Vacuum Tube Computer P.02 - Heaters and Debounce 

Thank you! Switch bounce is one of those things that's hard to imagine but once you see it on the scope, it's clear as day! There are indeed quite some beefy 6V supplies out there that would work great for what we're doing, and the world of transformers is a rabbit hole of epic proportions, haha. But that Hammond transformer is wicked - 12V at 20A is mind blowing! I may actually have a look at some other Hammond transformers when it comes time to start thinking about power supplies for the full replica. I've never actually built a power supply from scratch, so that could be quite a fun little journey!

@@UsagiElectric I think for filament supply, we don't need anything fancy for regulation. In most tube guitar amplifiers, the transformer is a dual purpose filament supply as well as high voltage plate supply. The high voltage DC filtering supply is complicated, as we need high voltage capacitors and diodes and usually a choke to make a pi filter. But the filament output is a 6.3v centre tapped output from the transformer. They ground the center tap (that has to do with helping suppressing the AC hum artifacts making their way into the audio signal), and then they just feed the filament suppy from the transformer directly to the tubes with twisted 22 ga wires. Really all we would need to do here is have a proper and safe AC line voltage connectors. Something like an IEC power cord socket, a fuse, a switch, and then using the terminal lugs on the wires to connect these and some heat shrink tubing. I guess some metal chassis to hold it as well. This power supply design is kind of my hobby as well, I am happy to help if I can anyway.

Weirdly, this one showed up on the actual video this time! Maybe the RU-vid algorithm is finally learning... I have thought about running ballast resistors, but I'm not a huge fan of trying to dissipate all the heat that'll be generated by the resistor. I do really like the light bulb idea! I may end up giving that a shot for some future projects. The issue with the D-Flip Flop is that for the full-size MC14500 I'm building, everything will be broken up into modules cut into boards that are 150mm x 60mm, and I just barely managed to squeeze all six 6AU6 tubes for the D Flip Flop on one board, but there's no spare room at all. Having said that, I think I'm definitely going to experiment with the light bulb idea, it might save my bacon on a few other boards that use an odd number of tubes!

Thank you very much! I always found it weird that old tube schematics had an aversion to nF. It's not just IBM, I've seen radio schematics that used uuF. I imagine the engineers were like "this capacitor is small, micro even, but this one is even smaller, so, uh, micro micro?" I'm really glad I decided to start with this proof of concept because as I'm building it up, it's highlighting all sorts of issues that wouldn't scale up well or need to be tweaked. I'm hoping to start building the first bits of the full-size MC14500B replica around Christmas (although the videos will lag behind to give my poor mill some time to cut all the boards).

Me on the day of the apocalypse: "Weird, the internet is down, must be our country bumpkin DSL acting up again. Oh well, time to build more stuff with old tubes!" It'd probably take me a week to realize something has happened, haha. Also, not gonna lie, I thought about what my hypothetical vault number would be for way too long! I'm partial to "42". Aside from being the answer to life, the universe, and everything, according to the Fallout Wiki, it was supplied with light bulbs that were only 40 watts or less, which is perfect for seeing the glow from my vacuum tubes!

@@UsagiElectric 42 comes from the book “Hitch Hikers Guide to the Galaxy”.

@@UsagiElectric SURELY you could use 127.0.0.1 as your vault number to 'keep it local'???

Thank you so much! You're totally right, as long as the whole build is divisible by 4 (and I think it is, if I remember right the final tube count is going to be 180) I could run the tube heaters entirely off of the 24V rail. There's a few reasons I'm running two off of the -12V rail on the D Flip Flops though. The DFF that I'm using here is six logic gates, I did some testing with other types of flip flops that had fewer gates, but I had the best luck with the six gate ones. But, I wanted all six tubes to be on one board and I wanted that board to be as universal as possible. In the entire build I have seven of these six tube D Flip Flops, and all seven of the DFF boards are totally interchangeable. That meant that I needed to find a way to power all six on a single PCB. So, the primary reason is less of an electrical one and more of a physical packaging one!

@@UsagiElectric That makes sense, thanks for taking the time to reply!

This is actually an awesome question! As far as high voltage vs. low voltage goes, that one is tough because the manufacturers never intended the tubes to be used at just 24V, like we’re doing here. As a result, the charts and graphs on the data sheets just don’t have enough resolution to really see what’s going to happen. However, in my (albeit limited) testing, I’ve found that whether the tube is at a high voltage or a low voltage, the characteristics tend to pretty close to the same. The biggest difference is in how much power we can produce. That’s more a limit of the plate resistor and our low voltage than anything else though. For the characteristics, that depends a lot on what you’re building. If it’s an audio amplifier you want a big, wide linear curve. For the digital switching stuff we build here, we want something that transitions much quicker. Now, I’m still a massive noob at reading datasheets, so I may be completely wrong, but here’s my theory behind deciphering the graph data. So, let’s take a look at 12AX7 here: frank.pocnet.net/sheets/127/1/12AX7A.pdf On page 3 you can see the “plate voltage vs plate milliamperes” graph. I believe this is the best indication of how the tube will react. Pick a plate voltage and go up and down that line to see how the mA changes in accordance with grid voltage. Next, check out the 6AU6 here: frank.pocnet.net/sheets/127/6/6AU6.pdf Again, on page 3 you can see the same “plate voltage vs plate milliamperes” graph. The lines look dramatically different. For example, let’s look at the change in plate mA at 100V with the grid at -1V and then at 0V. On the 12AX7 we go from about 0.5mA to 2mA. On the 6AU6 we go from about 10mA to about 17mA. Now, the tubes are obviously extremely different, but it seems that the 6AU6 plate characteristics change much more dramatically with change in grid voltage. And I’ve kind of confirmed this with a bit of my own testing here: i.postimg.cc/N0PbZtS8/6-AU6-vs-12-AX7.jpg My testing was done at just 24V, but you can see that the 12AX7 output only goes from 23.5V to 20.5V as the grid changes from -1V to 0V. But, the 6AU6 goes from 19V to 8.5V. A massive change comparatively! This means the 6AU6 is way better for digital switching where I want the tube to either be in cutoff or saturation. But, for audio, where we want nice linear response, the 12AX7 is much better, even at lower voltages. Sorry for the crazy long winded answer, but that’s kind of where I get my line of thinking on tube choices!

@@UsagiElectric You might find this useful, I don't know: Vacuum Valves in Pulse Technique (1959) nvhrbiblio.nl/biblio/boek/Philips_ElectronicValves_BookIX_1959.pdf

@@winstonsmith478 Excellent resource! It's been saved in my resource folder and I'll give it a proper read soon! Thank you!

I agree 100%! A scope is one of those things I didn't think I needed until I got one and now I can't imagine not having one. Thanks for checking the videos out!

Thank you! The idea behind running the heaters in series with 24V DC is to eliminate the need for a separate tube power supply. The tube heaters are slightly undervolted (6V instead of 6.3V), but that doesn't really have an affect on the circuit at all. Since the tube heaters are in series, all I need are two power supplies: +24V and -12V. And, I can use the -12V supply to power two tube heaters in series as well, so I can design large circuits around two tube modules or four tube modules, giving a lot of flexibility in design while still keeping good simplicity in the power supply. With digital tube circuits where all we really care about is logic high or logic low signals, tubes at low voltages work wonderful. We can still see large gain, just at extremely low amperage. So, my standard inverting amplifier design uses a 33k plate resistor at 24V, which means at most I can supply just 0.7mA. But, if all I'm using that output for is for an input to another inverter, it's not a problem at all because there's no impedance mismatching going on. However, if I wanted to drive an audio amplifier, we start to run into issues. Even using the 6AU6 as a cathode follower driver for an output transformer, at most I'd be able to get about 200mW of power. So, that's generally why low voltage audio designs are not great.

@@UsagiElectric 6.3 VS 6 is nothing at all is still is considered normal. They typically list 6.3 volts because most 6 volt VDC power supplies used tube or selenium rectifiers that didn't have much voltage drop compared to a silicone rectifier. Typically in tube circuits to run a filiment cool would be 5 volts. Running them in series, isn't that how the old christmas tree lights worked. One bulb goes out the whole strain goes out?

@@lelandclayton5462 Yup! If one filament dies, it breaks the circuit. But, since I'm only ever running at most four tube filaments in series, if one tube does go bad, troubleshooting should be fairly simple (just feel around for the group of four that are cool).