Dan MaloneyHi everyone, welcome to the last Hack Chat of 2020! I'm Dan, I'll be moderating again today for David Lovett, aka Usagi Electric, and we're going to be talking about Vacuum Tube Logic, or pretty much anything about vacuum tube circuits.
David, can you kick us off with a bit about how you got interested in vacuum tubes circuits?
Sure thing!
flat panel array antennas are amazing these days. You can fabricate somethat will pick up Blutooth from miles away
Hello everyone
Hi Marc! Welcome back!
My father was an EE (although his profession was a pilot), so there was always a little bit of interest in old electronics in the house. But, growing up in the 80s, I got used to really low voltage stuff (even cutting my teeth on BEAM robotics, which is ultra low voltage stuff). So, I've always had this fear of high voltage, which kept me away from vintage stuff too much. I always thought tubes were gorgeous though, so earlier this year, I decided to finally give tubes a shot, only at low voltages. To my surprise, they worked decently well!
For us non-Japanese speakers, what does Usagi mean (I think I get the Electric part)?
Once I found that I could actually get the tubes to do stuff, it became this goal of mine to try to build something substantial out of tubes at low voltages. There's many times when 150V would have been much easier, but I'm nothing if not stubborn.
Ah, yeah, Usagi just means "rabbit" in Japanese.
Sweet. We have pet rabbits too. Ours are not electric though :-)
Love the logo, BTW -- nice ears
There's kind of a story there. My buddy and I are also huge into classic cars and while out drinking in an Izakaya in Nagoya one night, we came up with this joke idea to create a blog or something called "Usagi Motors" (うさぎモーターズ). It just kind of stuck, so I adapted it a bit for the electronics stuff
Haha, thanks, that was a lot of fun to make.
And Marc, that's awesome! Pet rabbits are fantastic animals! We've had a pet bunny for years now and I would take a bullet for her, haha
I'm not telling my daughter (who has a pet rabbit) or it will inspire her to get another just to name it rabbit in japaneese
Haha, if she needs ideas for Japanese names, I can get my wife involved and you'll have a ton of rabbits before you know it
This conversation is going down a rabbit hole...
so... what sort of logic circuits have you built using low voltage tubes?
now....back to electronics.
Yeah, I was going to ask just how low you've pushed the voltage in some of these circuits. Seems like you use 24 volts a lot -- have you found a floor?
it would be great to look at building some VT computers .... Mr. Carlson's lab has a nice session with a NOS opamp tube
So, aside from the ones I've done videos on (SR, D, edge-D and T flip flops, a half adder and a full adder), I've also done a few combos of those build other things. Like mixing an XOR gate, a D Flip Flop and a Full Adder to build a bit-serial full adder.
Though, I've kind of shelved that idea for now. I may return to it in the future
link to Mr. Carlson's lab on opanmp
amp...
Dan, I've actually done a bit of testing at just 6V, but not much. 6V is great because you just need one supply for the entire thing, but you really start running super tight margins. Swings of just 1V to 2V. It could work with if you build some hybrid stuff, using a microcontroller or transistors to detect the small changes
I mostly stick with 24V because it works out well for running four tube heaters in series
I just watched your video where you talked about that.
I was surprised to see you used prototyping plug boards for building tube circuits, but I find the anachronism delightful. Taking the problem from the other end, how high a voltage do you think you could get away with on such boards?
McChavez, Mr Carlsons video on the OpAmp was great! I actually rebuilt that schematic a bit at just 24V. I managed to get it working pretty well. I actually built two of them and used them to control an SR Flip Flop (like a 555 timer) and it was working decently well!
On the breadboard, I've run 150V so far without any problems. And even at that high of a voltage, I don't think it's all that dangerous even (as a matter of a fact, the only components I've destroyed, I've done so at 24V). I don't know if I would comfortably go higher than 200V on a breadboard though
A vacuum tube 555 -- a video I now must absolutely see
shout out to curiousmarc....love your stuff. as well as the spelling of your first name as I am a Marc as well
Dan, that's something I was working on, but had to take a break because I ran head first into a problem with figuring out the discharge pin of the 555. I just couldn't come up with a good way of doing it with just tubes. I'll revisit it again in the future once my subconscious has had a while to chew on the problem
McChavez: thanks! Marcs with a 'c' unite!
I bet one of you engineers could tell us when things would arc over..as an architect I just hire engineers ....math get scarry
Oh yeah, Marc's stuff is absolutely fantastic! And Marc may not remember, but I actually sent him an e-mail about 6 months ago asking about an old switchboard I had along with my Model 14 Teletype. Marc gets massive respect for taking the time to respond to a random e-mail sent out of nowhere!
McChavez, I try to avoid math at all costs, which is probably not a good thing, haha
Usagi Electric: oh that was you! I had not made the connection
well I do have excel for storing formulas....and as i get better at excel that helps.
Yup, me indeed! Makes sense, totally different usernames, haha
It could be the fact that I spent 10 years being a Japanese businessman, but I use Excel almost daily. It's one of my favorite programs, despite all the frustration it gives me
are there better materials for emitting at low voltages....that were not used in the old days that are available now...I wonder
Excel is the lingua franca of all database and math programs.....i love to hate it
I agree that HV doesn't have to be frightening. Just be careful. I had more 'mishaps' due to HEAT than to voltage! Also, you really don't need much more than 100V or so, unlike old RF/HAM stuff with plate voltages in the several-hundreds of volts.
I'm sure there is, but I'm not nearly smart enough in materials science to figure out what, haha. I would love to build a tube from scratch though, and I have a shopping cart full of supplies on Amazon that I haven't yet took the plunge on.
maybe emitting is the wrong word...
all that and glass blowing.
@MCCHAVEZ I'd bet that with the advancements in materials science and fabrication, that better emitters are indeed possible!
Tomcircuit, heat is definitely a big issue. Fortunately, using those belton (I think that's right) style sockets on the breadboard keeps the tubes pretty elevated, so heat from the tubes themselves hasn't been much of a problem on the breadboard
I'd love to be able to use the tubes that just sit at the swap meets as they are not used for much....there are many of them...just learning
@Usagi Electric (David) - I think I already shared this with you, but for the rest of the class:
https://hackaday.com/2016/05/04/home-brew-vacuum-tubes-are-easier-than-you-think/
Home Brew Vacuum Tubes Are Easier Than You Think
It all began with a cheap Chinese rotary vane vacuum pump and a desire to learn the witchcraft of DIY vacuum tubes. It ended with a string of successes - a working vacuum chamber, light bulbs, glow tubes, diodes, and eventually this homebrew power triode and the audio amplifier built around it.
joking....well they liquid cool gaming computers....let's wrap a water jacket around 400V of DC....what could go wrong?
I haven't actually done much with old 8-pin octal tubes, which are what most of the really vintage tubes are. They're just a little too unwieldy for the digital logic stuff I've been doing. Great amplifiers though!
What's the max switching speed you've had success with, or have you even evaluated?
And BTW, I'll post a transcript after the chat in case anyone needs to refer to anything.
Who said vacuum tube logic?
@Usagi Electric (David) Ugh, didn't even think about tubes on a breadboard. Woof. Maybe some PCB "breakout boards" are in order, to bring neonoval (9 pin all glass - like 12AU7, etc.) to breadboard-friendly footprints. (no, haven't looked yet - somebody has probably already done this!!)
@Ethan Waldo I have actually done a bit of testing with switching speeds. It all depends on where you draw the line on failure. The bottleneck comes in the transition from saturation to cutoff (ie. low output to high output). At anything faster than about 220kHz, the tube can't transition fast enough to keep up with the input, so the tube never actually reaches a proper high output.
Have you tried to fire a Thyratron? How low of a voltage can you go on these?
Hi, I'm Artem kashkanov, And I'm trying to create vacuum and cold-cathode tube computer from soviet miniature lamps with brainfuck instruction set - https://hackaday.io/project/45538-dekatronpc DekatronPC project
@Artem Kashkanov That's a gorgeous SR Flip Flop implementation!
@tomcircuit I actually just soldered some wires to a standard tube socket then bent those around to fit into the breadboard and works brilliantly. Like this:
This is my first logic prototype via usual 6N3P lamps (double triode). I used IBM 604 schemas to create it
The all glass tubes "miniature" tubes were commoditized by TV and post-war (WW2) radios. They tend to be much cheaper than their octal cousins. They also tend to run a bit cooler, too, IIRC
@Artem Kashkanov ...and ferrite core memory? you are brave
@curiousmarc I have done a bit with the 2D21 Thyratron. At 24V it seems to work no problem, but there's definitely a floor on those. The 2D21 has an 8V voltage drop across it, so once you get lower than about 15V, it starts to get kind of useless
But for DekatronPC I want to get 1MHz clock frequency, so long-tail circuits are expected
Yep, I have few ferrite cubes, but they it's too tricky.
yes, I loved @curiousmarc 's videos on those...
@Artem Kashkanov The IBM 604 is probably the best resource I've come across yet. The customer engineering manual for it is magnificent (for reference, this is the manual: http://www.bitsavers.org/pdf/ibm/604/). That's an awesome looking implementation! What logic specifically is it doing?
@Usagi Electric (David) there are 2OR and 2NOR gates.
I need to get my hands on some Dekatrons one of these days. They obviously wouldn't work at low voltages, but they'd be a blast to play with!
@Usagi Electric (David) I also have docs for 650 and 704.
I have the fastest ones - soviet A110, which can do up to 1M clock count.
@Artem Kashkanov Ah, very nice! I cheat a bit with my designs and use silicon diodes for both OR and NOR. I'm mostly using the tubes to invert and amplify, but I have built using the 6AL5 dual diode instead and it works great!
@Usagi Electric (David) For a dekatron logic you may use docs for a Harwell dekatron - I will reuse some ideas from it.
IBM's circuits all seem to revolve around using a multivibrator in different ways, which is brilliant, but extremely hard to keep balanced at 24V. At higher voltages, it works an absolute treat though!
My rule for DekatronPC - no silicon in the core. germanium is allowed, but I want to minimize. I have about 250 double vacuum diodes for a logic, but not sure It's enough for me
That's awesome man! I love the idea of a completely hollow state computer. Do you have a website or YT channel I can follow along?
https://www.youtube.com/watch?v=8PEWXZFjvvA
This is my previous project - reed relay computer. This is only one ENG video.
Awesome! I'll give it a watch after the chat!
Speaking of relay computers, this is a fantastic resource as well: https://www.relaiscomputer.nl/
Thanks! You can follow me on YT: https://youtube.com/c/ArtemKashkanov/ but it's totally in russian :)
Artem - even though I have much trouble following the translated YouTube captions, I love your videos!
I'm gonna have to put YT's auto-translater to the test, haha
ditto me too
@curiousmarc it's very nice to know that you know me and watch me :) I really like your videos!
While we have Russian tube folks on here. I might need some of these miniature Russian tubes to repair this great spy radio. Where can one find these?
@Usagi Electric (David) Unfortunately, the is no update for MERCIA project for a few years:(
That thing is amazing and I want it...
I messed up, picture of radio above my post
@curiousmarc so cute device... Which one tube do you need?
Yeah, it's a shame the Mercia hasn't been updated for so long. It was my primary resource when I built my 10-bit binary relay calculator and my hexadecimal relay calculator
Don't know yet, have not started the restoration. Other problem is that the manual that comes with it is in Russian! Collab opportunity maybe?
I mostly get new tubes as a gifts from my subscribers.
@curiousmarc that spy radio looks like it's straight out of a movie!
@curiousmarc Sure! It seems that I have mostly all types of tiny tubes. 1.2V and 6.3v
Artem: it might take me a while before I get to it, but if you could contact me directly through the email on my about page so we can talk directly /about
Oh, while we've got @curiousmarc here, I had a question about decapping ICs. I really
Ah, damnit, hit enter at the wrong time
As for the main talk. I want to achieve 10-15MHz for a one logic gate - to get 1MHz clock.
2MHz is easily achievable, but if I use open-anode circuit - there are the problems with the tube closing process and output capacitance charge.
So my current selection is long-tailed circuits, as Alan Turing used in his ACE
Want to decap a tube ;-) ?
I made computer tomography of the tube...
Let me find the link :)
I've been doing a lot of research on a specific chip and wanted to try to get a die shot of one, but I don't have a microscope or anything. Suppose a solid macro lens on a DSLR could get the job done, or does it really have to be a microscope
He already did, I think
https://www.youtube.com/watch?v=6DM1qBgtxrs
Haha, yeah, I've destroyed a few tubes
So, the impetus behind this is trying to learn how a specific chip works at the logic level (this chip to be exact https://en.wikipedia.org/wiki/Motorola_MC14500B)
so....are there vacuum channel transistors out there or are they just a web page? Now that would be a tiny tube
Usagi: just send the chip to me if you want. Me or Ken can take a picture.
If you are interested in what's inside the tube WITHOUT it's destroying - follow the links:
That would be epic! I'll send you an e-mail!
Oh that's one for Ken. He will have it reverse engineered in no time!
It's a pretty simple chip, I've got it mostly figured out (and converted to NOR gates), but there's some quirks I'm still trying to wrap my head around
That mess of lines and gates is my best stab at reverse engineering it at the gate level, which gives me a great starting point
(although the actual chip is probably mostly NAND gates or something)
Oh, looks more like a bit slice ALU than a processor! Very interesting! I'm sure Ken will be all over it and delighted to help.
@MCCHAVEZ I've read about those vacuum channel transistors a couple of times, and I really want to believe they're still being developed. They have the potential to be totally game changing. What I've read says switching speeds into the THz level!
@curiousmarc It's a very weird chip indeed! Here's the block diagram for it:
My goal (as a bit of a spoiler for the upcoming series of videos) is to rebuild this chip from tubes
@Usagi Electric (David) 1-bit microprocessor... I used the same one, but on a discrete logic in soviet rare PLC.
Nice! The MC14500 I think was actually designed to replace industrial ladder logic stuff and is probably really similar to the one you used!
oh, I'll save up my popcorn budget for that series!
@Usagi Electric (David) - you will use normal tubes or tiny tubes? My current calculations show that I need about 1300 tubes for my DekatronPC.
For a 1-bit processor, you will ned much less. but few hundred also.
My goal is to use purely 6AU6 tubes. I can save a lot of tubes by using silicon diodes for the NOR gates, which keeps my total tube count to around 100 to 150 ish
Both of you are going to need to beef up your electric service ;-)
That's just for the 1-bit processor though, which doesn't include a program counter. There's a lot more headache in the future when it comes time to build the supporting circuitry for it, but one step at a time!
Dan, no joke! I think with about 150 tubes, I'll be using something like 300W of power just for the heaters
I'm starting the development from logic emulation on FPGA device. My idea is to implement the whole logic on SystemVerilog and them - start building blocks and modules, connect them to the emulator and replace emulation parts with the physical ones. At the end I can swith-off my emulator and my computer would be pure vacuum.
So, you can use the same approach - implement vacuum tube logic and IP counter on emulation side, and then - replace it.
Bump that up to the 1300 tubes for Artem's dekatron PC, and it's at least 2,500W of power!
My great problem on relay computer - that I couldn't turn it on for a year.
I want to say 10kWatt?
Haha, you're power company must love you
And that's a good idea using the FPGA! My theory was to build supporting systems with the actual chip, then implement those into the tube version as it grows
In Russia we have very cheap electricity. Just 0.05$ per 1Kwh
Usagi: Awesome project and very intriguing chip. A one bit ALU hooked up to an instruction decoder. No address bus or program counter in sight. I wonder how they do jumps.
In later editions of the RCA RTM (a.k.a. "the bible") you can see that there were a whole slew of tubes that there created in the late 1950's and early 1960's for TV sets, that have a relatively high level of "integration" - three triodes in one package, etc. That might cut your tube count (and your heater consumption) quite a bit. Also, these weird TV tubes are often very, very cheap these days.
@curiousmarc they do not jumps. Its for PLC. Just calculate logic function in the each line of a program. lines of logic and states. no jumps.
Artem: that makes sense
@curiousmarc Jumps are actually one of the reason I was looking into decapping the chip. They have a "skip next instruction is RR = 0" instruction, but I'm having a headache coming up with the best way to implement it. They also have generic flags you can set and I think the idea with that was that things like jumps could then be handled off chip with the flags
@tomcircuit I will mostly use double triodes and double diodes, but there are no double penthodes.
Welp, that's our allotted hour gone by in a flash again. I have to say a big thanks to David for coming on and sharing his infectious enthusiasm for tube circuits with us. As I told him before, I really dig the idea of being able to build with tubes -- something about it just seems so sturdy and solid. I'm going to have to take the plunge one of these days.
Thanks Dan for a fantastic year of Hack Chat presenters.
@tomcircuit I actually have one of those triple triode tubes, it's awesome! The primary reason for using the 6AU6 is that it's dirt cheap. I can usually find them for about 20 to 30 cents a tube if you get a big enough lot of them. I've been slowly stockpiling them for a while now
Thank you so much Dan! This was an absolute blast!
Everyone should feel free to stay on and continue the chat as time allows. of course. But for now, thanks to David for his time t
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.