Recommend shorting the resistor side of the relay so there is no brief disconnected state while switching. Once switched the relay will just short out the resistor.
That's a brilliant idea and I have no idea why I didn't think about that! I think I'm going to rebuild the boards that hold the relays so I can switch over to some proper 50s automotive relays (to keep the aesthetic in line), when I do that I'll change up the connection so the resistor side connects directly to common. Thanks for the recommendation!
Thank you! I actually bought like 80 2D21s off eBay about a year ago because I wanted to learn more about thyratrons, and if I'm going to buy one, I might as well buy almost 100, haha. As a result though, I've been looking for use for them and this seemed like an excellent candidate!
Hey :) It's even easier if you use a SPST NO relay and use the contact to shunt the resistor that is hardwired into the circuit. That's how it's done in soft-start circuits for toroid transformers. By the way, you may look into some long heatup time tubes like a PY88 booster diode, they take like a minute or so which is plenty of time for other tubes to fully heat up. This tube can even pass a few tens of milliamps at low plate voltages, and I've used such a circuit (PY88 and a 24VDC DPDT relay powered from -50V bias supply) for automatic B+ delay on one 4xEL34 monoblock I rebuilt ca. 17 years ago.
Heya Keri! I'm actually planning a rework of the soft start to use different relays that will use pretty much the same idea! Though, I'll still be using SPDT, because I already had some Omron G2R-1s that I want to use, but I'll be connecting the NC together with the common, so that it works pretty much exactly the same as the shunt idea with an SPST. That's really interesting about the PY88, I hadn't thought about that one. It has a 30V filament, which is insane, and I'm not sure if I'm reading the datasheet right, but it looks like at 24V, it'll move something like 400mA which is epic for a non-thyratron!
@@UsagiElectric The way you wired the relay contacts there will be a momentary loss of power when switching the current limiter resistor out of the circuit. The way Keri suggests eliminates that problem. It is a one wire change to make it work.
Thank you so much! It's honestly getting difficult to film, haha. I had to put the camera tripod on top of the backboard the computer is mounted to in order to get enough height to fit it all in one shot. I'm going to have to start coming up with inventive ways to get the camera high enough to see the whole thing as it continues to grow!
Thank you so much! I'm super happy with how it's shaping up, I think it looks absolutely amazing. Ultimately, the goal is to get it to calculate the Fibonacci sequence. That seems like a super simple goal, but to build a computer that can do that requires getting almost all of the fundamentals for a Turing complete machine in place, so I figure that's a decent goal to aim for! I'm also hoping to get it mounted on a simple rack to make a nice little display piece. The microprocessor will be about 1m x 0.7m, so I want to build the memory into the same form factor and have two of the backboards on display on a single 2m tall rack. Like this: i.postimg.cc/ydMKXvjw/Rack.png
@@k-8520 While the MC14500 is definitely intended for use as a PLC, I'm hoping to build this into a more universal type of computer, with a proper ALU, general purpose registers, RAM, input, output, etc. The primary goal is to get it to where it can calculate the Fibonacci sequence, but I would like for it to be capable enough to be considered Turing complete at some point!
@@UsagiElectric Thats super awesome :) Is Fib achievable .. ie: it gets into big numbers pretty quick, though maybe its just shifting away and can fake it without actually having all the numerals in RAM .. never really looked into binary representation of Fib sequences.
@@skeezixcodejedi So, even though it's technically a 1-bit computer, I do plan on building the general purpose registers around 8-bits. Hold on to your butts, I'm about to spew a bunch of word soup here in the hopes that it makes some sort of sense, haha. I'm imagining something like 4-bits for processor control, 2-bits for register/memory select, and 3-bits for addressing. Also, using the processor "write" pin I can kind of get an extra bit for output control. This makes things super confusing and a little difficult to wrap your head around, but it means that I have a lot of freedom in how I set up memory access. For example (CPU instruction of 1000 or 1001 = output): XXXX 00 AAA => 8-bit General Purpose Register A XXXX 01 AAA => 8-bit General Purpose Register B XXXX 1X AAA => 16-bits of Random Access Memory 1000 00 AAA => 8-bit Output Register A 1000 01 AAA => 8-bit Output Register B To do Fibonacci with a memory layout like this, the pseudo program would look like this: ADD GPRA and GPRB 1-bit at a time storing result back into GPRA COPY GPRA to OUTA ADD GPRA and GPRB 1-bit at a time storing result back into GPRB COPY GPRB to OUTA LOOP This should calculate the Fibonacci sequence from 0 up to 255, storing the output into our output register each calculation. The output register will ultimately be connected to our display module, which should show the result in human readable fashion, or I’m hoping so at least, haha. At any rate, that was a really long winded way of saying, I think Fibonacci should be doable! And, by using IEN and OEN masking, I should be able to implement rudimentary jumps and conditional statements, which means ultimately, we should be considered technically Turing complete. That’s a seriously long ways in the future though, so we’ll take it one step at a time and slowly work our way towards it!
Thank you so much! I actually thought about using a delay relay tube, but most of those tubes are designed to be used with much higher voltage. Finding a 12V delay relay in a tube package is surprisingly difficult (or maybe I just didn't know how to search properly). But, I had some thyratrons already laying from precious experiments and the 12V relays were a leftover from working on the Bellett, so it was nice to build something using parts I already had on hand!
@@Lee-il5kc Yeah, it's surprisingly difficult to find anything form the 1950s and 1960s that can work on just 24V! I'm really kind of pushing the limits of the tubes by running them at such low voltages, but it means that despite my clumsiness, I've yet to shock myself, haha.
This is great. I love how you're 'exploiting' the way the tube heats up, which then slowly increases the current to the relay coil! That is awesome, I wouldn't have thought of that in a million years. One question though, Is there many issues with back-emf from the relays? Do you need some backwards-diode loop or something to 'flush' the 'excess' flux after the relay coils are disconnected? Or does the direct ground connected to the relay coils do the flushing? Great work, I look forward to seeing more videos. Edit: Fixed typos.
Thank you! What's really great about the thyratrons is that they don't conduct at all until the gas inside ionizes, so the relay doesn't see any current at all until the moment the thyratron ionizes, at which point it sees enough to kick over. I think ideally I would have a flyback diode to prevent a spike from suddenly turning the power off, but I totally forgot to put one in. I'm not sure how absolutely necessary it would be in this situation, but if I ever redesign these boards for different relays or something, I might put one in there just to be safe. Having said that, so far I haven't seen any ill-effects from not having one!
I really like this soft start solution. I know it is not entirely the feature of this video, but that little plug in LED diagnosis board is also absolutely a wonderful idea, and I feel inspired to try to make something like this, the next time I am into building a microcontroller circuit. Thank you for the video!
Thank you! The little LED board was just something I knocked up because I got tired of plugging a ton of wires into the breadboard, but it turned out way cooler than I was expecting! I especially love how they all glow and dim when the tubes are warming up in this beautiful and chaotic display, which really shows how each individual tube is warming up at a slightly different pace. Also, even though the layout isn't ideal, I do like the way the staggered LEDs with 4-bits per color are split. It's a shame that ultimately I plan on not having any LEDs on there, but with the jumper board setup like it is, I can plug this little diagnostic board in at any time, which should make for some pretty cool light shows whenever I need to demonstrate the inner workings of the system in a little more detail.
As this scales larger, you might have issues with this soft-start concept. One solution would be a cascade-delay idea, where the enable signal is passed from one stage to another to stagger the turn-turn on transient over a large number of sub-blocks, as the computer grows, just add more delay stages.
I think you might be right. For now, this will let me build and test a few more modules and boards before we run into issues again, but I do really like the cascading relay idea! If for nothing else than that it would sound epic, haha. But, I think it would work really well for getting it powered up progressively without causing too much stress on the PSU and while keeping the entire design simple yet expandable.
Time delay relays are so cool! In the future I may redesign the soft start bit at the top and switch over to a proper time delay relay like that. In this one, I was a little impatient and built the entire thing with parts I had laying around as extras from other projects!
A megaprocessor! Although, if we're being really honest, it's actually a space heater that can do simple calculations. So, the first "smart" space heater!
@@leyasep5919 Being Turing Complete is an actual goal though! The 1-bit architecture this is based on is pretty weird, but by having the IEN and OEN registers control whether data can be input or output, you can actually make pretty complex programs with jumps and conditional statements. Here's a short bit on how IF-THEN-ELSE statements are implemented with such a limited instruction set: i.postimg.cc/RVLwy5KK/If-Then-Else.jpg Once memory is implemented and we can store data in general purpose registers, then it comes down to being able to figure out how to write the appropriate programs to calculate whatever it is we're trying to calculate!
Nice project! Question: Isn't it simpler just to use a large capacitor for the delay and just to short the current limiting resistors with the relays? If you switch in this way, you won't have this "dead time" when the relay is switching. Greetings from Germany.
Thank you! The issue I was running into was that I wanted about a 10 second delay, and I was having difficulty finding a capacitor large enough to give that long of a delay. Also, I already had the thyratrons hanging around and the warm up delay of them was just about perfect, so this delay circuit was more about what was easiest for me to build with what I already had at hand rather than what was the best way to build it. But, I should definitely switch short the current limiting resistors with the relays, that was a bit of a design oversight on my part! Fortunately, it's not hugely important since the relays switch fast enough that the heaters don't really lose any of the initial heat they already built up. But, I'm pulling the boards off here soon to change out the current limiting resistor, and when I do that, I'll do some quick bodge wiring underneath to get rid of that dead time when the relay is switching!
I have a couple of questions. 1. Will your computer be located in a server rack or something like that? (to save space) 2. What is the name of the rabbit that is at the end of each video?
Sure thing! The computer will definitely be located in a rack of some kind. The backing board for the processor is 100cm by 70cm, so I'm going to use that size as the backing board for all the other aspects of the computer. Ultimately, I think I'll have four backing boards: processor, memory, program control, and output. At the moment, I'm planning to build a rack to hold two boards at an angle at a time, and ultimately have two of those racks. Something like this: i.postimg.cc/QM43nPYb/DualRack.jpg That should make it easy to access and see all the indicator VFDs, as well as give easy access for repairs or changes. Plus, I think it'll look great for taking to vintage computing shows, haha. The bunny at the end of most of the videos is named Koma, she's the one always sniffing my weird electronic contraptions looking for treats, haha! Sometimes I post up videos of our previous buns too. The light brown colored bun was our very first bunny we had a long time ago and she was named Kono. The tiny bun with really short ears was our second bunny and she was named Kome.
@@akkudakkupl I'm not sure about other Thyratrons, but the 2D21 has a relatively slow de-ionization time of between 35us to 75us. The ionization time is crazy fast though at 0.5us. At the voltages I'm at, I might be able to squeeze 10kHz out of it, but 5kHz should be totally doable. And a 5kHz thyratron buck regulator would epic!
The 6AS6 is a really cool and the perfect computing tube! Unfortunately, I don't have any of them. However, they're really pretty similar to a pentagrid converter/heptode which has two control grids (and two screen grids) stuffed in a single tube. I did a short 1-minute video on heptodes here: th-cam.com/video/XRRqC0llNtM/w-d-xo.html I would have loved to build the entire computer out of heptodes, but boy they're rare and expensive when compared to the 6AU6 I ended up using!
I have actually thought a bit about this! Thermistors are not nearly tough enough to handle the amount of power I'm drawing here, which means I'd have to redesign each individual board to contain its own thermistor. It's probably the more elegant solution, but I'm way too deep into the design now to change that over, so I'm just going to brute force my way through it, haha.
Thank you so much! I do try to go into as much detail as I can (reasonably) so that if anyone wants to try to use what I've come up with here and build their own version of it, they totally can. It means a lot to hear you're enjoying the content!
A few people have mentioned this idea to me, and I do really like the concept of it! The inductor would most likely have to be a custom wound piece in order to get the wire thickness and number of turns just right. Which unfortunately mean that as the processor grows and more tubes get added on, the custom wound inductor would need to be rebuilt. When the processor is completed it should be at right about 180 tubes, which is going to be mental in terms of in-rush current, so at that point a custom wound inductor may make a lot more sense!
Thank you for your reply! I effectively didn't consider that high currents are driving the whole circuit so I didn't think about needing a custom inductor and the thickness of the conductor itself. Now I appreciate your solution even more, as other people said already it is very clever! I'd still love to see the custom big-ass inductor at the end of the project, so that we can compare the two solutions and see which one performs better/is easier to implement! I'm learning a lot of things from your videos, you're very good at it and especially you're great at building excitement over the project! It is very apparent that you're passionate about older technology and the fascinating world of tubes! See you soon then with the next part, I'm excited!!!
@@aliceitc8380 Thank you! It's kind of counterintuitive to think of tube circuits as being low voltage high current affairs because they're really terrible at that, haha. But, I'm pretty happy with how well everything has been performing so far! I think once I get up to the full 180 tubes, the current soft start solution may not be able to handle it, so a custom high current inductor with a really low resistance may be the best solution! Plus, it would look super cool, haha. Thank you so much for the kind words on the videos and projects! I'm not an engineer at all (I went to school for language), so when the electrons do what I want them to, I get super excited, so I'm glad I can share that feeling of "I can't believe that actually worked"!
I'm sure you've shown before (?) ... what schematic capture tool you using? how are you making the boards? You've got so many custom guys there .. you're CNC carvinmg one sided copper boards? They're way nicer than hand drawn and etched, thats for sure; or perhaps using print out and transfer type? Given how fast you're knocking boards out, you're not waiting on shipping ... and you do seem to have a wicked workshop there, so surely a CNC. Also, the two tube slow start diagram you showed on paper.. could have been shaped like your Usagi logo :)
Thank you! To design all the PCBs I'm using a program called DesignSpark PCB ( www.rs-online.com/designspark/pcb-software ), for a free tool I feel like it's one of the best out there. I was able to get the design rules setup just right for the single sided PCBs I make and even make custom layouts that I copy paste from design to design for the tube sockets. To get the PCB design from DesignSpark into G-Code for our mill, I use a program called FlatCAM ( flatcam.org/ ). It took a little bit of time to get used to and get setup for the mill and bits that I use, but once I got all the parameters figured out, it works really well! Our mill is a Bridgeport EZ-Trak with an Acu-rite Millpwr CNC conversion. It's actually way too large and very nearly too old to be doing this level of precision PCB cutting on, but after a lot of trial and error I got to a method that works. All my PCBs are designed with pretty fat traces (1mm) and lots of space between traces to help with any inaccuracies in the mill (or laziness in my setup). The mill seems to be accurate to about 0.02mm, which is amazing, but not quite precise enough for narrow SMD stuff or anything like that. The PCBs that I cut are just single sided PCBs from Paramount 3D ( www.paramount-3d.com/prototypingboards ). I find these to cut really well, and I like the see through appearance they have after cutting a design out! Those little VFDs are actually Russian copies of the DM-160, which is designated IV-15. They're not super cheap, but boy do they look great!
Awesome stuff! Looks like you've managed to tackle that inrush current in a wonderful way. I love it! I've seen others mention that this timed relay thing might not cope with another hundred tubes, and that perhaps you need to split the relays across different sections of the computer. How about chaining several delays together, so that once the first section becomes powered, this activates a set of thyratrons and relays within that section, which then click on and power up a second section? Depending on the number of sections required to keep inrush current at safe levels, you could add additional delays in series with the different sections. Only issue would be that if you had a number of delays in series, it would take a while to get the whole thing warmed up... but I'm sure your power supply would thank you ;)
Thank you so much! I definitely think you're right in that this solution may not make it all the way to the 180 tubes the processor portion will eventually end up having. I do like the idea of multiple delays with different resistors that click clack their way to way to completion though! That's a relatively simple solution that should be pretty easy to implement. As it sits now, the boards that hold the relays are only half height, so if I need to stuff more relays or power resistors in that space, there is enough room to make new boards that expand upwards. I'll keep testing it and keep a close eye on it as the processor grows, so we may be revisiting this soft start again in the future!
@@UsagiElectric I just realised a slight flaw in the original idea I had - the thyratrons in the first stage would preheat with the other tubes and quickly trigger the second stage, probably resulting in a large current spike. Perhaps a double pole relay could work - one pole shorts across the preheat resistor for the first section, and the other pole is connected to the preheat resistor and thyratrons for the second section. These thyratrons then warm up, and trigger the relay to fully power the second section and power the preheating process for the next section (if needed). That way, the delays should remain consistent and safe. This is some hair-brained evening scheming, so hopefully you can make sense of this!
Well I finished binge watching this series, and can't wait for more. I don't understand half of it, but I love it. Best of luck and keep these coming! ;)
Thank you so much! We've still got a very long ways to go. Even after we finish the processor, we still need to tackle memory, program control and I/O, each of which is most likely going to be the same physical size as the processor!
I wonder if you might get some chatter/squegging on the -12V buss if the main load steals enough current to cool down the cathode at all once the relay engages..
I think the -12V rail should be alright since it's essentially got its own relay and thyratron. The -12V supply can only give about an amp of power, so the -12V bus sits below that currently. In all honesty, I totally didn't need a soft start for the -12V rail at the moment, but as the computer grows, I might get to a point where a soft start could help, so I went ahead and built it in just in case.
The ENIAC machine from 1944-ish had 18,000 vacuum tubes. Every time they turned it on, a few would blow. The solution they came up with was to leave it on all the time.
I love your mixture of simply filming diagrams on paper and "highlighted" animations... it makes your explanations very clear indeed. If I was doing this "in my world" I'd just use a 555 timer, but your "limitation" of having to use vacuum tubes has actually made a far simpler and more elegant solution. I had a few vacuum tube devices in the past and had always seen the warm-up time as an inconvenience, I love this because it "turns a bug into a feature". :)
Wow. For me, this is the second project "DIY Programmable Lamp Computer" on the entire world wide web, that I found. Both for the last 2 years. LOL, One dude makes a one-bit computer and other dude makes a "brainf*ck compatible pc" from decatrons. so different... . . . . . . . . . . . . . . . . . . . Very strange.. why only last 2 years... and only 2 projects in whole wwweb. Is this some kind of re-imagining of DIY retro computing electronics?
Oh, you're talking about Artem's PC projects! Both his Dekatron PC and Reed Relay PC are absolutely amazing! He's a seriously smart builder and I'm envious of his projects, they tend to make my little 1-bit project look like child's play, haha. Links for anyone else that may be reading along: hackaday.io/project/45538-dekatronpc hackaday.io/project/18599-brainfuckpc-relay-computer
@@UsagiElectric I apologize, very much, when I said "different", I did not mean that your project is somehow worse or simpler. Your project is as good as Artem's. (I've been watching Artem for a long time and his projects are just wonderful) I just can't put into words my admiration for your work. I really like the idea of a 1-bit computer and how you implement it (As you said, "... so one guy in a small room..." with VACUUM TUBE Computer XD), along the way explaining how the machine works. I wish you success in your project. . And thank you for the cool episode with the reverse engineering of the Motorola MC14500. As you said, "...truly amazing people". . . . ================================================================= Maybe one day it will be able to interpret the "Brainf*ck" language. (of course, using 8 bits for memory addressing)
@@k-8520 No need to apologize! I didn't take it that way at all! Artem is just doing really, really amazing stuff and his understanding of complex computing really blows my mind. Artem's a properly smart builder and I got tons of respect for the amazing projects he builds! Thank you for the super kind words on the 1-bit design I have here! I really wanted to try to build something I could go along step-by-step and follow the data around processor to see exactly what is happening and where. That and vacuum tubes are comparatively massive, so I needed something small enough to fit inside, haha. CuriousMarc and Ken Shirriff were absolutely amazing, and working with them to reverse engineer the MC14500 was one of the greatest experiences ever! Of course, now I've got a new goal of actually getting out to California one of these days to thank them in person (and see Marc's awesome collection of stuff).
hmm after watching the video something does not seems right to me. it works more as a delayed start used to power multiple separate sections from single power source so the power spike would spread in time over entire device than a soft start which would slowly bring voltage up and then switch over you still enabling entire circuit at once from the relay and those funny plasma tubes only dlay that, voltage spike might be smaller somewhat but a real soft start would switch over seamlessly whane the voltage on the output would reach certain level key components that are missing are capacitors for each power rails on the computer side and even larger resistors , not smaller so you produce voltage ramp and switch to full power after full voltage is achived i don't know how much the project got advanced by today and i might be totaly wrong but oh well, this is the internet
Outstanding content! But I do have to ask, just out of curiosity: Why do you pronounce the word “QUITE”, like “QUIET”? … As in, like, instead of saying, “QUITE a lot of current.”, you say, “QUIET a lot of current.”… Just curious as to why. Perhaps there’s some linguistic etymology here that I need to learn. Thanks for great content! Subscribed!
@@pahaahv … LOL! There’s being different, and not normal, and then there’s being just flat-out WRONG, or blatantly (or willfully) ignorant… If it serves zero purpose, there’s no point in a particular behavior or habit. That said, David here is WWWAYYY too smart to not know the correct pronunciation of the word “quite”, so I’m guessing he has a VERY GOOD REASON for pronouncing it “QUIET”. I’m just curious, and wanting to know WHY he has made that conscious decision. I am FASCINATED by language…
Thank you so much! And I have to say, this is probably the most difficult question I've been asked! I've been thinking about this for like a day now and honestly, I'm not sure why I mispronounce "quite," haha. I mean, I do know the difference between the two, and intellectually I know that "quite" is one syllable and "quiet" is two, but you're definitely right that I pronounce "quite" with two syllables. I've been thinking about this though, and I've got two explanations that might make a bit of sense. First, when i was kid I probably mispronounced "quite", just saying it the same as "quiet," and by the time I learned that it was pronounced differently my brain had already etched the mispronounced version in and old habits die hard. Or, I have been known to intentionally (though unconsciously) exaggerate syllables and weighting in words when speaking to place more emphasis on them. Kind of like placing a pause after "A" in "Amazing" so that ends up sounding like "A...MAZEing". "Quite" seems like a strange word to do that on, but by extending it out into two syllables, I can create emphasis on the fact that I'm talking about a lot of something without needing to switch to a different adjective. Probably terrible for accurate linguistic purposes, but that type of speaking style can be quite effective when talking in front of people, which was something I had to do for work for a fair few years. Either way, thank you for the question! I love thinking about language and linguistic patterns and all the weird intricacies of conversation that make absolutely no sense. It gets a lot wilder when trying to think about it for a second language too; Japanese is a hilariously difficult language, not just because of the stark differences from English, but also from the massive amount of non-verbal communication that is so different too!
@@UsagiElectric … Absolutely FASCINATING! Thanks! I too am fascinated by languages, and I am now working on my fourth language (English; native, Spanish, Korean, and LATIN, YES, LATIN! EXCEPTIONALLY DIFFICULT, and will soon also start GERMAN!). This doesn’t mean I’m very good at ANY of them. I’m actually, probably pretty terrible, BUT, they are still fascinating to me! And I certainly LOVE TO LEARN! I have a “good ear” so I catch things; nuances that other people don’t catch, and I may very well be the first person to notice your QUITE/QUIET nuance. To be sure, I hear things; accents for example, that other people don’t hear, that I can often pin down, not only to a state, but often to a particular CITY! Anyway, thanks very much for the fascinating explanation, and especially for such wonderful content!
@@stephaniejane1811 Oh man, language is one of those things I wish I had a knack for! I learned Japanese through sheer brute force and power of will, haha. Living in Japan for 7 years and using it on a daily basis as my primary language at work and in life made a massive difference though. Especially once I snagged a part time job on the weekends at a shop where no one spoke English. Although, probably the most important thing I learned while picking up Japanese is that rhythm and pronunciation is king. I discovered that in the office, even though some of my foreigner co-workers had far greater grammar and vocabulary skill than I did, I would often get other Japanese staff asking me to do things because I placed all my focus on learning rhythm and pronunciation. I think it made it easier to get the conversational "catch ball" going. Watching children's shows like Pokemon without subtitles made a huge difference in that too. The grammar and vocabulary was simple enough to follow, but the speed and conversational tones were much closer to real life. Speaking of which, now that I'm back in the states, I can feel the ability slipping, it's time to fire up Pokemon again and get into practice again. And actually, one of my personal goals is to start a series of YT videos talking about tubes and electronics in Japanese, so that may be a good opportunity to get some practice in too! That's awesome that you're learning so many languages! I can't imagine learning a fourth or fifth language, much less one that you don't use on a daily basis! In University I minored in Korean, but it's 10+ years since I used any Korean at all, so I've pretty much forgotten it all. Although, I'm curious what you think about my English accent! I think in the background of many of my videos there's a few Texas license plates floating about, so my state is obviously Texas, but having lived in Japan for so long before, I feel like my accent slipped into a more general American accent. At any rate, thank you for the awesome comments and questions!
Love the videos in this series, I built a 74 series based TTL SAP-1 for my final year project at University and your content has inspired me to start experimenting with vacuum tubes as a hobbyist!
Thank you so much! That's awesome that you built an SAP-1, that's a mega achievement! Definitely keep us updated with your future projects and if there's any insight I can offer on playing with tubes, don't hesitate to ask.
Thank you for the suggestion! For the moment, I think the delay is probably okay, it's just the heaters sit at such a low resistance, that the 5 ohm resistor sucks up all the power. We can see that once the heaters receive power, they only spike for about 1 second before they come up enough in resistance to no longer be a problem. With a smaller power resistor (say 2 ohms), I should be able to get more power in the heaters before the delay times out. I'll have a larger initial spike up to maybe 5A-ish when first kicking it on, but then it should stay at that level even after the relays kick over.
@@UsagiElectric yes ,you are right but the point of soft starting the filaments of tubes are generally to warm them up so they dont flash , in ur case it will help both low starting resistance and also the filament flash . Generally they use about a 10 seconds delay before kicking the main relay .
@@thevoidedwarranty Thank you so much! Soft starting is definitely good for filament flash, but I was actually planning on dealing with filament flash by just replacing any tubes that went bad, haha. I have a ton of 6AU6s (like over 1000 of them), so I kind of built the heater circuits a little fast and loose. I'm mostly just trying to give my power supply a fighting chance at surviving power on!
Amazing man, using the warm up time of the heaters as the time delay is incredibly simple and clever at the same time. Can't wait to see how the computer turns out. But also i don't want this series to ever end.
Thank you so much! I can't wait to see how it turns out too, haha. Don't worry though, the series still has a massive way to go. After we get the processor finished, we need to start in on memory, which will need some general purpose registers and some RAM, both of which are going to be serious hurdles. After that, it's onto program control, and I'm still thinking of ways to handle that one. Finally, I also want to build some kind of display output using old alphanumeric VFDs or something. It's going to be a massive project, but we'll take it one step at a time and slowly get there!
Thank you! Ultimately I think the processor is going to be at right around 180 tubes, so 'I'm not really sure how well this soft start will scale up to that level. I may have to do some more heavy redesigning of the soft start portion once we near completion. For the moment though, this lets me fire up quite a lot in one go so I can continue building and testing!
@@leyasep5919 Oh man, memory is going to be a serious challenge! I want the general purpose registers to be built from tubes, but that doesn't scale up well at all. I have a few ideas and plans I'm working on, so we'll see how we do!
@@UsagiElectric and you'll show us the results :-D Depending on the speed, you could use one technology or another. If you're not too fast, you could try dynamic RAM with 1 capacitor and 2 diodes per bit... But row/line decoders will still use a lot of tubes.
There were a couple reasons, but the biggest is because I already had some thyratrons and relays here and didn't need to order any parts. But also, I'm trying to keep the computer somewhat close to period correct for something like the LGP-30, which was built in 1956. NTC thermistors were around, but from what I've been able to read, they weren't all that common or commercially viable so they weren't often used. Building it with thyratrons and automotive relays felt a little more representative of what an enthusiast in the 60s could have accomplished in their garage! But mostly it's because I didn't need to order anything, haha.
Recommend shorting the resistor side of the relay so there is no brief disconnected state while switching. Once switched the relay will just short out the resistor.
That's a brilliant idea and I have no idea why I didn't think about that!
I think I'm going to rebuild the boards that hold the relays so I can switch over to some proper 50s automotive relays (to keep the aesthetic in line), when I do that I'll change up the connection so the resistor side connects directly to common.
Thanks for the recommendation!
thyratrons, not 555, not transistors, but thyratrons ... that's being self-consistent.
Also, TIL, that word; thats damned near the best component name ever. Also, The Transformers would love it.
Thank you!
I actually bought like 80 2D21s off eBay about a year ago because I wanted to learn more about thyratrons, and if I'm going to buy one, I might as well buy almost 100, haha. As a result though, I've been looking for use for them and this seemed like an excellent candidate!
Hey :)
It's even easier if you use a SPST NO relay and use the contact to shunt the resistor that is hardwired into the circuit.
That's how it's done in soft-start circuits for toroid transformers.
By the way, you may look into some long heatup time tubes like a PY88 booster diode, they take like a minute or so which is plenty of time for other tubes to fully heat up. This tube can even pass a few tens of milliamps at low plate voltages, and I've used such a circuit (PY88 and a 24VDC DPDT relay powered from -50V bias supply) for automatic B+ delay on one 4xEL34 monoblock I rebuilt ca. 17 years ago.
Heya Keri!
I'm actually planning a rework of the soft start to use different relays that will use pretty much the same idea! Though, I'll still be using SPDT, because I already had some Omron G2R-1s that I want to use, but I'll be connecting the NC together with the common, so that it works pretty much exactly the same as the shunt idea with an SPST.
That's really interesting about the PY88, I hadn't thought about that one. It has a 30V filament, which is insane, and I'm not sure if I'm reading the datasheet right, but it looks like at 24V, it'll move something like 400mA which is epic for a non-thyratron!
@@UsagiElectric The way you wired the relay contacts there will be a momentary loss of power when switching the current limiter resistor out of the circuit. The way Keri suggests eliminates that problem. It is a one wire change to make it work.
Nice work! Looks like your build has overtaken mine for desk requirements!
Thank you so much!
It's honestly getting difficult to film, haha. I had to put the camera tripod on top of the backboard the computer is mounted to in order to get enough height to fit it all in one shot. I'm going to have to start coming up with inventive ways to get the camera high enough to see the whole thing as it continues to grow!
Clearly, this is art -- and its mounted; when this is done, you can just hang it on a wall and have it drive a clock or something :P
Open and close garrage door.. Open and close all windows.. am... A! Turn tv, when nobody at home.. am.... what else we can do with plc?
Thank you so much!
I'm super happy with how it's shaping up, I think it looks absolutely amazing. Ultimately, the goal is to get it to calculate the Fibonacci sequence. That seems like a super simple goal, but to build a computer that can do that requires getting almost all of the fundamentals for a Turing complete machine in place, so I figure that's a decent goal to aim for!
I'm also hoping to get it mounted on a simple rack to make a nice little display piece. The microprocessor will be about 1m x 0.7m, so I want to build the memory into the same form factor and have two of the backboards on display on a single 2m tall rack. Like this: i.postimg.cc/ydMKXvjw/Rack.png
@@k-8520 While the MC14500 is definitely intended for use as a PLC, I'm hoping to build this into a more universal type of computer, with a proper ALU, general purpose registers, RAM, input, output, etc. The primary goal is to get it to where it can calculate the Fibonacci sequence, but I would like for it to be capable enough to be considered Turing complete at some point!
@@UsagiElectric Thats super awesome :)
Is Fib achievable .. ie: it gets into big numbers pretty quick, though maybe its just shifting away and can fake it without actually having all the numerals in RAM .. never really looked into binary representation of Fib sequences.
@@skeezixcodejedi So, even though it's technically a 1-bit computer, I do plan on building the general purpose registers around 8-bits. Hold on to your butts, I'm about to spew a bunch of word soup here in the hopes that it makes some sort of sense, haha.
I'm imagining something like 4-bits for processor control, 2-bits for register/memory select, and 3-bits for addressing. Also, using the processor "write" pin I can kind of get an extra bit for output control. This makes things super confusing and a little difficult to wrap your head around, but it means that I have a lot of freedom in how I set up memory access.
For example (CPU instruction of 1000 or 1001 = output):
XXXX 00 AAA => 8-bit General Purpose Register A
XXXX 01 AAA => 8-bit General Purpose Register B
XXXX 1X AAA => 16-bits of Random Access Memory
1000 00 AAA => 8-bit Output Register A
1000 01 AAA => 8-bit Output Register B
To do Fibonacci with a memory layout like this, the pseudo program would look like this:
ADD GPRA and GPRB 1-bit at a time storing result back into GPRA
COPY GPRA to OUTA
ADD GPRA and GPRB 1-bit at a time storing result back into GPRB
COPY GPRB to OUTA
LOOP
This should calculate the Fibonacci sequence from 0 up to 255, storing the output into our output register each calculation. The output register will ultimately be connected to our display module, which should show the result in human readable fashion, or I’m hoping so at least, haha.
At any rate, that was a really long winded way of saying, I think Fibonacci should be doable! And, by using IEN and OEN masking, I should be able to implement rudimentary jumps and conditional statements, which means ultimately, we should be considered technically Turing complete. That’s a seriously long ways in the future though, so we’ll take it one step at a time and slowly work our way towards it!
Very clever with using the heaters as time delay. I'd have used a delay relay tube and wouldn't have been as elegant. :3
Thank you so much!
I actually thought about using a delay relay tube, but most of those tubes are designed to be used with much higher voltage. Finding a 12V delay relay in a tube package is surprisingly difficult (or maybe I just didn't know how to search properly). But, I had some thyratrons already laying from precious experiments and the 12V relays were a leftover from working on the Bellett, so it was nice to build something using parts I already had on hand!
@@UsagiElectric Yeee, the delay relay tubes I've fooled with were all 120v coils. :
@@Lee-il5kc Yeah, it's surprisingly difficult to find anything form the 1950s and 1960s that can work on just 24V! I'm really kind of pushing the limits of the tubes by running them at such low voltages, but it means that despite my clumsiness, I've yet to shock myself, haha.
This is great. I love how you're 'exploiting' the way the tube heats up, which then slowly increases the current to the relay coil! That is awesome, I wouldn't have thought of that in a million years. One question though, Is there many issues with back-emf from the relays? Do you need some backwards-diode loop or something to 'flush' the 'excess' flux after the relay coils are disconnected? Or does the direct ground connected to the relay coils do the flushing? Great work, I look forward to seeing more videos.
Edit: Fixed typos.
Thank you!
What's really great about the thyratrons is that they don't conduct at all until the gas inside ionizes, so the relay doesn't see any current at all until the moment the thyratron ionizes, at which point it sees enough to kick over.
I think ideally I would have a flyback diode to prevent a spike from suddenly turning the power off, but I totally forgot to put one in. I'm not sure how absolutely necessary it would be in this situation, but if I ever redesign these boards for different relays or something, I might put one in there just to be safe. Having said that, so far I haven't seen any ill-effects from not having one!
@@UsagiElectric you could "see" the effect with your latest oscilloscope and take the appropriate measures :-)
I really like this soft start solution.
I know it is not entirely the feature of this video, but that little plug in LED diagnosis board is also absolutely a wonderful idea, and I feel inspired to try to make something like this, the next time I am into building a microcontroller circuit.
Thank you for the video!
Thank you!
The little LED board was just something I knocked up because I got tired of plugging a ton of wires into the breadboard, but it turned out way cooler than I was expecting! I especially love how they all glow and dim when the tubes are warming up in this beautiful and chaotic display, which really shows how each individual tube is warming up at a slightly different pace.
Also, even though the layout isn't ideal, I do like the way the staggered LEDs with 4-bits per color are split. It's a shame that ultimately I plan on not having any LEDs on there, but with the jumper board setup like it is, I can plug this little diagnostic board in at any time, which should make for some pretty cool light shows whenever I need to demonstrate the inner workings of the system in a little more detail.
As this scales larger, you might have issues with this soft-start concept. One solution would be a cascade-delay idea, where the enable signal is passed from one stage to another to stagger the turn-turn on transient over a large number of sub-blocks, as the computer grows, just add more delay stages.
I think you might be right. For now, this will let me build and test a few more modules and boards before we run into issues again, but I do really like the cascading relay idea! If for nothing else than that it would sound epic, haha. But, I think it would work really well for getting it powered up progressively without causing too much stress on the PSU and while keeping the entire design simple yet expandable.
3:40 - I was reminded of the time-delay relay tubes,something like 6NO30.
Time delay relays are so cool! In the future I may redesign the soft start bit at the top and switch over to a proper time delay relay like that. In this one, I was a little impatient and built the entire thing with parts I had laying around as extras from other projects!
More like macroprocessor
or a processor...
A megaprocessor!
Although, if we're being really honest, it's actually a space heater that can do simple calculations. So, the first "smart" space heater!
@@UsagiElectric a bold claim indeed 😁
@@UsagiElectric how can it be "smart" if it's not even Turing Complete ?
@@leyasep5919 Being Turing Complete is an actual goal though!
The 1-bit architecture this is based on is pretty weird, but by having the IEN and OEN registers control whether data can be input or output, you can actually make pretty complex programs with jumps and conditional statements. Here's a short bit on how IF-THEN-ELSE statements are implemented with such a limited instruction set: i.postimg.cc/RVLwy5KK/If-Then-Else.jpg
Once memory is implemented and we can store data in general purpose registers, then it comes down to being able to figure out how to write the appropriate programs to calculate whatever it is we're trying to calculate!
Nice project!
Question:
Isn't it simpler just to use a large capacitor for the delay and just to short the current limiting resistors with the relays? If you switch in this way, you won't have this "dead time" when the relay is switching.
Greetings from Germany.
Thank you!
The issue I was running into was that I wanted about a 10 second delay, and I was having difficulty finding a capacitor large enough to give that long of a delay. Also, I already had the thyratrons hanging around and the warm up delay of them was just about perfect, so this delay circuit was more about what was easiest for me to build with what I already had at hand rather than what was the best way to build it.
But, I should definitely switch short the current limiting resistors with the relays, that was a bit of a design oversight on my part! Fortunately, it's not hugely important since the relays switch fast enough that the heaters don't really lose any of the initial heat they already built up. But, I'm pulling the boards off here soon to change out the current limiting resistor, and when I do that, I'll do some quick bodge wiring underneath to get rid of that dead time when the relay is switching!
I have a couple of questions.
1. Will your computer be located in a server rack or something like that?
(to save space)
2. What is the name of the rabbit that is at the end of each video?
Sure thing!
The computer will definitely be located in a rack of some kind. The backing board for the processor is 100cm by 70cm, so I'm going to use that size as the backing board for all the other aspects of the computer. Ultimately, I think I'll have four backing boards: processor, memory, program control, and output. At the moment, I'm planning to build a rack to hold two boards at an angle at a time, and ultimately have two of those racks. Something like this: i.postimg.cc/QM43nPYb/DualRack.jpg
That should make it easy to access and see all the indicator VFDs, as well as give easy access for repairs or changes. Plus, I think it'll look great for taking to vintage computing shows, haha.
The bunny at the end of most of the videos is named Koma, she's the one always sniffing my weird electronic contraptions looking for treats, haha! Sometimes I post up videos of our previous buns too. The light brown colored bun was our very first bunny we had a long time ago and she was named Kono. The tiny bun with really short ears was our second bunny and she was named Kome.
Amazing work done! Very impressive milestone achieved.
Also, nice delay mechanism is used.
Thank you so much!
It's slowly making progress and getting more and more functionality in it!
Thyratrons call for tube buck regulator :D
A long term goal of mine is to someday build a buck regulator like the thyraton PSU used in CuriousMarc's Teletype!
@@UsagiElectric i thought something closer to 10kHz than 100Hz, but that PSU is also amazing. It needs a lot of iron though 😉
@@akkudakkupl I'm not sure about other Thyratrons, but the 2D21 has a relatively slow de-ionization time of between 35us to 75us. The ionization time is crazy fast though at 0.5us. At the voltages I'm at, I might be able to squeeze 10kHz out of it, but 5kHz should be totally doable. And a 5kHz thyratron buck regulator would epic!
6AS6 TUBE CAN BE CONTROLLED FROM GRID #1 AND THE SUPPRESSOR GRID #3. CAN YOU DO A DEMO VIDEO?
The 6AS6 is a really cool and the perfect computing tube!
Unfortunately, I don't have any of them. However, they're really pretty similar to a pentagrid converter/heptode which has two control grids (and two screen grids) stuffed in a single tube. I did a short 1-minute video on heptodes here: th-cam.com/video/XRRqC0llNtM/w-d-xo.html
I would have loved to build the entire computer out of heptodes, but boy they're rare and expensive when compared to the 6AU6 I ended up using!
Would thermistors suffice for the soft start? I know it's not valve/tube driven. 😃
I have actually thought a bit about this! Thermistors are not nearly tough enough to handle the amount of power I'm drawing here, which means I'd have to redesign each individual board to contain its own thermistor. It's probably the more elegant solution, but I'm way too deep into the design now to change that over, so I'm just going to brute force my way through it, haha.
I'm not sure I've seen that thorough and understandable tutorials and explanations on TH-cam before. Beautiful content!
Thank you so much!
I do try to go into as much detail as I can (reasonably) so that if anyone wants to try to use what I've come up with here and build their own version of it, they totally can.
It means a lot to hear you're enjoying the content!
couldn't you use an inductor to limit the transient currents? What would have been the downsides to that?
A few people have mentioned this idea to me, and I do really like the concept of it! The inductor would most likely have to be a custom wound piece in order to get the wire thickness and number of turns just right. Which unfortunately mean that as the processor grows and more tubes get added on, the custom wound inductor would need to be rebuilt. When the processor is completed it should be at right about 180 tubes, which is going to be mental in terms of in-rush current, so at that point a custom wound inductor may make a lot more sense!
Thank you for your reply! I effectively didn't consider that high currents are driving the whole circuit so I didn't think about needing a custom inductor and the thickness of the conductor itself.
Now I appreciate your solution even more, as other people said already it is very clever! I'd still love to see the custom big-ass inductor at the end of the project, so that we can compare the two solutions and see which one performs better/is easier to implement!
I'm learning a lot of things from your videos, you're very good at it and especially you're great at building excitement over the project! It is very apparent that you're passionate about older technology and the fascinating world of tubes!
See you soon then with the next part, I'm excited!!!
@@aliceitc8380 Thank you!
It's kind of counterintuitive to think of tube circuits as being low voltage high current affairs because they're really terrible at that, haha. But, I'm pretty happy with how well everything has been performing so far!
I think once I get up to the full 180 tubes, the current soft start solution may not be able to handle it, so a custom high current inductor with a really low resistance may be the best solution! Plus, it would look super cool, haha.
Thank you so much for the kind words on the videos and projects! I'm not an engineer at all (I went to school for language), so when the electrons do what I want them to, I get super excited, so I'm glad I can share that feeling of "I can't believe that actually worked"!
I'm sure you've shown before (?) ... what schematic capture tool you using? how are you making the boards? You've got so many custom guys there .. you're CNC carvinmg one sided copper boards? They're way nicer than hand drawn and etched, thats for sure; or perhaps using print out and transfer type? Given how fast you're knocking boards out, you're not waiting on shipping ... and you do seem to have a wicked workshop there, so surely a CNC.
Also, the two tube slow start diagram you showed on paper.. could have been shaped like your Usagi logo :)
Which part number ids that 1 bit VFD? I need to buy a pile of those :O
Thank you!
To design all the PCBs I'm using a program called DesignSpark PCB ( www.rs-online.com/designspark/pcb-software ), for a free tool I feel like it's one of the best out there. I was able to get the design rules setup just right for the single sided PCBs I make and even make custom layouts that I copy paste from design to design for the tube sockets.
To get the PCB design from DesignSpark into G-Code for our mill, I use a program called FlatCAM ( flatcam.org/ ). It took a little bit of time to get used to and get setup for the mill and bits that I use, but once I got all the parameters figured out, it works really well!
Our mill is a Bridgeport EZ-Trak with an Acu-rite Millpwr CNC conversion. It's actually way too large and very nearly too old to be doing this level of precision PCB cutting on, but after a lot of trial and error I got to a method that works. All my PCBs are designed with pretty fat traces (1mm) and lots of space between traces to help with any inaccuracies in the mill (or laziness in my setup). The mill seems to be accurate to about 0.02mm, which is amazing, but not quite precise enough for narrow SMD stuff or anything like that.
The PCBs that I cut are just single sided PCBs from Paramount 3D ( www.paramount-3d.com/prototypingboards ). I find these to cut really well, and I like the see through appearance they have after cutting a design out!
Those little VFDs are actually Russian copies of the DM-160, which is designated IV-15. They're not super cheap, but boy do they look great!
Awesome stuff!
Looks like you've managed to tackle that inrush current in a wonderful way. I love it!
I've seen others mention that this timed relay thing might not cope with another hundred tubes, and that perhaps you need to split the relays across different sections of the computer.
How about chaining several delays together, so that once the first section becomes powered, this activates a set of thyratrons and relays within that section, which then click on and power up a second section?
Depending on the number of sections required to keep inrush current at safe levels, you could add additional delays in series with the different sections.
Only issue would be that if you had a number of delays in series, it would take a while to get the whole thing warmed up... but I'm sure your power supply would thank you ;)
Thank you so much!
I definitely think you're right in that this solution may not make it all the way to the 180 tubes the processor portion will eventually end up having. I do like the idea of multiple delays with different resistors that click clack their way to way to completion though! That's a relatively simple solution that should be pretty easy to implement. As it sits now, the boards that hold the relays are only half height, so if I need to stuff more relays or power resistors in that space, there is enough room to make new boards that expand upwards.
I'll keep testing it and keep a close eye on it as the processor grows, so we may be revisiting this soft start again in the future!
@@UsagiElectric I just realised a slight flaw in the original idea I had - the thyratrons in the first stage would preheat with the other tubes and quickly trigger the second stage, probably resulting in a large current spike.
Perhaps a double pole relay could work - one pole shorts across the preheat resistor for the first section, and the other pole is connected to the preheat resistor and thyratrons for the second section. These thyratrons then warm up, and trigger the relay to fully power the second section and power the preheating process for the next section (if needed).
That way, the delays should remain consistent and safe.
This is some hair-brained evening scheming, so hopefully you can make sense of this!
Well I finished binge watching this series, and can't wait for more. I don't understand half of it, but I love it. Best of luck and keep these coming! ;)
Thank you so much!
We've still got a very long ways to go. Even after we finish the processor, we still need to tackle memory, program control and I/O, each of which is most likely going to be the same physical size as the processor!
I wonder if you might get some chatter/squegging on the -12V buss if the main load steals enough current to cool down the cathode at all once the relay engages..
I think the -12V rail should be alright since it's essentially got its own relay and thyratron. The -12V supply can only give about an amp of power, so the -12V bus sits below that currently. In all honesty, I totally didn't need a soft start for the -12V rail at the moment, but as the computer grows, I might get to a point where a soft start could help, so I went ahead and built it in just in case.
The ENIAC machine from 1944-ish had 18,000 vacuum tubes. Every time they turned it on, a few would blow. The solution they came up with was to leave it on all the time.
I love your mixture of simply filming diagrams on paper and "highlighted" animations... it makes your explanations very clear indeed.
If I was doing this "in my world" I'd just use a 555 timer, but your "limitation" of having to use vacuum tubes has actually made a far simpler and more elegant solution.
I had a few vacuum tube devices in the past and had always seen the warm-up time as an inconvenience, I love this because it "turns a bug into a feature". :)
Wow. For me, this is the second project "DIY Programmable Lamp Computer" on the entire world wide web, that I found. Both for the last 2 years. LOL, One dude makes a one-bit computer and other dude makes a "brainf*ck compatible pc" from decatrons. so different...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Very strange.. why only last 2 years... and only 2 projects in whole wwweb.
Is this some kind of re-imagining of DIY retro computing electronics?
Oh, you're talking about Artem's PC projects! Both his Dekatron PC and Reed Relay PC are absolutely amazing!
He's a seriously smart builder and I'm envious of his projects, they tend to make my little 1-bit project look like child's play, haha.
Links for anyone else that may be reading along:
hackaday.io/project/45538-dekatronpc
hackaday.io/project/18599-brainfuckpc-relay-computer
@@UsagiElectric
I apologize, very much, when I said "different", I did not mean that your project is somehow worse or simpler. Your project is as good as Artem's. (I've been watching Artem for a long time and his projects are just wonderful)
I just can't put into words my admiration for your work.
I really like the idea of a 1-bit computer and how you implement it (As you said, "... so one guy in a small room..." with VACUUM TUBE Computer XD), along the way explaining how the machine works.
I wish you success in your project.
.
And thank you for the cool episode with the reverse engineering of the Motorola MC14500.
As you said, "...truly amazing people".
.
.
.
=================================================================
Maybe one day it will be able to interpret the "Brainf*ck" language.
(of course, using 8 bits for memory addressing)
@@k-8520 No need to apologize! I didn't take it that way at all!
Artem is just doing really, really amazing stuff and his understanding of complex computing really blows my mind. Artem's a properly smart builder and I got tons of respect for the amazing projects he builds!
Thank you for the super kind words on the 1-bit design I have here! I really wanted to try to build something I could go along step-by-step and follow the data around processor to see exactly what is happening and where. That and vacuum tubes are comparatively massive, so I needed something small enough to fit inside, haha.
CuriousMarc and Ken Shirriff were absolutely amazing, and working with them to reverse engineer the MC14500 was one of the greatest experiences ever! Of course, now I've got a new goal of actually getting out to California one of these days to thank them in person (and see Marc's awesome collection of stuff).
hmm after watching the video something does not seems right to me. it works more as a delayed start used to power multiple separate sections from single power source so the power spike would spread in time over entire device than a soft start which would slowly bring voltage up and then switch over
you still enabling entire circuit at once from the relay and those funny plasma tubes only dlay that, voltage spike might be smaller somewhat but a real soft start would switch over seamlessly whane the voltage on the output would reach certain level
key components that are missing are capacitors for each power rails on the computer side and even larger resistors , not smaller so you produce voltage ramp and switch to full power after full voltage is achived
i don't know how much the project got advanced by today and i might be totaly wrong but oh well, this is the internet
awsome video
and great use of the heaters as a delay as we discussed
ps you have a pm on discord
Thank you!
Outstanding content! But I do have to ask, just out of curiosity: Why do you pronounce the word “QUITE”, like “QUIET”? … As in, like, instead of saying, “QUITE a lot of current.”, you say, “QUIET a lot of current.”… Just curious as to why. Perhaps there’s some linguistic etymology here that I need to learn. Thanks for great content! Subscribed!
I once hanged out with a pastor and he said "brekafast" instead of "breakfast". When I asked why does he say it like that, he asked "Why be normal?".
@@pahaahv … LOL! There’s being different, and not normal, and then there’s being just flat-out WRONG, or blatantly (or willfully) ignorant… If it serves zero purpose, there’s no point in a particular behavior or habit. That said, David here is WWWAYYY too smart to not know the correct pronunciation of the word “quite”, so I’m guessing he has a VERY GOOD REASON for pronouncing it “QUIET”. I’m just curious, and wanting to know WHY he has made that conscious decision. I am FASCINATED by language…
Thank you so much!
And I have to say, this is probably the most difficult question I've been asked! I've been thinking about this for like a day now and honestly, I'm not sure why I mispronounce "quite," haha. I mean, I do know the difference between the two, and intellectually I know that "quite" is one syllable and "quiet" is two, but you're definitely right that I pronounce "quite" with two syllables.
I've been thinking about this though, and I've got two explanations that might make a bit of sense.
First, when i was kid I probably mispronounced "quite", just saying it the same as "quiet," and by the time I learned that it was pronounced differently my brain had already etched the mispronounced version in and old habits die hard.
Or, I have been known to intentionally (though unconsciously) exaggerate syllables and weighting in words when speaking to place more emphasis on them. Kind of like placing a pause after "A" in "Amazing" so that ends up sounding like "A...MAZEing". "Quite" seems like a strange word to do that on, but by extending it out into two syllables, I can create emphasis on the fact that I'm talking about a lot of something without needing to switch to a different adjective. Probably terrible for accurate linguistic purposes, but that type of speaking style can be quite effective when talking in front of people, which was something I had to do for work for a fair few years.
Either way, thank you for the question! I love thinking about language and linguistic patterns and all the weird intricacies of conversation that make absolutely no sense. It gets a lot wilder when trying to think about it for a second language too; Japanese is a hilariously difficult language, not just because of the stark differences from English, but also from the massive amount of non-verbal communication that is so different too!
@@UsagiElectric … Absolutely FASCINATING! Thanks! I too am fascinated by languages, and I am now working on my fourth language (English; native, Spanish, Korean, and LATIN, YES, LATIN! EXCEPTIONALLY DIFFICULT, and will soon also start GERMAN!). This doesn’t mean I’m very good at ANY of them. I’m actually, probably pretty terrible, BUT, they are still fascinating to me! And I certainly LOVE TO LEARN! I have a “good ear” so I catch things; nuances that other people don’t catch, and I may very well be the first person to notice your QUITE/QUIET nuance. To be sure, I hear things; accents for example, that other people don’t hear, that I can often pin down, not only to a state, but often to a particular CITY! Anyway, thanks very much for the fascinating explanation, and especially for such wonderful content!
@@stephaniejane1811 Oh man, language is one of those things I wish I had a knack for! I learned Japanese through sheer brute force and power of will, haha. Living in Japan for 7 years and using it on a daily basis as my primary language at work and in life made a massive difference though. Especially once I snagged a part time job on the weekends at a shop where no one spoke English.
Although, probably the most important thing I learned while picking up Japanese is that rhythm and pronunciation is king. I discovered that in the office, even though some of my foreigner co-workers had far greater grammar and vocabulary skill than I did, I would often get other Japanese staff asking me to do things because I placed all my focus on learning rhythm and pronunciation. I think it made it easier to get the conversational "catch ball" going. Watching children's shows like Pokemon without subtitles made a huge difference in that too. The grammar and vocabulary was simple enough to follow, but the speed and conversational tones were much closer to real life. Speaking of which, now that I'm back in the states, I can feel the ability slipping, it's time to fire up Pokemon again and get into practice again. And actually, one of my personal goals is to start a series of YT videos talking about tubes and electronics in Japanese, so that may be a good opportunity to get some practice in too!
That's awesome that you're learning so many languages! I can't imagine learning a fourth or fifth language, much less one that you don't use on a daily basis! In University I minored in Korean, but it's 10+ years since I used any Korean at all, so I've pretty much forgotten it all. Although, I'm curious what you think about my English accent! I think in the background of many of my videos there's a few Texas license plates floating about, so my state is obviously Texas, but having lived in Japan for so long before, I feel like my accent slipped into a more general American accent.
At any rate, thank you for the awesome comments and questions!
Love the videos in this series, I built a 74 series based TTL SAP-1 for my final year project at University and your content has inspired me to start experimenting with vacuum tubes as a hobbyist!
Thank you so much!
That's awesome that you built an SAP-1, that's a mega achievement! Definitely keep us updated with your future projects and if there's any insight I can offer on playing with tubes, don't hesitate to ask.
Beautiful build! Nicely done :D
Thank you very much!
U need to add more delay to the relay ,u can add a resistor to the heaters of thyrotrons so they warmup and turn on slower .
Thank you for the suggestion!
For the moment, I think the delay is probably okay, it's just the heaters sit at such a low resistance, that the 5 ohm resistor sucks up all the power. We can see that once the heaters receive power, they only spike for about 1 second before they come up enough in resistance to no longer be a problem. With a smaller power resistor (say 2 ohms), I should be able to get more power in the heaters before the delay times out. I'll have a larger initial spike up to maybe 5A-ish when first kicking it on, but then it should stay at that level even after the relays kick over.
@@UsagiElectric yes ,you are right but the point of soft starting the filaments of tubes are generally to warm them up so they dont flash , in ur case it will help both low starting resistance and also the filament flash . Generally they use about a 10 seconds delay before kicking the main relay .
Huge fan of ur videos by the way, great job 👍👍
@@thevoidedwarranty Thank you so much!
Soft starting is definitely good for filament flash, but I was actually planning on dealing with filament flash by just replacing any tubes that went bad, haha. I have a ton of 6AU6s (like over 1000 of them), so I kind of built the heater circuits a little fast and loose. I'm mostly just trying to give my power supply a fighting chance at surviving power on!
@@UsagiElectric nice solution ! Those tube go for 10 to 20usd where i live ! (I live in Iran )
This is so amazingly cool! I love this :)
Thank you so much!
Simple & clever solution.
Thank you!
Amazing man, using the warm up time of the heaters as the time delay is incredibly simple and clever at the same time.
Can't wait to see how the computer turns out. But also i don't want this series to ever end.
Thank you so much!
I can't wait to see how it turns out too, haha. Don't worry though, the series still has a massive way to go. After we get the processor finished, we need to start in on memory, which will need some general purpose registers and some RAM, both of which are going to be serious hurdles. After that, it's onto program control, and I'm still thinking of ways to handle that one. Finally, I also want to build some kind of display output using old alphanumeric VFDs or something. It's going to be a massive project, but we'll take it one step at a time and slowly get there!
@@UsagiElectric i didn't realize how much things still had to be done. Seems we're in for a treat :)
YAY !
instead of reducing the resistance,
add another in parallel and power 2 separate power buses ?
Thank you!
Ultimately I think the processor is going to be at right around 180 tubes, so 'I'm not really sure how well this soft start will scale up to that level. I may have to do some more heavy redesigning of the soft start portion once we near completion. For the moment though, this lets me fire up quite a lot in one go so I can continue building and testing!
@@UsagiElectric it's going to be fun with the memory.
You had it easy with the processing part.
/me grabs some more popcorn
@@leyasep5919 Oh man, memory is going to be a serious challenge! I want the general purpose registers to be built from tubes, but that doesn't scale up well at all. I have a few ideas and plans I'm working on, so we'll see how we do!
@@UsagiElectric and you'll show us the results :-D
Depending on the speed, you could use one technology or another.
If you're not too fast, you could try dynamic RAM with 1 capacitor and 2 diodes per bit...
But row/line decoders will still use a lot of tubes.
Why not just an NTC?
This way is much cooler and uses a tube.
There were a couple reasons, but the biggest is because I already had some thyratrons and relays here and didn't need to order any parts.
But also, I'm trying to keep the computer somewhat close to period correct for something like the LGP-30, which was built in 1956. NTC thermistors were around, but from what I've been able to read, they weren't all that common or commercially viable so they weren't often used. Building it with thyratrons and automotive relays felt a little more representative of what an enthusiast in the 60s could have accomplished in their garage!
But mostly it's because I didn't need to order anything, haha.
@@UsagiElectric Fair enough. Great to learn about new things regardless!
Lol 😂 can we run Linux on this ?
Haha, someday I might be able to get it to output the word "Linux"!