Your one of the few people keeping this knowledge alive and sharing it! Keep it up!!

@UsagiElectric

3 жыл бұрын

Thank you so much! Tubes are so cool and a great way to learn electronics I think. Hopefully, I can inspire some people to get into them and build some cool projects!

I'm building a "Unity for Atari 2600 game development", but compared to me, you are a GOD. I'm just amazed by your knowledge.

@UsagiElectric

2 жыл бұрын

Thank you so much, but honestly, I'm just "faking it until I make it"!

@M3MB3Rrr

2 жыл бұрын

@@UsagiElectric We are all like that. I always say that I'm good with Fortran 90, but since I had not done any of it since november, I should look at my older codes to remember everything again. Some months ago I wrote an Adlib emulator in Python, but I don't really remember how I had done much of it. :D The only thing I'm really good at is understanding the 2600, but since that console is my hobby for 6 years, I would be mad at myself I I would be bad at it. :D I'm into old mainframe / midrange workstations, so I'm really amazed by your work, and I guess we can agree that tubes are beautiful. I have some old tube radios, but that's all. :D

It will get more and more interesting to see it all work together, the more parts you build. Especially in high impedance systems like vacuum tubes, there are a lot of things that might become a problem in the end. But this far you definately deserve respect for getting it going. :) Another intersting thing might be to see if the system can be optimized later on to reduce the total amount of tubes. Maybe you are able to reuse some parts of the system for another task if their original task can't be executed at the same time? Greetings from Germany, Michael

@UsagiElectric

3 жыл бұрын

Thank you! You're absolutely right that as it grows in size, it gets harder to predict how certain aspects of the machine will interact with other bits. My hope is that with buffers in the right spots, everything will go smoothly, but it's going to get really interesting with the clock circuit and timing issues! I could drop the total number of tubes in half by using dual triodes (particularly something like the 6DJ8 or 6CG7), but those tend to be much more expensive than the 6AU6. But, it'll be interesting to see if there are other avenues for optimizations in the future!

This is soo underated. Building a vacuum tube computer has always been my dream, thanks.

@UsagiElectric

3 жыл бұрын

Thank you for checking the video out! It's a pretty massive undertaking, but I think I've got a solid gameplan and I've ever got some interesting ideas for a good program counter and memory, so hopefully it'll all come together in a package that can actually compute some stuff!

So super cool! Loved watching that power up and the VFD's are a great touch!!

@UsagiElectric

3 жыл бұрын

Thank you! The camera had a hard time capturing the light show when powering up, but it sure did look awesome!

I can't wait to see it fully assembled and working! Lately I got intrigued by the vacuum tubes as part of digital logic and began to search stuff about it. And now I'm watching you, building a vacuum tube computer! Nice work, awesome way of telling your experience, and thank you so much for documenting this tecnology so well!

@UsagiElectric

3 жыл бұрын

Thank you so much! I can't wait to see it fully assembled too, haha. As it gets larger and larger, we're definitely going to run into some power supply issues. My current 24V supply can give plenty of current, but the inrush current draw of the heaters is ridiculous, drawing massive current until the heaters warm up. So, I'm going to have to look into how to overcome the inrush current problem soon. Building something this large out of vacuum tubes is illuminating all sorts of super interesting challenges I never would have thought of! Thank you again for the excellent praise!

This is amazing :) subscribed!

@UsagiElectric

3 жыл бұрын

Thank you so much!

Looking good. I am enjoying the heck out of this series! all good wishes.

Good

@UsagiElectric

3 жыл бұрын

Thank you!

Beautiful.

@UsagiElectric

3 жыл бұрын

Thank you!

Awesomeee

@UsagiElectric

3 жыл бұрын

Thank you!

This is amazing. Instant subscribed :D

@UsagiElectric

3 жыл бұрын

Thank you so much! It means a lot to hear that people are enjoying the content!

Here is a neat effect. Connect Not Q to D and then fee the output to the next D. If I remember right, you will get a counter if it is connected right. IT is no wonder why ENIAC took up a 100 foot long room, can you image the space and heat over over 17,480 tubes?

@UsagiElectric

3 жыл бұрын

That is an excellent way to make binary counter! Though, I'm not sure if the D Flip Flop design I use here would actually be capable of doing it without issue. When I used this design in the proof of concept we ran into all sorts of weird racing problems, which we shouldn't get with this design. I was able to rectify that by redesigning the D Flip Flop to a master-slave (primary-secondary) design, which should work great setup as a binary counter as well. That starts to get pretty tube intensive for large counters though (8 tubes per D Flip Flop means an 8-bit counter is 64 tubes!). Although, the bigger issue is how to integrate that into the design as the program counter. If I want to decode the binary count into individual signals, that's a decoder that has 256 tubes in it! I still haven't thought of a good solution for the program counter just yet, but I'm thinking paper tape loops are the best bet. At first, I had no clue why or how the first computers were the size of rooms and used thousands and thousands of tubes, but I can totally see it now. I'm trying to build the most watered down and simplified computing circuit I can think of and we're still looking at several hundred tubes! Also, even with just 32 tubes for the instruction register, I can feel the heat radiating off it. I can't even imagine the level of heat something like ENIAC put out!

@stphinkle

3 жыл бұрын

@@UsagiElectric I read somewhere that ENIAC took special fans and air conditioning units just to keep the room cool enough. It also consumed about 150,000 watts of electricity.

@UsagiElectric

3 жыл бұрын

@@stphinkle I can believe it! I can’t even imagine how hot it must have been in the room with that machine, even with hardcore air conditioning. And, it’s crazy to think that most likely, about 90% of all that power was just to run the filament heaters!

@stphinkle

3 жыл бұрын

@@UsagiElectric It you think about computing history and that the average microprocessor today has the equivalent of billions of tubes in just that one chip of silicon, it is pretty amazing.

@UsagiElectric

3 жыл бұрын

@@stphinkle Oh yeah, absolutely! Even within the microprocessor world, things have changed so much it's hard to wrap your head around. Ken made an excellent post comparing the ARM1 processor from 1985 to the Apple M1 processor from last year, and it's mind boggling to think of the advancements! Edit: Just realized I forgot to post the link to Ken's Twitter post! twitter.com/kenshirriff/status/1327021630636212224

I am amazed at the construction of this computer with tubes.I have a question: what tubes do you use to make it, 12AX7?

I am not sure that the uneven heater pattern is acceptable with a system intended to be robust. It might be worth the effort to swap tubes around so that each string of 4 heats evenly. The heaters that glow brightest may be the first to fail.

One day it'll run Tunix

@UsagiElectric

3 жыл бұрын

Not gonna lie, I actually googled "tunix" before I figured it out, haha!

You using TELEFUNKEN Space Charge vacuum tubes, may it might be possible to have the heaters connected in parallel

@UsagiElectric

3 жыл бұрын

I'm actually using kind of a weird mix of tube manufacturers here: Sylvania, RCA, GE, etc. But, they're not specifically low voltage space charge tubes like could be found in late model tube car radios. They're just run of the mill 6AU6s, which usually expect about 100V to 150V on the plate I believe. It could be possible to run the heaters in parallel, but I don't actually have a second 6V power supply in use here. The only two power supplies I'm planning on using are a +24V and -12V supply, so I have to run the 6V heaters in series to get the proper voltage across them. The upside is a simplified power supply setup, which should make life easier as this thing starts to grow to ridiculous sizes!

Very neat. Out of curiosity, what's the power drawn by this register?

@UsagiElectric

3 жыл бұрын

Thank you! Oooh, power draw is an interesting thing to calculate here. 99% of power draw comes from the filament heater in each tube. I’m using the 6AU6 and the heater is rated at 300mA at 6.3V. Since I’m running at 24V, I can run four heaters in series and get a 6V drop across each one. That’s slightly under-volting each heater, but it actually saves some power with no real noticeable impact on performance. Four heaters in series at 24V should draw 0.285mA of current, which is 6.84 W of power. There are 32 tubes used in the instruction register, which is eight groups of four. Which should be about 2.2A, or 52W of power. Now, it should be noted that 24 of the tubes are setup as groups of four powered from 24V, and the remaining 8 tubes are setup as groups of two powered from -12V. So, the 24V supply is providing 41W and the 12V supply is giving the remaining 11W. Of course, that’s just when it’s running. The inrush current draw is actually enough that my 24V 15A power supply gets upset and thinks there’s a short, so in the future I’m definitely going to have to think of some kind of slow turn solution. That’s a phenomenally large amount of power for something so simple, haha.

@grotsoftsadventuresinelect8868

3 жыл бұрын

@@UsagiElectric Ouch! I suspected it wouldn't be pretty. When I was at Uni back in the 80's one of my colleagues built a 4 bit CPU out of LS TTL chips. It ended up a nightmare finding a couple of very large PSUs for a student project.

@UsagiElectric

3 жыл бұрын

@@grotsoftsadventuresinelect8868 Yeah, as this thing gets bigger and bigger, the power supply is going to become more and more of a headache, haha. Fortunately, with the abundance of 3D printers, there's a lot of 24V power solutions out there. And, I think with a thyratron, a big timer capacitor and a massive power resistor, I can setup a sort of time delay that allows the heaters to start warming up a bit and come up in resistance before letting the power supply do it's full thing, helping to battle the inrush problem. That's awesome that y'all built a 4-bit TTL CPU! I imagine that was a ton of fun to figure out, especially in the 80s! It's easy to forget how amazing of a resource the internet is when it comes to researching things. I wouldn't even dream of undertaking this project (or even building a TTL CPU) without all the reference material I've been able to collect over the past year.

I think if you use higher anode voltage, maybe a cathode follower is not necessary. And im curious about what you are going to use as memory of the computer. Anyway, that will be a splendid project!😁

@UsagiElectric

3 жыл бұрын

Thank you very much! Most definitely a lot of my headaches would be solved if I had a reasonable anode voltage! But, I like the idea of a low voltage vacuum tube computer and how accessible it makes vacuum tube electronics to the average enthusiast. Plus, it's fun to just hilariously misuse components, haha. Memory is going to be a tough one! There's going to need to be three different types of memory I think. First is program storage. I think for this one I'm going to use paper tape since it only needs to be read only and there's a lot of feasible ways to tackle that. Second is general purpose registers. For these, I think I may end up just having to bite the bullet and build my registers out of tubes for ease of integration into the machine. Third is random access memory (ie. memory much larger than a general purpose register). Whereas the registers will probably be 8-bit, I would like around 64-bits of RAM. That's still mega low, but enough that building it from tubes isn't really possible. The obvious choice is core memory, but that carries with it a lot of difficulties as well. The current option I'm exploring is actually light based RAM using VFDs and photoresistors. Here's a test I did just today actually: imgur.com/gallery/yz4yWDY I still have a long way to go, but the deeper I get into it, the more interesting it gets!

Hey I would like to commend you on your work! I would also like to know about your VFD displays? and your PCB software. Thanks for a great channel

@UsagiElectric

3 жыл бұрын

Thank you so much! The VFDs I'm using here are little Russian IV-15 displays, they're a copy of the DM160 VFD, which I believe was the very first commercial VFD. I ordered them off eBay for decently cheap, though I had to wait a month for them to arrive as they came all the way from Russia. The PCB software is DesignSpark PCB (www.rs-online.com/designspark/pcb-software). I quite like it, though it's not as popular as some other programs, like Kicad. I can't really vouch for any of the auto-routing features, as I hand route every one of my PCBs. This is mostly because I've learned how our Bridgeport EZTrak cuts boards and have kind of memorized good and bad practices specific to the mill when laying out a PCB. Also, I find it super therapeutic, haha.

@Sam-jw1yn

3 жыл бұрын

@@UsagiElectric thanks for feed back, do you use the free version or the 499?

@UsagiElectric

3 жыл бұрын

@@Sam-jw1yn No problem! I use the free version. I haven't come across any limitations so far that have warranted upgrading to the paid version.

@Sam-jw1yn

3 жыл бұрын

@@UsagiElectric Thanks again! You should do a video on the PCB software and your CNC setup. Thanks Sam

@UsagiElectric

3 жыл бұрын

@@Sam-jw1yn I actually did some earlier videos about the software and CNC setup, but they were some of the first videos I ever made, and so definitely could use to be redone, haha. If you're curious for now though: kzread.info/dash/bejne/e2acx6pulrvairA.html kzread.info/dash/bejne/eGZ6tLBuqpW5Z5s.html I think I may just remake these for the MC14500 Breadboard CPU series I'm doing simultaneous to the tube CPU build. Since eventually that build is going to move over to PCBs, it'll be a good opportunity to talk about the setup!

So will it play GTA IV?

Wonderful. But some of the filaments look a bit over powered. I love the new Russian VFD's.

@UsagiElectric

3 жыл бұрын

Thank you! I noticed that some of the filaments were brighter than others as well and was slightly concerned. So, I did actually checked the voltage drop across them, and each filament is seeing right around 6V. I think that since it's essentially just a random smattering of different tube manufacturers (GE, Tung Sol, Sylvania, Philco, etc.) that even though the filaments have the same specs (6.3V and 300mA), some glow brighter or are more visible from a tube construction standpoint than others. And I couldn't be happier with those Russian VFDs, they look epic!