Vacuum Tube Computer P.16 - Building a Full Adder for the ALU
ฝัง
- เผยแพร่เมื่อ 9 ก.พ. 2025
- In this episode we continue work on our ALU, this time adding in the “Arithmetic” part! We explore what a Half and Full Adder is, then we go ahead and build up the Full Adder that will be used in the machine. And then I go on a complete tangent talking about output displays, hexadecimal, 7-segment VFDs and so on. This one’s kind of all over the place, but it sure was a fun one to make, haha.
Check out the previous episodes on Half and Full Adders here:
• Let’s Build a Vacuum T...
• Let’s Build a Vacuum T...
If you want to support the channel please hop over to Patreon:
/ usagielectric
Also, we now have some epic shirts for sale!
my-store-11554...
Timelapse Music:
Artist: Shnabubula
Title: Chrono Trigger "Trigger, Please"
ocremix.org/re...
Intro Music adapted from:
Artist: The Runaway Five
Title: The Shinra Shuffle
ocremix.org/re...
Come join us on Discord and Twitter!
Discord: / discord
Twitter: / usagielectric
Thanks for watching!
Chapters
0:00 Intro
1:42 Half adder design
3:59 Half adder testing
5:16 Full adder design
7:24 Building the full adder
9:04 Hmmm. How to test?
10:33 Testing the full adder
12:28 7-segments to Decimal
14:10 Why are we talking about this?
15:59 Outro
16:25 Bunny!
Giant bonus point for the VFD display output! It improves so much on both the craziness and the beauty factor. Which in my book are the only two measures that count ;-)
I'm right there with you! Looks and insanity is the whole ethos behind this build, haha.
@@UsagiElectric May I encourage you to keep adding a few more of these debugging displays around your CPU so we can follow what it does (or pretend to)!
@@CuriousMarc The more VFDs the better! I'm totally on board with that line of thinking!
Great stuff! The VFD was a nice touch. Those little dot type VFD's are very unique as well. Maybe a neat memory project :^)
Thank you!
Those little dot VFDs combined with some LDRs I think will end up working alright for a type of crude SR flip flop, which would save me a ton of tubes on implementing memory (and hopefully be a bit easier to manage than core memory). Plus, light based memory just sounds cool, haha.
Very cool. Probably the only tube computer built by a hobbyist. Keep up the excellent work! Looking forward to the next series.
Thank you so much!
Surprisingly, there's a few other hobbyist tube computers out there! The ena computer recently came to my attention and it's a brilliant 8-bit design that seems to work really well. Also, Artem Kashkanov on TH-cam is currently building up a tube computer that's going to be very, very cool too!
Your tube build and James Sharman's pipelined CPU build are the two most impressive things I've seen on the net in a long time.
Quite disgracefully I'd never heard of James. I looked him up and immediately subscribed. Very cool project. Thanks for the pointer.
To be mentioned in the same vein as James Sharman is humbling (but also feels amazing)!
James' build is one of my all time favorite builds and I pretty much drop everything to watch the newest episode everytime he uploads. He's been a massive inspiration, and is all-around just an awesome guy!
@@vincei4252 Ooh, glad we could get you over there! James' build is mind blowingly good, and he's a really great guy too! He has a Discord for his build too if you're interested in chatting with him about it (I think he links it in the description of his newest videos).
@@vincei4252 No disgrace in not bumping into random corners of the Web, 'tis a big place, but it is indeed an honor to turn anybody on to James' channel. His whole process is completely inspirational.
@@UsagiElectric Thanks. Looking forward to your next installment.
Nearly or maybe my favorite YT channel :)
Thank you so much! It's awesome to hear that the videos are still interesting and fun!
Bloody hell, that is fantastic. BTW, I love the new intro.
Thank you so much!
And thank you for the praise on the intro, I spent like two full days putting that 11 second intro together, so it's a huge relief to hear it was worth it!
@@UsagiElectric I was definately worth it. It's realy smooth and slick.
Great work as always! Keep pushing forwards as sometimes in the middle-part of any project it all seems too difficult and time consuming, but this project's end-game is definitely going to be worth it!
Thank you!
I kind of hit that wall about two weeks ago, I was finding it hard to get out there and do the leg work to finish the project. But, seeing the Full Adder light up the VFD with decimal numbers lit that fire again and I've been cutting new circuit boards on a daily basis lately!
Still love seeing you cut PCBs on a big mill. 😍
This brings back a lot of memories of the digital logic labs at university. Though we never got to play with valves 🤣
Sometimes I get a little jaded and forget that I'm using a massive Bridgeport to cut 0.2mm deep into a PCB, which is kind of like the equivalent of using a rocket launcher to drive in a nail! Though, it's still always mesmerizing to watch it draw out the lines on the board.
I wish I had taken a digital logic class in University, I think it would have been massively fun!
@@UsagiElectric exactly, it's a brilliant demo of just how accurate those machines are. "Hand me my sledge hammer! I need to drive in a thumb tack"
this is so much awesome one day I will try build those projects using regular components such as transistors and LEDs, it would be a great introduction to electronics.
Thank you so much!
I've actually started working on something similar using the 2N3904 a bit in my spare time. Kind of like a miniaturized version of this one (though it's still going to be very big). I really like the idea that I can clearly follow the bits as they move through the system, but also, a part of me really wants to see what kind of new issues I run into building a transistorized version of the tube machine!
Don't wait forever.. its not as insane or hard as you may think, and it is super fun :)
@@skeezixcodejedi You're totally right, it's been massive amount of fun working on the transistorized design!
I've already got the instruction register and decoder figured out as well as parts of the ALU and the IEN and OEN registers. I'm planning on ordering the PCBs from JLCPCB for this one, so it's been particularly fun trying to build within the constraints of their cheapest PCB size as well as making designs that can be used in multiple places to take advantage of the minimum order of five.
I'm pretty excited about it!
@@UsagiElectric I use elecrow (been awhile mind you) and they had 10* 4" pcbs (now 15?) for $10 sort of thing going on, and I think its cheaper now; just insane. I found it easier to just design and order boards (and pay shipping) and have the in a week, and at 10+ copies.
I've been itching to set up a mill to do one sided boards for myself, but just another machine I don't have room for :P
@@skeezixcodejedi I've had good luck with JLCPCB in the past, and like Elecrow, they're so ridiculously cheap, it makes you wonder how they stay in business. Problem is, I'm way too impatient to wait the one to two weeks for my PCBs to get here. When I finally commit to a PCB design, I want it right now, haha.
Love the new intro!
Thank you!
It took about two days of solid work to get that 11 second intro done, so it's a huge relief to hear that y'all like it!
I look forward to new videos in this series
Thank you so much!
I'm really not much further ahead than what is shown in the videos, so I'm pretty excited to see what's going to happen next too, haha.
NICE
Thank you!
new intro?
very cool!
Thank you!
I spent a solid two days working on the new intro, so it's a massive relief to hear that it's not a total flop of an intro, haha.
Wow-a FULL Adder was made by a MAD Adder-lol! I love your channel and all your projects David! PS: I have a Panasonic clock radio I got for Christmas '77 or '78 that has VFD's & it STILL works well!
I love your channel and your videos … 😂 I wish I could volunteer to work at your garage
Thank you so much!
There's so many projects going on in the garage, I can barely keep up, but it sure is fun!
Easy wah to remember gate equivalence is that if you have all inverted signals going into the inputs, you can remove them all, flip the base symbol (AND to OR, OR to AND), and invert the output. This helps one see that a NOR with all invertered inputs is the same as an AND without invetering any inputs.
I really enjoyed this video. The only thing is that in my earlier comments I forgot to mention how gargantuan this project will become at the end of it all.
You broke the tube count and complexity of the original half added down which was the sensible thing to do. But buffers were still need in the end for a clean high and low output, that is an issue with transistors and tube based computers, but not relay based system?
But I am happy that you can share some of your knowledge with us.
Thank you!
It's going to be a massive project indeed! The processor will have right at around 180 tubes, and I'm aiming for about the same number of tubes for each of the remaining three sections (memory, program control, and I/O). So, in total, the tube count should be somewhere between 500 and 800-ish!
It should be noted though that my judicious use of buffers is actually due in large part to my use of low voltages. If I were running the same +150V/-100V that IBM used in the 604 tube computer, I could actually run a smaller plate resistor and get much better "fanout" properties, eliminating a lot of the buffers. Still, buffers are indeed needed at times, even on the high voltage tube computers (or even transistorized machines), but as you noted, relays don't need them.
And this is for a few reasons. The first is that there is no voltage drop across a relay's contacts, which is something that isn't quite possible with tubes or even semiconductors. The second is that relays are essentially connecting or disconnecting source voltage with with other relay coils through the circuit, meaning there's practically no use for resistors. Relay computers and systems are really awesome and can do some amazing things. Definitely check out Jeroen Brinkman's computer, it's absolutely amazing: www.relaiscomputer.nl/
Nice video! For full 4bit to 7-segment if you use inverted outputs in a diode rom you can get it down to 34 diodes! It's the de multiplexer bit that I imagine will use lots of valves.
Thank you!
Well now, that sounds interesting indeed! I'm going to have to hit you up on that one and see if we can't find an optimal place between tube count and diode count!
Will you eventually implement at set of registers for your creation? Memory, storage? Theoretically how slow would this thing be, max clock speed? Mips speed?
That's my goal!
One of the reasons I chose the MC14500 architecture as a starting point is because it's designed in a way to offload memory to other parts of the entire system. What I'm building here is just the processor, but ultimately, I plan to also build up memory, program control and input/output ( i.postimg.cc/pdBCC70m/Whole-Computer.png ).
I'm working on trying to figure out memory now behind the scenes, and I'm running into some serious headaches (mostly with addressing and control), but the goal is to have a 6-bit memory address to access 64-bits of memory. 8-bits will be for external input, another 8-bits for system related stuff, and the remaining 48-bits as general purpose storage (with a specific 8 of those 48 directly linked to output).
Program control is the next thing to tackle after that, and it's also going to be... difficult. Most vintage computers would read programs in off a tape or punch card and store the program locally in memory. But, memory is brutally difficult to build and programs will have to be extremely long due to the 1-bit nature of the computer. Son in this case I think I'm going to have to sacrifice clock speed for simplicity, and have the program executed directly from tape. So, the actual clock speed of the computer will be dictated by the tape speed. Which means it'll be super slow! My current thinking is to use both left and right tracks on a standard 1/4" audio tape to store the program and clocking data. Ultimately, I'll need a 10-bit control word, and it comes down to how dense I can pack the data per inch of tape. If I can get the full 10-bit control word in 1-inch of tape, and then I run the tape at 7-inches per second, that gives a clock speed of about 7Hz. Which is a whopping 7 instructions per second, haha. So, uh, real slow.
The program control section is still up in the air, and personally, I would like to see a clock speed quite a bit faster than that. I would be happy with about 50Hz or so. That should be enough to do some pretty basic computing things, but shouldn't stress my design too much. I still have a long ways to go before I get there though!
Love it!
Thank you!
That's pretty cool. I'm eventually going to build a frontpanel bus display for my z80 computer that uses those vfd tubes. I am planning to implement a binary to 7segment hex decoder with a GAL, and then drive the segments in the tube with a darlington array. I guess that would be too cheaty to use in a tube computer, though eh? ;D
Thank you!
Not necessarily cheaty, especially considering all of my PCBs have been designed on a veritable supercomputer and machined by a 3-axis CNC machine, haha.
Your build sounds awesome though! If you have any pictures of it, I would love to see it, I love checking out other's homebrews!
One of these days I definitely need to get around to building up a proper 8-bit machine from the ground up. And, I think I have the perfect project for something like that in mind. I just so happen to have a spare ADDS data terminal from the Centurion Minicomputer that I most likely won't end up using, so it would make a perfect data terminal to connect up to a homebrew 8-bit micro!
YAY a new video
YAY VFRAM
Yup! Time to start thinking about memory now that the processor is nearing completion!
Have you thought about using core memory for the future memory part?
I have thought a bit about core memory, but the difficulty comes in the control logic for it. Where core memory really excels is in larger memory sizes, however for the relatively small amount of memory I’m planning on, the number of tubes required to get core memory working correctly would be about the same number of tubes as needed to just build the memory directly from vacuum tubes.
For this build, I’m probably going to split the memory into two: working registers and RAM.
For the working registers, I’m planning on building two general purpose registers, one output register and one input register. The GPRs and output register will probably be built around a slightly modified SR flip flop like this:
twitter.com/UsagiElectric/status/1404454767548583939
This means two tubes per bit, so 16 tubes per 8-bit register. The input register is only there for external input, so I can simplify it or even offload it entirely, saving some tubes.
The RAM will be quite a bit more difficult. I’m thinking I’ll probably have a 6-bit address, which means 64-bits of memory, 24-bits will be used by the registers, 8-bits for input, and the remaining 32-bits for RAM. Implementing ram without using a colossal number of tubes is going to be tough, even just multiplexing and demultiplexing is going to use quite a lot. So, I’ve been experimenting with using VFD displays as memory cells as you can see at 14:54.
This essentially uses a VFD and an LDR to create a feedback loop that latches the VFD on. A simple positive pulse sets it and negative pulse resets it. This could allow me to build the remaining 32-bits of RAM while saving around 60 tubes. And if I decide to expand to 128-bits or even 256-bits of memory, this setup should expand pretty easily.
But, that’s just where I’m thinking at the moment, that’s all very likely to change in the future!
@@UsagiElectric Understandable about core memory. Your VFD+LDR idea reminds me of mercury delay line memory, sounds great!
That vfd driver is so cool!
Can you do a googol counter like Look mum no computer did but with tubes?
Thank you so much!
Interestingly, I think something like that could definitely be done! It would be massively tube intensive though, requiring about four dual triodes per digit. A googol has 101 digits, so you would need 404 tubes, which is a massive number, but it's possible! HP had a really brilliant design for counting 0 to 9 in decimal using binary and displaying it for their frequency counters. I went in to a bit more detail on one of these units in a previous episode ( th-cam.com/video/WrBPVVKvWfE/w-d-xo.html ), but these would be perfect for it!
@@UsagiElectric don't worry i wasn't being serious! I just think it's amazing how you come up with all these things. I've been experimenting with tubes for 2 years now and i know how little useful information there is about doing these kind of things. Keep it up man :)
@@stefandumoulin1872 Thank you so much!
And while I wasn't planning on building a Googol counter, I do think it would be awesome to build either binary, decimal or hexadecimal counters based on HP's design! They have a really compact and brilliantly engineered design that someday I would like to borrow to build something neat, I just haven't decided on what yet.
You're right though, there is precious little information out there for using tubes for things other than audio, but tubes are super versatile and can be coerced into doing all sorts of fun stuff, you just have to pretty stubborn about it, haha.
@@UsagiElectric a counter would be amazing :)
VERY cool. Where are you getting the tube socket pins you're soldering into the PCB? For program storage for the completed system, I'll recommend DIY delay line, DIY magnetic drum, or DIY core memory to be period correct. 😁
Thank you!
The sockets pins that I solder in are just 1mm PCB sockets that I get on Mouser. These in particular are Harwin H3161. Then I just machine out a custom foot print on the PCB for them and solder them in. It's not the best option as rigid mounts for tubes can cause thermal stress and may eventually cause the tube to crack, but I haven't had that issue yet with any of the boards I've built.
For memory, I would love to do a magnetic drum, but man that would require a redesign from the ground up to better work with it! I'm not planning on having much memory available as I don't really think it'll be necessary given the architecture. A program that fills up the entire memory would have to be immensely long, so I think that'll be our primary limitation.
I’m probably going to split the memory into two: working registers and RAM.
For the working registers, I’m planning on building two general purpose registers, one output register and one input register. The GPRs and output register will probably be built around a slightly modified SR flip flop like this:
twitter.com/UsagiElectric/status/1404454767548583939
This means two tubes per bit, so 16 tubes per 8-bit register. The input register is only there for external input, so I can simplify it or even offload it entirely, saving some tubes.
The RAM will be quite a bit more difficult. I’m thinking I’ll probably have a 6-bit address, which means 64-bits of memory, 24-bits will be used by the registers, 8-bits for input, and the remaining 32-bits for RAM. Implementing ram without using a colossal number of tubes is going to be tough, even just multiplexing and demultiplexing is going to use quite a lot. So, I’ve been experimenting with using VFD displays as memory cells as you can see at 14:54.
This essentially uses a VFD and an LDR to create a feedback loop that latches the VFD on. A simple positive pulse sets it and negative pulse resets it. This could allow me to build the remaining 32-bits of RAM while saving around 60 tubes. And if I decide to expand to 128-bits or even 256-bits of memory, this setup should expand pretty easily.
For program control, well, I don't have a great solution for that one yet. I'm leaning towards using stereo audio tape like used on reel-to-reel players. I could build looping programs by physically looping the tape, and could use a low frequency carrier and high frequency carrier to read both data and clock pulses. It's not the best solution, but it's one that I think I should be able to build in the garage!
A version of the pdp-8 (pdp8s?) does serial operations.
maybe some hardware for quickly doing serial additions is good idea?
Would be neat. You could boot OS8 or even TSS8 maybe. Would be very retro as the PDP8S was transistor logic. The ultimate number of tubes would likely be awesome.
The PDP-8S is a brilliant machine! But the overall layout of it quite a bit different and wouldn't really mesh very well with the architecture that I'm using here. The MC14500 was truly meant to do operation on a single bit at a time. While it may be slow, I think it's excellent for visualizing how data is moving about the system. In the future I definitely want to explore other 1-bit serial type architectures though, so I'll definitely be looking into the PDP-8S in more detail at that time!
Very nice video
Thank you very much!
Thanks for these videos! Iv been watching your builds. Really am impressed and inspired. Love the vdf. do you have a schematic of the vdf circuit? Id love to breadboard this adder using 6205 submini pentodes. cheers!
Thank you for checking the build out!
Here's the schematic for the full adder: i.postimg.cc/fTsnwnWC/Full-Adder.png
And here's the schematic for the VFD portion: i.postimg.cc/x8BFFZjX/7Seg.png
If you get it breadboarded and working, definitely let me know, I'd love to see it!
Very nice, what does your Usagi intro board do or is it just nice to look at?
I think your boards are getting better with every one - and I don't want to say that your first ones were bad but I think the density and "beautifullnes" is getting better each time.
I also just love how many other great electronics youtube people watch your videos, very cool to see that they are all one big "society".
Another thing I have to remind myself of every now and then is that you are doing all of this with low voltages, which is pretty amazing.
Oh and this weeks time laps soundtrack kind of has a Nier Automata vibe to it, very cool.
Greetings,
Michael
Thank you so much for the praise on the boards! When I designed the Instruction Register and Decoders, there was a lot of repetition in the design, and then the interconnect was kind of a mess because I needed to get a bunch of signals to a whole lot of different places, but I feel like I started to really get into a good groove when designing this next section that has the ALU and RR and Carry Registers! This 8 tubes per board design idea actually comes up a lot in this section.
The Usagi Intro board really doesn't do anything special other than flash the LEDs hidden behind the cheeks. It's a basic astable multivibrator using the two tubes that are the ears and running on just a single +12V wall wart. Here's the schematic for it: i.postimg.cc/0NDsDxXv/UELogo-Schematic.png
Ooh, I'll have to look into some Nier Automata stuff in the future if it has a similar vibe! This week's music was by Shnabubula, who is probably the most common artist I use music from. They do some really amazing work!
@@UsagiElectric
Very cool, I love how you actually used the ear tubes for the LEDs and diddn't just put in a 555 or microcontroller.
For the Nier Automata stuff, try "city ruins" or "weight of the world", "Amusement park" is also very good.
the first time "hello and welcome back" was said on this channel i think
WOW!! NEW INTRO!! JUST GREAT!! ☺
Well, now it's 100% Usagi's Electronic Computer 14500. Because now it has its own full-fledged ALU and Instruction set.
Eh, it all started with simple idea - "just create a MC14500B computer using the vacuum tubes"! {Apparently, this is how new things are born}
Once again, I wish you good luck in the project! :-)
... and... I.. didn't quite understand from your schemes, ..but how are you going to clear this carry "register"? After several addition operations and before starting the next ones, it is necessary to somehow clear the carry register.
..
.
.
.
(did you really not like the idea of using the standard "LU" sooooo much?).
Thank you so much!
I spent like two full days working on that 11 second intro, so it's a huge relief to hear that it was worth it!
Yup, we've got our own unique instruction set and a proper ALU now! Well, almost, we've got the "A"rithmetic, but we haven't build the "L"ogic part yet. That's coming up next.
As for clearing the Carry Register, I have an instruction that forces a "1" into the Results Register and forces a "0" into the Carry Register, regardless of what is actually stored in either register. It's not the most ideal solution, but with this architecture, it's often beneficial to have the ability to force a "1" into the Results Register, so I piggybacked clearing the Carry register off that instruction.
And I was actually planning on totally using a regular LU like in the MC14500, until I tried to program a simple increment program on the actual chip and it was a nightmare ( th-cam.com/video/l7fpswJrG04/w-d-xo.html )! It was like 12 instructions for a simple binary add, which I can now knock out in just 1 instruction. Nothing breeds innovation quite like spite, and so the ALU on this build is mostly just to spite the LU on the MC14500, haha.
Here's the full instruction set if your curious!
0000: NOP - No operation (Set Flag O)
0001: LD - Load value on Data bus into RR
0010: ADD - Perform binary addition of value in RR and value on Data bus
0011: SUB - Perform binary subtraction of value in RR and value on Data bus
0100: ONE - Force a "1" into RR and "0" into Carry
0101: NAND - Perform logic NAND on value in RR and value on Data bus
0110: OR - Perform logic OR on value in RR and value on Data bus
0111: XOR - Perform logic XOR on value in RR and value on Data bus
1000: STO - Output value in RR onto Data bus (Set Write Flag)
1001: STOC - Output inverse of value in RR onto Data bus (Set Write Flag)
1010: IEN - Store value on Data Bus in IEN register (enable/disable input to ALU)
1011: OEN - Store value on Data Bus in OEN register (enable/disable Write Flag)
1100: JMP - Set Flag JMP
1101: RTN - Skip next instruction (Set Flag RTN)
1110: SKZ - Skip next instruction if value in RR is "0" (inhibit Flag F)
1111: NOP - No operation (Set Flag F)
My digital lecturer's wife's maiden name was Norgate....
Have you ever used Logic Friday? It'll convert from truth tables to logic diagrams for you.
Now that was a match made in heaven!
I hadn't heard of Logic Friday, but that seems like it would have saved me so much time trying to come up with a logic circuit for the 2-bit to decimal display. I'll definitely check it out, thanks for turning me on to it!
very cool to make things out of tubes. Reminds me of Spock, Star Trek, making a computer out of tubes. But, should you be trying to make things out of tubes that you cannot make out of solid-state, like free energy devices or something?
Thank you!
This project is my kind of love letter to the vacuum tube computers of yesteryear! There's honestly not much that you can make out of tubes that you can't make out of transistors. Vacuum tubes make brilliant audio amplifiers, but I don't have too much interest in audio stuff (largely because I'm deaf in one ear and just can't hear the difference in quality between them).
Also, I like that when building large logic circuits out of vacuum tubes, you have think in new and interesting ways. Transistors are so small and cheap that adding in a few more transistors here or there is super easy, but with tubes, they're relatively huge and not cheap at all, so it requires some really fun problem solving to come up with a viable circuit, and that's been the best part of this build so far!
Had you gone for the Raytheon/Philips Wireended Vacuum tubes you could have used a smaller board.
You're absolutely right!
Those submini tubes would have saved a ton of space, but with the size of this project, I really have to take into consideration tube availability in large quantities, and more importantly, how happy my wallet is. Those submini tubes, cost anywhere from $5 to $15 per tube. Assuming I could find enough of them to build the entire project, it'd end up costing me around $800 in tubes just for the processor!The 6AU6 on the other hand I often find for around $0.80 per tube, which makes my wallet much happier.
Plus, all the 7-pin tubes sticking straight up out of the PCB looks kind of cool in my opinion, haha.
@@UsagiElectric TELEFUNKEN GMBH
@@DAVIDGREGORYKERR I'm not sure I'm following you. I checked Telefunken's site and new submini tubes are $20 a tube! Even on eBay I can't find Telefunken tubes for anywhere near affordable prices for the volume that I would need here.
@@UsagiElectric That is a shame sorry.
awww the rabbit
The bunny didn't know what to think of the vacuum tube bunny, but after a couple good sniffs, she came to the conclusion that the vacuum tube bunny wasn't going to give her any treats, haha.
Hey! Cool as always, this is incredible, I was trying to use that russian vfd tubes but use them with vacuum tubes is awesome, do you have the schematic? I recently entered to discord again
Thank you so much!
I don't exactly have a schematic, but I do have a logic diagram for the entire setup as it sits: i.postimg.cc/26BgcHVH/1bit-FA-2bit7seg.png
As for wiring up the VFDs, it's a little confusing because you have to count from the short pin, but this website has some excellent information on getting the VFD working: zw-ix.nl/blog/tag/vfd/
Hit me up on the Discord and I'll be happy to work through them with you!
Can anyone explain why my comments keep getting rejected ?
It seems to fail if I mention the code repository -- maybe some word is triggering it?
Any suggestions would be useful !
Ok got it now. TH-cam doesn't just reject comments with links, but any mention of other popular sites.
Yeah, TH-cam has been having a difficult time with spam lately, so any comments referring to external sites or links cause TH-cam to automatically nuke the comment before I get a chance to see it.
@@UsagiElectric Thanks. Been loving this project, and your delivery is brilliant -- really laid back and fun. I was so inspired by the diversions in this episode, I wanted to build a 7-seg display driver, but *in UE14500 code* . So I wrote a little code simulator in Python and worked it out. Great exercise for the cct design -- just the kind of thing it was meant for! However, it's a much bigger calculation than I realised. Best effort so far took **67** instructions, based on the logic diagram you showed, but my first naive code took about 140 instructions!
If you're interested I can post on discord.
Hope this one message gets though. All the best for future projects !
6:17: This makes me wonder how much more complex it would be and how much it would increase the gate/tube count to build a two-bit CPU? One that would still be similar to the MC14500B and UE14500 but theoretically more capable? I'm not sure if anyone has ever designed a "real" 2-bit CPU. There was the Intel 3000 bit-slice family, but I'm not sure if that's quite the same, and also I've not been able to google how many logic gates or transistors that contained. I guess the UE14500's 192 tube count and the Intel 4004's 2,300 transistor count do provide some upper and lower bounds, but it would be interesting to know more. Also interesting: Wikipedia has 1-bit computing and 4-bit computing articles, but no proper article on 2-bit CPUs as such.
Late to the party 😊 Why invert a NAND instead of a diode AND for the half adder?
Just found your channel through HackADay! Liked, subscribed and commented for you and the youtube alg. Your videos are awesome and you deserve more views, Sir :D
Thank you so much! The best part of the TH-cam thing is reading all the comments, so thanks for chiming in!
@@UsagiElectric Some might say it's the opposite. I suppose it depends on what types of viewers your videos attract. Your videos are highly specialized and extremely technical and therefore probably only attract the VERY technical type. They are less harsh. If you review phones and computers for a living... I'd have someone else to read them for you and PLEASE not have that person tell you about what's going on in the comments :P
Intrageted
VLSIC
Transistor ic
I keep trying to post a comment, but I think it is rejecting me, they all disappear.
Is it because I had a link in it?
Tried again, still having trouble for some reason. I have built a hex to 7-segment decoder in UE14500 code, and would like to share it !! Fun problem I thought, and just the kind of thing the original MC14500 was meant for !
But it seems to block me any time I include something looking remotely like a url ??
P.S. some details of the code --
The real surprise was how much complexity this requires !!
I see your sample circuit is using @24 gates, or "16 tubes, plus >100 diodes" .
A semi-rational human approach, with some manual optimizations, still took 173 instructions (!)
With code based on your circuit from the video, I now have it down to 67 instructions =about 3 for each of your 24 gates.
Building stuff with valves is just a bit TOO RETRO....
I dare say it's perfectly retro!
But, I do have some logic behind that assertion. The phrase "Creative Limitation" or "Limitation breeds Creativity" is something that I find fascinating. The human brain is amazing at problem solving, and working within severe constraints or limitations really allows the brain to wild. Massive hardware and software size limitations bred some of the most amazing video games in the history of gaming. As hardware got faster and faster and hard disk space grew wild, I feel it became harder and harder for devs to really tap into that creativity that made those early games so endearing.
So, in a way, but building stuff with vacuum tubes, I'm placing massive limitations on my self and really forcing my brain to get creative with problem solving, just like the legendary game devs from the early micro era!
@@UsagiElectric uh, no...
Unsubbing.
@@Theineluctable_SOME_CANT Sad to see you go, hope you check back on the occasion to see the progress on the tube build!
Take it easy!
@@UsagiElectric thank you. And you too, sir. I would like to see you use transistors.... much more k3wl....
@@Theineluctable_SOME_CANT Funny you should mention that because I've been working on a transistorized version of this processor using 2N3904s in my spare time:
i.postimg.cc/XvvN5GW8/Transistor-Version.jpg
I'm planning on ordering the PCBs for this one, so once I get further along and actually have some usable results, I'll be making a video about it!