There was a benchmark done by the Computer History Museum of how many digits of the fibonacci sequence a computer could compute in 15 seconds. Would be fun to see how well the UE1's performance compares to other vintage computers!
That requires having two numbers in memory to add up. With 1 Byte of memory, you can only store two numbers in the range 0-15. So you will only get up to the 7th digit, which presumably this machine running at >100 Hz for 15 seconds will be able to do, but then it will just have to bive up when it gets to the un-imaginably large number of 21.
@@ShieTar_ you can definitely pair it with an arduino to serve as memory in a turing machine style, it could only move left or right in offsets of 127 and still be turing complete.
How is this comment made 19 hours ago (as I write), yet the video was only released 50 minutes ago. Seems the UE1 also has temporal travel characteristics.
Hi, I don't understand a word of English or American, but with the UA-cam translator your channel in Italian is magnificent, I also collect ancient telecommunication equipment, this video hasn't been translated yet, I can't wait for it to be available in Italian , thanks for sharing your work
I'm glad the UA-cam Italian translation is alright! I've been really excited about UA-cams work towards getting the videos available in other languages. Hopefully, the Italian text-to-speech doesn't sound too weird!
@ Hi, the translation is almost perfect, the summary loses a bit of emotion and passion, but it doesn't matter, I saw your original audio videos with the subtitle, I think this tube computer is brilliant
Great job on this and my dad commented that the power supply issues that you're running into remind him of trying to run early computer equipment when he was stationed in Greenland. He said to let you know to go for it and if it works you'll be pretty warm shortly!
Oh man, I can't imagine the difficulties of running computers in extreme cold environments like Greenland! Once they're up, you're golden, but that initial power up is going to be a challenge for sure.
An astrophysicist I used to work with had previously managed the server room at the US base in Antarctica. Apparently they had some interesting issues there with controlling the temperature and humidity. As you bring the cold air inside and it gets heated by the computers the relative humidity drops to almost nothing, causing huge issues with static electricity.
Hey, you probably know this already, but in the video it seems like you may have overlooked this. The ground of your multimeter should be at the ground of the board you're measuring. In higher current applications, the current flowing back through the conductor to ground can lead to a significantly higher ground level, which leads to an overall lower effectice potential difference at the devices you're trying to measure. So the values you've been getting might actually be quite a bit lower in reality. Just a little heads-up! :-)
Congratulation to the completion! It was quite a journey: not just fried metal, but also broken glass is part of the history. But like the engineers that pushed ahead to success, so YOU did. - Beside bringing join to you, it brought memories back from the time I entered the technology and investigated MC14500 ICU. I even restarted some of the activities. I think you extension/enhancements and providing the program by life-reading a paper tape are really exciting. Thank you for going back to the future.
Congratulations David! Maybe the bell is ringing randomly because, having imparted your own positivity into the machine, it's just as excited as you are!
You certainly have a ton of perseverance. Congratulations. You tackled some very interesting challenges along the way, in the design. Love the homemade paper tape reader! I have a homebrew 8085-based college project from 42 years ago, that was never completed, and I am finally getting to it. My challenges will be far fewer than yours were, so it should be easy peasey.
Just going to throw out some suggestions regarding chasing weird logic issues. 1.) I'm not sure if this is contributing to the weird issues, but you might want to install a clamping diode across all inductors (relays, bells, etc.) that get switched off and on, because of the counter EMF spike that happens when power is removed from them. That spike can cause all sorts of random issues toggle stuff off that was on, and stuff that was on, toggling it off. 2.) Also, be aware that long parallel wires with changing data, can couple their signals back and forth between them because the paths are not balanced with their complementary signal which cancels out fields emanating from them. That is why they use twisted pairs for signal wires instead of single wires with one common ground return. On circuit boards, designers often use guard lines to help minimize the radiating fields around the single data lines 3.) Beer and Doritos. Because the solution is probably a lot simpler than you realize and it helps prevent over thinking the problem.
put on a soft-start on the powersupply so that it slowly brings up the current on turn-on, it will extend the life of the tubes at the same time as prevent the tubes from drawing too much current. on old vacuum tube TV's they put in a thermistor that's resistance goes down as it warms up.
There is, that's what the couple tubes and big fat resistors with relays at the top of the modules are. It was so cold that the resistors just blew up from the inrush.
Maybe it needs a sequencer thingy like the Bendix? Split the tube filament supplies into two sections (with a common.) have the sequencer turn on the tubes on the I/O-tape reader board first with a soft-start. Then - a few seconds later - have it start the tubes on the processor board, again, with a soft-start circuit.
It's not necessary in this case IMO, just current limit all filaments the with high-ish value PTC thermistor for a minute, then bypass the thermistor. The filaments will slowly but surely warm themselves up. The main problem is not the surge itself, it's that it causes filament burn-out.
@@jwhite5008PTC is a good idea. They are using it on primary of SMPS PSUs all around for that purpose. Light bulb for this kind of vacuum computer would make better visual effect, but at such high power needed by filaments it is not really an option.
@@swrzesinski the idea of PTC is to dampen the inrush current and get to like half the required filament voltage, then it should be bypassed by a relay. A stripe or sector indicator tube such as soviet 6Е1П or 6Е3П can be used to see how much the filaments warmed up - they were sometimes included in early radios for that reason.
Awesome to have this project finally finished! I’ve been following it for a long time, and this is one of the inspirations for my own homebrew computer architecture. Back on topic, congrats on finishing the ue-1, and I can’t wait to see what you get up to next! Cheers!
its been a journey I followed this project from the first episode with the goal to build one myslef...now ..I actually realized how I underestimated the complications and power draw of such a system. so thanks for doing it so I dont have to do it.
This man is a genius! I love these early systems. I can't believe a system like the centurion with it's drives, terminal, and printer could still be found today, and wasn't scrapped for extra space long ago, nevermind be fully functional!
@@volvo09 They surely built them well back then. It is incredible to me how much smaller, more powerful, and _how cheaply_ modern computers are built nowadays compared to the "big iron" of days gone by.
Looks like the bell relay was chattering. Didn’t see a diode across the coil to prevent that surge from the collapsing magnetic field when the relay is opened…
And he has this 100µF capacitor in parallel to the coil. That should eliminate all back EMF just like a diode would. Maybe putting a small (100 Ohms?) resistor in series to quickly dampen that resonance circuit?
@@ttyR265 I had a crazy one once where a bell solenoid was destabilizing a power supply, causing a MCU to reboot when the bell rang. The MCU was powered by a SMPS and the filter capacitor on the voltage regulation feedback mechanism (you know, through an optocoupler) hand dried out. So the SMPS tended to overreact to noise coming from the bell solenoid, causing the MCU to undervolt and reboot!
Well done. Everytime I watch that paper tape looping and jumping around I'm just waiting for it to catch on something, or somehow fail. I'd like to see a bit more restriction on where it can go - a much longer program for example would be frightening to behold.
Congratulations! You completed a tremendously complicated project and also learned some things not to do in the future. I'd call that a massive success.
Isn't this a little anti-climatic? Shouldn't we at least see a small program source, that takes 2 numbers, and add them or something, and then see the result?
With how little time he's spent talking about the program he's running (we've seen snippets but no explanation) I think he worries we wouldn't enjoy seeing the programming - but on the contrary I personally would be so psyched to see the gritty details of the programming.
When you were testing the bell, it was flat on the table. On the wall, it's vertical. Maybe test it with the breadboard while vertical and see if the results are different. Gravity can sometimes affect how things work.
Holy crap, congratz! I am a bit sad we didn't get to hear the bell again, in this video, but I am looking forward for your future exploits. Love your channel, and thank you for sharing. Also, you're making me want to get back to my abandoned 68000 homebrew plans :P
David, I've been following your work a couple years now. I just hope the retro computer community can count on your harde work for many years to come, with many big milestones as this one. Very nice indeed! Congrats!
Nice job. You are an excellent troubleshooter. I always say it's okay to be a troublemaker if you are also a good troubleshooter. hyper-interesting stuff and I'm excited about any multi-bit processor you are planning! All good wishes.
Congrats dude! i'm happy for ya! Still looks pretty awesome too! (someone with a background in component level digital electronics etc) Looks great aesthetic gets A+ :)
It's done! Congratulations, glad to see it completed. Your self critique half-way through shows how much you learned in the process, meaning whatever you build next will be way better!
I just watched this 2 am in Australian Time, was ready and waiting all along until the Usagi Alert appeared for this Video, It was worth while staying up ! Me and my Rabbit Felix watched your UE1 together with a home made brewed Coffee !
Since it's new project time, I have a couple of things I've had in mind that I've never had time to get around to starting the groundwork on, maybe one will be an earworm (mindworm?) and start you off on something really unique. 1 - building with diode logic... but using LEDs as the diodes. That's just a concept though, haven't thought too hard about it. 2 - A CPU based on an architecture that uses concurrent, free-flow operation. It's hard to word this out because it's simple, but very different from anything I've ever run across. On "normal" architecture, there's a logic unit with a given circuit that's activated by operand and moves through it's operation based on clock cycles. That description is likely wrong but it's not important, I gave 'it works stepwise' about 30 seconds of thought across both Von Neumann and Harvard architectures, and there you go. So what am I proposing that works differently? An incredibly reduced instruction set that at the base level only manages timing and movement of bits or bytes into registers. The register set though, is the entire memory space. Why? Because each operation the computer is capable of is a separate module keyed to individual memory locations for input and output, and is running in a constant state. That adder, it's always functioning. Tremendous power hog, but that's not the point... the point is with modules keyed to input registers, the moment you complete moving bytes to those registers, the output is already calculated because it's always operating. "Programming" is then as easy as knowing which output registers hold the results of each module, and plucking that and putting it in the next step. Chain modules are possible too because there's no reason the output of one module can't be the direct input of another. Consider you have an add, subtract, multiply, shift, and invert module all wired for the same two input registers. Your program moves a byte to one input, and to the other input, and after an appropriate wait for the result to settle, you already have the result for each of those operations ready in their respective result register, the program's next step is simply to pluck the desired result and put it where it needs to go next. There's no reason you couldn't have multiple operational modules with the same functionality, just watching different or even matrixed registers. if one so desired you could have three input registers, A B C, and three addition modules with three outputs D E F.... A+B=D, A+C=E, B+C=F, and by filling the three inputs you just automatically have all three outputs ready to use. Modules could even be much more complex than simple operations, a module can be literally whatever you want taking inputs and providing outputs, all you need to know is which registers to fill and which registers to pull the answers from. Memory is simply another module with flipflops and more complex address decoding, that provided an output on the same location that was input (which would actually require a select line but only so it's not contending for the bus to return the value, as far as the CPU is concerned the wait between poke and pull is managed the same) Harvard would be easier on this with the amount of addressing that goes on in this schema, the 'on the ground' device would be a backplane with a 'control card' and any number of modules, and power draw would be the simple addition of power draw for every single part because they're all running constantly. It would need a compiler and the control board some circuitry for retrieving instructions from ROM space. The compiler would need to keep track of reserved registers by module activity, but I would be surprised if the actual logic control needed more than 16 instructions to accomplish everything, no matter the scale (depending on address bus width, you want 1024 modules? Go for it, prepare for the power bill from hades)
Now that it all works and is troubleshot and debuggified, the modularity of the thing would lend itself to building a cool midcentury retrofuturistic cabinet for it! Front panel toggle inputs, gauges, blinkenlightzen, and lots of fans!
Hooray! Congratilations! As I know now your computer is world first modern vacuum tube computer which was FINISHED! All other projects were stucked or scraped. I also have now idea when I would be done with my own...
Great Project ! Some (unwanted) suggestions: 1. Bell relay: Put a snubber diode across the coil, to prevent a high voltage spike when deactivated. Also, since you basically have a capacitor across the relay coil (as a snubber cap), make sure it is a snubber rated capacitor (or just remove it and replace it with the mentioned diode). 2. Since you power buss is so long, I would put some bypass capacitors every so often (.01-.1 uf). 3. Add some soft-start circuits, arranged in zones (several options on ebay).
Epic! A few years ago I stumbled across your first experiments with the MC14500 chip on a breadboard, as I was also building a homebrew on breadboards. Ever since I followed your UE1 build. Truely an awesome archievement. Congrats!
Congratulations! I loved watching this series, your unbridled enthousiasm and watching the UE1 go from crazy idea via lots of debugging to something actually working. Can't wait to see what the UE2 will look like!
It's been a fun journey. I hope we get to see the adjustments which you're planning to make to the UE1. Maybe you should design a progressive power-up into the UE1 so that it soft-starts different sections to avoid overloading the power supply? At 7:30 the "like" counter rolled over to 666 so I hit the button to roll it to 667. You're welcome! 😁
Congrats! It’s been interesting & fun watching this project develop from (for me) fairly early on… an immense amount of work on your part with its fair share of highs and lows and now it’s there! And it looks like a piece of art on your wall too 😊
@@SnakebitSTI You're right of course, but I just love the idea.If that was available as a kit I'd buy and build it. With tons of LEDs... add peripherals and run CP/M on it. Yum!
You could tackle the power on surge quite easily: you could manually stage the bring-up, one board at a time. Modify your little connector PCBs to have a physical switch on the heater power line. Congraturation on a job well done!
A true triumph is making a real true retro computer by hand. The sheer complexity of working with tubes as your switches and all the variables of dealing with the materials like you've been through makes any more complex computers using tubes just an incredible engineering achievement. Thanks for sharing!
The Bendix G-15 behind you showed you how to do it: raise the voltage gradually/in steps! And/or use some kind of stepping relay to power up the sections one at a time. Stereo amps/receivers having been dealing with that for a long time too.
I like your energy level and the way you think, this is coming from a life long embedded HW/SW guy 🙂 The whole project is beyond amazing and lets us old enough to remember learning on tubes a great vicarious experience! Bravo! 👍👍
3 years is fast! My own insane homebrew project was started about 4 years ago, and I haven't even started building the hardware yet... must get back to it soon!
This is such an epic project. I'd love to see more of it in the future. Only today I finished my MC14500 computer following your series from a few years ago :)
Somethin that's used on large vac tubes is a preheat circuit If you insert a power resistor in series with the heaters that gives you 50-70% of the power, then a relay contact across the resistor, you can let it warm up on low power Wire the relay like a start/stop that's on motors, when you press a button it latches, but when the machine is turned off it drops out to low power ready for restart.
If you look at the top of each board, you'll see big honking power resistors. That's my soft-start mechanism! It's been there for a while and I cover it in a few previous episodes in the series, but even with that, the power supply tends to be right on the ragged edge during power up on a really cold day.
@@UsagiElectric But are those resistors just to drop the current to normal operation my ideaa lowers that current so the heaters are barely on, not even glowing, untill they have warmed up sufficiently
"I'm endeavouring, to construct a mnemonic memory circuit using stone-knives and bear-skins." Congratulations getting the machine operational! And the bell is classic - love the callback to the other project.
A very long time ago in Elektor magazine, i read about inrush current limitation for audio amps with large dc reservoir capacitors. The solution was to insert series resistors with a bypass relay on a timer or based upon voltage drop across the resistor (as a form of current measurement). Perhaps this solution would allow you to limit inrush current by increasing the filament circuit resistance presented to the power supply, limiting the inrush current to the filaments and, as a final benefit, preheating the filaments to prolong their life.
Awesome, it’s funny how quite often I find the journey more enjoyable than the results. It’s probably a good time to take inventory of what you’ve learned, reflect on that and absorb the learnings that you want to take forward. We tend to call that part a “retrospective”, do it physically, like write it out or, better, PowerPoint and document it. Develop your own best, and worst, practices that you want to have on the table for every future project. Here’s an example of one I’ve added to be box of things I’ve learned that I want to reinforce remembering: ALWAYS check input voltages.
This project has me itching to design a ternary computer (don't know why, but I've just always found the possibility of three-state gates fascinating). I'd love to do something physical, but I might even just write an emulator at this point. It would be cheaper and less time consuming. Super cool project.
There was a benchmark done by the Computer History Museum of how many digits of the fibonacci sequence a computer could compute in 15 seconds. Would be fun to see how well the UE1's performance compares to other vintage computers!
Yeah, we need a benchmark!
That requires having two numbers in memory to add up. With 1 Byte of memory, you can only store two numbers in the range 0-15. So you will only get up to the 7th digit, which presumably this machine running at >100 Hz for 15 seconds will be able to do, but then it will just have to bive up when it gets to the un-imaginably large number of 21.
@@ShieTar_ you can definitely pair it with an arduino to serve as memory in a turing machine style, it could only move left or right in offsets of 127 and still be turing complete.
Would be cool to release a short of the bell getting fixed and working
That's a good idea! Once I sort out the problem, I bet it's a quick fix and would work well for a short video!
How is this comment made 19 hours ago (as I write), yet the video was only released 50 minutes ago. Seems the UE1 also has temporal travel characteristics.
@@andrewdavie386 patreons get the videos early, calm down, you're not going insane 🤣 the date on the video is solely the published date
@@andrewdavie386 the video was available to patrons eariler
@@andrewdavie386 Patreons get early access
We want... no, we NEED the bell to ring. Get back to work ! The opera's not done until the fat bell sings.
The bell rang in the forest. That's why we didn't hear it.🌲🔔🌲
@@youjustlikeit3774 In the forest of vaccum tubes lol
Hi, I don't understand a word of English or American, but with the UA-cam translator your channel in Italian is magnificent, I also collect ancient telecommunication equipment, this video hasn't been translated yet, I can't wait for it to be available in Italian , thanks for sharing your work
I'm glad the UA-cam Italian translation is alright! I've been really excited about UA-cams work towards getting the videos available in other languages. Hopefully, the Italian text-to-speech doesn't sound too weird!
@ Hi, the translation is almost perfect, the summary loses a bit of emotion and passion, but it doesn't matter, I saw your original audio videos with the subtitle, I think this tube computer is brilliant
David, you make my Sunday mornings joyful. Looking forward to the next one.
Thanks!
Great job on this and my dad commented that the power supply issues that you're running into remind him of trying to run early computer equipment when he was stationed in Greenland. He said to let you know to go for it and if it works you'll be pretty warm shortly!
Oh man, I can't imagine the difficulties of running computers in extreme cold environments like Greenland! Once they're up, you're golden, but that initial power up is going to be a challenge for sure.
kinda not a problem when they also brought along a nuclear reactor for those early SAGE computers
By modern standards the UE1 is not much of a computer, but, if marketed as a programable space heater, it might get some marketing interest. 😜
An astrophysicist I used to work with had previously managed the server room at the US base in Antarctica. Apparently they had some interesting issues there with controlling the temperature and humidity. As you bring the cold air inside and it gets heated by the computers the relative humidity drops to almost nothing, causing huge issues with static electricity.
Hey, you probably know this already, but in the video it seems like you may have overlooked this. The ground of your multimeter should be at the ground of the board you're measuring. In higher current applications, the current flowing back through the conductor to ground can lead to a significantly higher ground level, which leads to an overall lower effectice potential difference at the devices you're trying to measure. So the values you've been getting might actually be quite a bit lower in reality.
Just a little heads-up! :-)
Congratulation to the completion! It was quite a journey: not just fried metal, but also broken glass is part of the history. But like the engineers that pushed ahead to success, so YOU did. - Beside bringing join to you, it brought memories back from the time I entered the technology and investigated MC14500 ICU. I even restarted some of the activities. I think you extension/enhancements and providing the program by life-reading a paper tape are really exciting. Thank you for going back to the future.
Congratulations David! Maybe the bell is ringing randomly because, having imparted your own positivity into the machine, it's just as excited as you are!
A wise friend once commented: "Some times things just don't make sense, until you figure them out." Seems relevant.
Congratulations!
You certainly have a ton of perseverance. Congratulations. You tackled some very interesting challenges along the way, in the design. Love the homemade paper tape reader! I have a homebrew 8085-based college project from 42 years ago, that was never completed, and I am finally getting to it. My challenges will be far fewer than yours were, so it should be easy peasey.
Just going to throw out some suggestions regarding chasing weird logic issues. 1.) I'm not sure if this is contributing to the weird issues, but you might want to install a clamping diode across all inductors (relays, bells, etc.) that get switched off and on, because of the counter EMF spike that happens when power is removed from them. That spike can cause all sorts of random issues toggle stuff off that was on, and stuff that was on, toggling it off. 2.) Also, be aware that long parallel wires with changing data, can couple their signals back and forth between them because the paths are not balanced with their complementary signal which cancels out fields emanating from them. That is why they use twisted pairs for signal wires instead of single wires with one common ground return. On circuit boards, designers often use guard lines to help minimize the radiating fields around the single data lines 3.) Beer and Doritos. Because the solution is probably a lot simpler than you realize and it helps prevent over thinking the problem.
Congratulations!!
The only thing missing is an Usagi Electric Mk 1 nameplate, perhaps on the front of the paper tape box.
put on a soft-start on the powersupply so that it slowly brings up the current on turn-on, it will extend the life of the tubes at the same time as prevent the tubes from drawing too much current. on old vacuum tube TV's they put in a thermistor that's resistance goes down as it warms up.
Or maybe divide the filaments into banks that can be switched on sequentially.
There is, that's what the couple tubes and big fat resistors with relays at the top of the modules are. It was so cold that the resistors just blew up from the inrush.
Maybe it needs a sequencer thingy like the Bendix? Split the tube filament supplies into two sections (with a common.) have the sequencer turn on the tubes on the I/O-tape reader board first with a soft-start. Then - a few seconds later - have it start the tubes on the processor board, again, with a soft-start circuit.
It's not necessary in this case IMO, just current limit all filaments the with high-ish value PTC thermistor for a minute, then bypass the thermistor. The filaments will slowly but surely warm themselves up.
The main problem is not the surge itself, it's that it causes filament burn-out.
@@jwhite5008PTC is a good idea. They are using it on primary of SMPS PSUs all around for that purpose. Light bulb for this kind of vacuum computer would make better visual effect, but at such high power needed by filaments it is not really an option.
@@swrzesinski the idea of PTC is to dampen the inrush current and get to like half the required filament voltage, then it should be bypassed by a relay. A stripe or sector indicator tube such as soviet 6Е1П or 6Е3П can be used to see how much the filaments warmed up - they were sometimes included in early radios for that reason.
Awesome to have this project finally finished! I’ve been following it for a long time, and this is one of the inspirations for my own homebrew computer architecture. Back on topic, congrats on finishing the ue-1, and I can’t wait to see what you get up to next! Cheers!
Congratulations!
fantastic! this has been an amazing saga to watch unfold!
congrats, such an awesome project. love the tubes and the paper tape reader!
its been a journey I followed this project from the first episode with the goal to build one myslef...now ..I actually realized how I underestimated the complications and power draw of such a system. so thanks for doing it so I dont have to do it.
Well, you are getting there, man. First the Centurions, then the G15, then the PDP-11, now the UE-1. Keep 'em comin' bartender!
This man is a genius! I love these early systems. I can't believe a system like the centurion with it's drives, terminal, and printer could still be found today, and wasn't scrapped for extra space long ago, nevermind be fully functional!
@@volvo09 They surely built them well back then. It is incredible to me how much smaller, more powerful, and _how cheaply_ modern computers are built nowadays compared to the "big iron" of days gone by.
Looks like the bell relay was chattering. Didn’t see a diode across the coil to prevent that surge from the collapsing magnetic field when the relay is opened…
back emf isnt usually a problem with valve/tube circuitry...
I was thinking the same thing...wonder if the coil is setting up some sort of resonance in operation.
And he has this 100µF capacitor in parallel to the coil. That should eliminate all back EMF just like a diode would. Maybe putting a small (100 Ohms?) resistor in series to quickly dampen that resonance circuit?
@@ttyR265 I had a crazy one once where a bell solenoid was destabilizing a power supply, causing a MCU to reboot when the bell rang. The MCU was powered by a SMPS and the filter capacitor on the voltage regulation feedback mechanism (you know, through an optocoupler) hand dried out. So the SMPS tended to overreact to noise coming from the bell solenoid, causing the MCU to undervolt and reboot!
What a journey. I came in probably about halfway through it. Congrats.
hope we see a few more pieces of content involving running programs on the UE1
I always love the joy, excitment, and positive can-do energy of your videos! You keep making them I'll keep watching them. Thanks!
Well done. Everytime I watch that paper tape looping and jumping around I'm just waiting for it to catch on something, or somehow fail. I'd like to see a bit more restriction on where it can go - a much longer program for example would be frightening to behold.
Maybe the bell relay coil needs a fly back diode? Congrats on getting it running!!
yes thought so too.
Or the bell itself?
@@RJTCThe bell doesn't need a flyback diode, it's a bell. Unless you're talking about the thing that strikes the bell electronically?
@@nikkiofthevalley It's an electromechanical bell - containing a solenoid.
Congratulations! You completed a tremendously complicated project and also learned some things not to do in the future. I'd call that a massive success.
Sunday morning coffee with Usagi here in Arlington, TX! I always look forward to this.
19:18 I have followed this project from the get go. It has been an honour to witness every step. THANK YOU.🎉❤🎉❤🎉❤🎉❤
Isn't this a little anti-climatic? Shouldn't we at least see a small program source, that takes 2 numbers, and add them or something, and then see the result?
With how little time he's spent talking about the program he's running (we've seen snippets but no explanation) I think he worries we wouldn't enjoy seeing the programming - but on the contrary I personally would be so psyched to see the gritty details of the programming.
Same
When you were testing the bell, it was flat on the table. On the wall, it's vertical. Maybe test it with the breadboard while vertical and see if the results are different. Gravity can sometimes affect how things work.
❤❤❤❤❤❤
What an incredible achievement. I can't wait until you start version 2.
Holy crap, congratz! I am a bit sad we didn't get to hear the bell again, in this video, but I am looking forward for your future exploits. Love your channel, and thank you for sharing.
Also, you're making me want to get back to my abandoned 68000 homebrew plans :P
David, I've been following your work a couple years now. I just hope the retro computer community can count on your harde work for many years to come, with many big milestones as this one. Very nice indeed! Congrats!
Nice job. You are an excellent troubleshooter. I always say it's okay to be a troublemaker if you are also a good troubleshooter. hyper-interesting stuff and I'm excited about any multi-bit processor you are planning! All good wishes.
Congrats dude! i'm happy for ya! Still looks pretty awesome too! (someone with a background in component level digital electronics etc) Looks great aesthetic gets A+ :)
It's done! Congratulations, glad to see it completed. Your self critique half-way through shows how much you learned in the process, meaning whatever you build next will be way better!
I just watched this 2 am in Australian Time, was ready and waiting all along until the Usagi Alert appeared for this Video, It was worth while staying up ! Me and my Rabbit Felix watched your UE1 together with a home made brewed Coffee !
Give Felix some extra international pets from me!
Since it's new project time, I have a couple of things I've had in mind that I've never had time to get around to starting the groundwork on, maybe one will be an earworm (mindworm?) and start you off on something really unique.
1 - building with diode logic... but using LEDs as the diodes. That's just a concept though, haven't thought too hard about it.
2 - A CPU based on an architecture that uses concurrent, free-flow operation. It's hard to word this out because it's simple, but very different from anything I've ever run across.
On "normal" architecture, there's a logic unit with a given circuit that's activated by operand and moves through it's operation based on clock cycles. That description is likely wrong but it's not important, I gave 'it works stepwise' about 30 seconds of thought across both Von Neumann and Harvard architectures, and there you go.
So what am I proposing that works differently? An incredibly reduced instruction set that at the base level only manages timing and movement of bits or bytes into registers. The register set though, is the entire memory space. Why? Because each operation the computer is capable of is a separate module keyed to individual memory locations for input and output, and is running in a constant state. That adder, it's always functioning. Tremendous power hog, but that's not the point... the point is with modules keyed to input registers, the moment you complete moving bytes to those registers, the output is already calculated because it's always operating. "Programming" is then as easy as knowing which output registers hold the results of each module, and plucking that and putting it in the next step. Chain modules are possible too because there's no reason the output of one module can't be the direct input of another.
Consider you have an add, subtract, multiply, shift, and invert module all wired for the same two input registers. Your program moves a byte to one input, and to the other input, and after an appropriate wait for the result to settle, you already have the result for each of those operations ready in their respective result register, the program's next step is simply to pluck the desired result and put it where it needs to go next.
There's no reason you couldn't have multiple operational modules with the same functionality, just watching different or even matrixed registers. if one so desired you could have three input registers, A B C, and three addition modules with three outputs D E F.... A+B=D, A+C=E, B+C=F, and by filling the three inputs you just automatically have all three outputs ready to use.
Modules could even be much more complex than simple operations, a module can be literally whatever you want taking inputs and providing outputs, all you need to know is which registers to fill and which registers to pull the answers from. Memory is simply another module with flipflops and more complex address decoding, that provided an output on the same location that was input (which would actually require a select line but only so it's not contending for the bus to return the value, as far as the CPU is concerned the wait between poke and pull is managed the same)
Harvard would be easier on this with the amount of addressing that goes on in this schema, the 'on the ground' device would be a backplane with a 'control card' and any number of modules, and power draw would be the simple addition of power draw for every single part because they're all running constantly.
It would need a compiler and the control board some circuitry for retrieving instructions from ROM space. The compiler would need to keep track of reserved registers by module activity, but I would be surprised if the actual logic control needed more than 16 instructions to accomplish everything, no matter the scale (depending on address bus width, you want 1024 modules? Go for it, prepare for the power bill from hades)
Now that it all works and is troubleshot and debuggified, the modularity of the thing would lend itself to building a cool midcentury retrofuturistic cabinet for it! Front panel toggle inputs, gauges, blinkenlightzen, and lots of fans!
Amazingly cool Been watching you build this step by step Gratz
Hooray!
Congratilations! As I know now your computer is world first modern vacuum tube computer which was FINISHED! All other projects were stucked or scraped.
I also have now idea when I would be done with my own...
Great Project !
Some (unwanted) suggestions:
1. Bell relay: Put a snubber diode across the coil, to prevent a high voltage spike when deactivated. Also, since you basically have a capacitor across the relay coil (as a snubber cap), make sure it is a snubber rated capacitor (or just remove it and replace it with the mentioned diode).
2. Since you power buss is so long, I would put some bypass capacitors every so often (.01-.1 uf).
3. Add some soft-start circuits, arranged in zones (several options on ebay).
Epic! A few years ago I stumbled across your first experiments with the MC14500 chip on a breadboard, as I was also building a homebrew on breadboards. Ever since I followed your UE1 build. Truely an awesome archievement. Congrats!
Congratulations! I loved watching this series, your unbridled enthousiasm and watching the UE1 go from crazy idea via lots of debugging to something actually working. Can't wait to see what the UE2 will look like!
With the beard coming back, I knew it would be successful 😅😅
Hi from Nashville, TN area, we also had a cold weather here too! Congratulation with your projects being done! One word: Wow!
It's been a fun journey. I hope we get to see the adjustments which you're planning to make to the UE1. Maybe you should design a progressive power-up into the UE1 so that it soft-starts different sections to avoid overloading the power supply?
At 7:30 the "like" counter rolled over to 666 so I hit the button to roll it to 667. You're welcome! 😁
Congrats on reaching this milestone!
Congrats! It’s been interesting & fun watching this project develop from (for me) fairly early on… an immense amount of work on your part with its fair share of highs and lows and now it’s there! And it looks like a piece of art on your wall too 😊
Congrats on completing your UE1 project and what an amazing prototype you have built ! Looking forward to more excellent videos.
Awesome! Next project suggestion: An 8-bit computer equivalent to an Z-80 in specs and instruction-set, but using only transistors.
Brilliant suggestion! And of course, it needs tons and tons of LEDs so every aspect of its operations can be monitored. Blinkenlights!
Check out the MOnSter 6502!
A Z80 would need more than double the transistor count though... That might be a bit too insane.
@@alpcns mmm yes, its probably too much - but one LED for every bit in every register?
@@alpcns would be hot to run CPM on it. or TRS-DOS.
@@SnakebitSTI You're right of course, but I just love the idea.If that was available as a kit I'd buy and build it. With tons of LEDs... add peripherals and run CP/M on it. Yum!
Congrats! And what a journey!
Thanks for sharing it with us.
You could tackle the power on surge quite easily: you could manually stage the bring-up, one board at a time. Modify your little connector PCBs to have a physical switch on the heater power line.
Congraturation on a job well done!
Congrats ! A job wel done. A wonderful 3 years watching your progress and I am looking forward to your next journey !
Congratulations! Cant wait for version 2!
Select your computer model:
- Laptop
- Desktop
- Floortop
- Wall mounted
Well done. I first learned about vacuum tubes in the 1960's when studying for my ham license. Glad to see they are still available and useful.
Love your enthusiasm, looking forward to seeing more.
Congratulations tube wizard Usagi ... your dedication is inspiring.
Onwards to the next crazy project!
Congratulations. I have been following these videos weekly since before the start of the UE1 project and it has been an amazing journey.
Many congratulations. It’s been a wild ride and I am excited for the next beast of a tube computer.
I thought you would fight that paper tape reader until the end of time. Glad to see it working! Well done!
David, it has been a pleasure riding along with you in the development of the UE1. Congratulations!🎉
The bell ringing from the last video has been living rent free in my head since then 🤩
A true triumph is making a real true retro computer by hand. The sheer complexity of working with tubes as your switches and all the variables of dealing with the materials like you've been through makes any more complex computers using tubes just an incredible engineering achievement. Thanks for sharing!
Congratulations on completion. How long until I can buy the kit to make my own ? ;-)
Congrats. What an inspiring homebrew project!!
Congratulations for this stunning job!!
The Bendix G-15 behind you showed you how to do it: raise the voltage gradually/in steps! And/or use some kind of stepping relay to power up the sections one at a time. Stereo amps/receivers having been dealing with that for a long time too.
5 hour since the upload and im seeing the likes just jump up. really nice to see all the support for your channel. I cant wait to see future projects
I like your energy level and the way you think, this is coming from a life long embedded HW/SW guy 🙂 The whole project is beyond amazing and lets us old enough to remember learning on tubes a great vicarious experience! Bravo! 👍👍
"I declare the UE1 (and 0x1000) OPEN!"
Awesome work Dave, both in computing terms and room heating terms.
3 years is fast! My own insane homebrew project was started about 4 years ago, and I haven't even started building the hardware yet... must get back to it soon!
Good job brother, it takes a lot of tenacity to complete that thing you did right there!
Should really try to make the next one as detailed as possible. I’d love to build the computer with you as you go along.
Gratz! Loved this project and I can't wait until the next project!
Congratulations, Dave! I look forward to seeing it sometime.
Congratulations!!! Been awesome seeing this thing come to life!
This is such an epic project. I'd love to see more of it in the future. Only today I finished my MC14500 computer following your series from a few years ago :)
COngrats with completing the UE1 :) It has been very interresting to follow the development of UE1
Your enthusiasm is infectious, your glee when things work is a joy. Keep doing what you do. Pontiac Turbo Firefly Forever!
Somethin that's used on large vac tubes is a preheat circuit
If you insert a power resistor in series with the heaters that gives you 50-70% of the power, then a relay contact across the resistor, you can let it warm up on low power
Wire the relay like a start/stop that's on motors, when you press a button it latches, but when the machine is turned off it drops out to low power ready for restart.
If you look at the top of each board, you'll see big honking power resistors. That's my soft-start mechanism! It's been there for a while and I cover it in a few previous episodes in the series, but even with that, the power supply tends to be right on the ragged edge during power up on a really cold day.
@@UsagiElectric But are those resistors just to drop the current to normal operation
my ideaa lowers that current so the heaters are barely on, not even glowing, untill they have warmed up sufficiently
Congrats! It's been fun watching the journey!
This makes me appreciate what went into building the old vacuum tube computers.
Well done Mr Usagi , has it really been 3 years wow, been with you since the first PCB ,on wards and upwards to the next.
Congratulation on a job well done! It was very interesting following you.
Congratulations for this amazing achievement.
Every CPU need Errata!
"I'm endeavouring, to construct a mnemonic memory circuit using stone-knives and bear-skins."
Congratulations getting the machine operational! And the bell is classic - love the callback to the other project.
A very long time ago in Elektor magazine, i read about inrush current limitation for audio amps with large dc reservoir capacitors. The solution was to insert series resistors with a bypass relay on a timer or based upon voltage drop across the resistor (as a form of current measurement).
Perhaps this solution would allow you to limit inrush current by increasing the filament circuit resistance presented to the power supply, limiting the inrush current to the filaments and, as a final benefit, preheating the filaments to prolong their life.
Excellent project. This has been a great ride. Make sure you documents all your latest updates so that you remember what you did 1 year from now.
Congratulations! This has been an interesting journey.
Congratulations!!! Labor of love for sure!
Awesome, it’s funny how quite often I find the journey more enjoyable than the results. It’s probably a good time to take inventory of what you’ve learned, reflect on that and absorb the learnings that you want to take forward. We tend to call that part a “retrospective”, do it physically, like write it out or, better, PowerPoint and document it. Develop your own best, and worst, practices that you want to have on the table for every future project.
Here’s an example of one I’ve added to be box of things I’ve learned that I want to reinforce remembering: ALWAYS check input voltages.
Absolutely loving the videos! Even though I don’t fully understand everything it’s a fun learning process 😀
PCB at 17:38 Looks like some of my Factorio factory sections, love it!
congratulations. your enthusiasm is inspiring. 🎉
This project has me itching to design a ternary computer (don't know why, but I've just always found the possibility of three-state gates fascinating). I'd love to do something physical, but I might even just write an emulator at this point. It would be cheaper and less time consuming.
Super cool project.