may find yourself having to explain it to someone who has no idea what your talking about. But even a non electronics person can see a graph of good data , and that gives them a reason to pass you on.

The kid I competed against when I was young who implemented switch/relay logic and and/xor/nand/flip flop in discrete transistors never made it past the second judge because all he did was build working logic. If he would have tested its performance against say, starved collector current , he would have had something to show and would have made it to the regional.

I got busted at my first county science fair (7th grade) for not having a variable to measure, I did etching of circuit boards. The blinky light circuit working was not enough. I learned a lot, but I just duplicated existing work. The next year I measured the light output of different colored lamp phosphors vs energy input and got to the regional.

No research, no hypothesis, no win. . If you just copy a design and get it working, while impressive, it is not gonna get you past a local or maybe a county competition. If you just build it, and you hit a purist judge like me, and I'll say "This is nice build, nice hardware, glad you learned a lot about theory, However I see no original work here with measurements and I can't pass you on to the next level. "

However if I saw you only could find a pentode and a triode when you needed two triodes, and you found a way to make it work, and explained and documented the difference, then I could pass you on.

So if you do the project, the first step, and it is encouraged these days, is to get yourself a mentor who can help you with your project. Most electrical engineers and physicists will jump at the chance to help you. You need to keep a lab notebook and document your contact with your mentor. If you have a university around where you live, call them and ask for help. You may need to be a bit persistent to reach the right person.

You need a hypothesis, a experiment to do with at least one variable, data tables, at least one conclusion.

So if your project title is something like :"reducing tube count in a 4 bit serial adder using magnetic memory and diodes" or " Speed and Accuracy in a vacuum tube flip-flop vs grid bias voltage", or "LEDS replacing neon lamps in a vacuum tube computer display" you would be on the right track. Those are complex subjects, you may not need to be that complex.

(hint neon lamps are slow, modern LEDS are fast, Take five minutes and search GOOGLE later and find out why ionizing neons are slow. Would replacing the neons with leds in a vacuum tube flip flop work? Are the neons in the anode circuit or not, if so, would it speed up the flip flop if they had leds in 1955?)

There is so much more available to you if you go further then the basic requirements. If you pick and do a project that gets you into ISEF, you get a free airline ticket to a big town, per diem, a week there in a hotel, at least two social events while there, usually a trip to a ball game, a chance at up to 10K in prizes or a possible scholarship, and a shot at a interview for MIT, Carnegie Univ etc

Young people doing stuff like this are rare these days, and there are people who may reward you for it. It is a government goal and a electronics industry goal to get more young people into science. If you do this right, you can get a shot at getting part of your college paid for, and use the project to get admitted to a better school. That is important.

You need to make some decisions soon and get started. For ISEF, the on line registration deadline (ie when you have to have your paperwork signed and a mentor/reviewer assigned) is Feb 28 if I read it correctly. Most districts have their first judging in March. Your running against a clock.

Take a look at this:

Steve Roberts/

Here is a good book on line for looking at counting tubes and indicators used in old vacuum tube computers: (ie dekatrons)

Steve

Oh, yeah, and Steve Roberts very wisely points out, all the complex technical wizardry in the world won't win you a science fair by itself, you need a hypothesis, some measurements, a conclusion, and more important than all, posterboards that explain all this in the context of the scientific method :-).

So if you're comparing calculator technologies from different eras, try proving/disproving Moore's law or your own variant.

Suggested variations of Moore's law: Cost for a two-digit adder vs year of technology. Energy requirements for a two-digit adder vs year of technology. Heat dissipation of a two-digit adder vs year of technology. (Really, it's unlikely a judge will know that "energy requirements" is, up to the tiniest bit of entropy, entirely the same as "heat dissipation".) Note that Moore's law is ALWAYS presented using a logarithmic scale. Do not belittle the logarithmic scale.

The broad technologies you are choosing (vacuum tube logic: 1930's through 1960's. Transistors: 1950's through 1960's. TTL: 1960's and

Intel is a big sponsor of science fairs in the US, so choose wisely whether you want to prove or disprove Moore's law, because if you make it to the national level, Moore himself might show up!

Tim.

Or build a Colossus clone. While the real colossus filled a room, a single function subset might be interesting. colossus was a WWII secret code cracking processor. There is a question if it was truly a computer, but it had one thing you might find interesting, the data input was a paper tape with holes in it. Could you make a vacuum adder using two paper tapes and a readout? That might be easier to find then a strowger switch or dekatons. I like the strowger switch , phone dial driven system.

Steve.

Going slightly off topic for second, didnt a phone dial add two or three counts to each digit? Steve

I hope not. It would be a silly project.

No. Other than the zero being 10 pulses.

--

aioe.org, Goggle Groups, and Web TV users must request to be white listed, or I will not see your messages.

If you have broadband, your ISP may have a NNTP news server included in your account:

There are two kinds of people on this earth: The crazy, and the insane. The first sign of insanity is denying that you're crazy.

That's kinda my point, he needs something much more research oriented then duplicating a old ALU. I have some experience in this, having sponsored/mentored 20 or so kids for ISEF, with many of them winning scholarships and the school taking team prizes at ISEF three years in a row. But its clear from his posts he wants to be a CPU builder, that is his area of interest, so let him try. This is why he needs a mentor and to read the requirements of the competition. A lot of this depends on his grade level.

Until he sets down with someone experienced in this area and sets up a project with a thesis, he's just building another toy. Hopefully he learns from it.

The kids I worked with did things like " Catalytic Polymer Nanofiber Filters for removing Formaldehyde From air" or " Electrospun Polymer Nanofibers for Condensing Oil Droplets in Diesel Exhaust", Or the one young lady who cultured skin cells on nanofiber substrates and got herself a free ride at Rice university. Or the other young lady that got tired of paying for paper and balsa ballet shoes and so borrowed our Instron test machine and fatigue tested the shoes to destruction, then made a polymer reinforced pair and tested those.

Congrats guys, we just scared him off.

Steve

You may be interested in two pages on my website:

To cook up resistor values, consult the plate curves and calculate the plate resistors to give a logic-low level that's just enough to turn off the following tube after the grid divider. Grid divider values are not critical, as long as enough grid leakage is provided (usually under 1Mohm, with a capacitor in parallel with Rg1 to compensate pulse response). Calculate the grid divider resistors for a high-state voltage a bit over 0V and low-state somewhere in the cutoff region (maybe -10V for 12AU7).

Typical values might be, guessing, +V = 100V, -V = -50V, Rp = 5k (I_OL =

Much larger values of Rp can of course be used (principally limited by required grid leakage resistance and fanout demands), and you probably won't be needing such bleeding speed anyway.

Cathode followers make excellent line drivers/buffers, and can "saturate" with less than 20V drop.

See also *The* Tube Clock;

Tim

Analog or Digital?

Not true; some simple but useful calculations can be done using less than 100 tubes.

Nope...

One does not need a Dekatron to count to 10.

adder:

If your knowledge is such that you need to be given that much detail, then you have absolutely zero chance of getting anything this complicated to work. Aim for something simpler.

For more points, he could build a higher-order machine of mechanical design based on the Digicomp 1.

Look ma, no wires.

My gawd! I have a copy of that book.

...Jim Thompson

Are you sure you know what your getting into?

You are obviously looking at a very simplified computer - here's info about just what you're looking for:

"The Rebuild by the Colossus Rebuild Project ... has taken 10 years and has involved over 6,000 man-days of volunteer effort. At a conservative daily rate of £200 this gives a value of £1.2 Million for the Colossus Mk 2 Rebuild.

The Rebuild is now working and can demonstrate most of the code breaking work against the German Lorenz cipher in WW II."

Or Google Colossus ...

The National Museum of Computing obvious have detailed schematics of the type you are looking for but I would suggest you consider "... we rely solely on your financial donations to run the museum ..." when contacting them.

Terry