Vacuum Tube Computer P.07 - Building the Worst OpAmp in the World
ฝัง
- เผยแพร่เมื่อ 1 ต.ค. 2024
- In this episode, we try to build a vacuum tube Operational Amplifier. It goes well by my standards, but my standards are incredibly low! The result is most probably the worst OpAmp in the World!
Artist: Shnabubula
Title: Earthbound “Psi Piano Omega”
ocremix.org/re...
Also, we’re now on Discord and Twitter, so join us!
Discord: / discord
Twitter: / usagielectric
Thanks for watching!
The observed closed-loop gain of your op-amp is about right given the low open-loop gain. In the first example, I eye-balled on your scope a gain of 0.9/0.5 = 1.8. If the open loop gain of the op-amp is 25, then Vo/Vi = A/(1+AB) where A = 25 and B = 1/2 giving Vo/Vi = 1.85, which is close enough. BTW - love your videos!
Thank you so much!
The closed loop gain did seem fairly close, although the open loop gain was laughably small in this first iteration. In some of my later iterations, I was able to get the open loop gain up to well over 5000, though that did introduce the tendency to oscillate pretty heavily if not kept in check. Still, the idea of building a low-voltage vacuum tube operation amplifier is just so much fun!
It can‘tbe the worst opamp, because these are in my Soundblaster…
Single-driver pentode test data:
"I've been running a few tests on some reasonably high-Gm pentodes, in pentode and triode mode. These are tubes that you could use as a single-tube driver for an SET, for example.
I've pretty much gone through all the likely suspects from my tube boxes. Most of the tubes I had can all be obtained for under $10.00, some (like the 12HL7) for as little as $0.70!"
www.pmillett.com/pentodes.htm
----------
Pentodes configured as triodes
by Tom Schlangen:
www.audiodesignguide.com/New2A3/ETF06TS.pdf
That pentodes strapped as triodes pdf has been a wonderful resource for a while! When I run a 6CB6 hooked up like that, it performs nearly as well as the 6AU6 and 6DJ8, it's really wild how differently the pentode can react!
"The worst op-amp... in the world." So what you're saying is, your op-amp is... _ambitious but rubbish._ Cue "Jessica". Roll main title.
Funnily enough, I was imagining James May uttering that reply. 😉
The resistor between each stage and the next together with the resistor from grid to ground are forming a voltage divider that reduces the signal by 1/2. This happens 4 times, so you're getting about 1/16 of the gain you could have if the coupling were more direct.
The resistor divider between each stage is actually incredibly important in order to bring the tube into full cutoff or saturation. Especially, at these super low voltages, I have to cause the voltage on the grid to swing between about -2V and +1V to push the tube close to saturation and cutoff. Additionally, I wanted this to be DC coupled, so I couldn't really use a more direct AC coupling capacitor, which means I'm limited to using a voltage divider as a level shifter to get that grid voltage level to where I need it.
Now, my resistor values in this example are all broken, and in the follow up episodes to this one I pick much better resistor values to get much more gain.
@@UsagiElectric I know that's important for digital use, but an op amp is a linear circuit, and you don't really want saturation or cutoff there. But you're probably doing about the best you can with only 24V. I expect it's easier with a higher B+, because then there's room to bias different stages at different levels and use direct coupling.
Your amplifier works well for what it is. It's just a regular amplifier, not an op amp. The low gain when you use feedback is because the gain isn't high enough.There are some things you can do to increase the gain.
This is the most important one: The resistor from the connected cathodes to the ground isn't working for you. People use high voltages and a high resistances to create a passive constant current sink. With low voltages, you'll have to make an active one.
I know very little about tubes. I think you can put a vacuum tube with the plate connected to the joined cathodes and the cathode connected to the 10k resistor. Connect a voltage divider between +24v, -12v, and the grid. Adjust the voltage until the top tubes are biased correctly.
This is less important: In transistor op amps it's important that the plate (collector) resistors match. This might be a place where using a high plate voltage reduces circuit complexity. Put the matching resistor in here and make sure the following stage works correctly.
Edit: Clarify parts of the explanation.
Great vacuum tube projects, well done! I am wondering why you are using 24V plate supply, is it for safety? Both analog and digital tube circuits have low gain, grid biasing related issues (as you are pointing out) at these low voltages. I'd go up to at least 50V plate supply, which is still safe for a short touch. And 12AX7 would be a good choice (if you have them on hand) since it has a high gain. Really enjoying your videos!
Great work, i love your videos!
I just got my first (high voltage) tube opamp working this week, it's just a clone of the k2-w.
I once read somewhere that in low voltage (space charge) mode, you can tie the pentode's control grid to positive voltage and then use the screen grid as control grid for more gain.
Have you tried this? It seems like a nice way to increase the gain and maybe even save some tubes in the design
Thank you so much!
That's awesome that you've got a clone of the K2-W. I really need to get my hands on one someday, but that +300V/-300V scares me, haha.
I have actually tried that, but I'm not sure if I'm getting something hooked up wrong or if the 6AU6 pentode just has a comparatively sparse screen grid, but I can't seem to get it to work at all. I've seen some designs that use the screen grid as a secondary input, essentially build a NAND gate out of a single pentode, but I get the same problems, the screen grid just doesn't have enough influence over the flow of electrons. But again, that could be due to the internal design of the 6AU6, some older octal pentodes may have better screen grids for this.
Fortunately, the 6AU6 is cheap enough that I don't mind using a few extra tubes to get a good low voltage design going (though in this video the design still isn't that great, it gets much better in the next episode).
@@UsagiElectric The high voltage is scary indeed. It's why i would very much like a better low voltage opamp. I'm excited to see the next video.
I don't have any octal pentodes but maybe about 50 or so noval philips pentodes and a lot of pentode/triodes. I will certainly let you know if i find anything interesting
Maybe it's in a previous video I haven't watched, but have you tried various combination of pentode grids for control to see if the hi/lo transition can be made sharper?
Here's a example from:
hackaday.io/project/169147-low-voltage-vacuum-tube-logic-gates
"V3 is supposedly half of V1 (not sure where I got that from), so this was produced by a potential divider formed of two 10K resistors to halve the input voltage. Additionally the two inputs were tied together to produce a NOT gate, which in turn was connected to the output of the OR gate (effectively pin A2 on the arduino)"
He got that information via a link to a forum that contains a link that no longer works (as happens far too often!). Here's the schematic he apparently got from that link - a pentode NAND gate:
cdn.hackaday.io/images/7226451577408254154.png
Which he built into this NAND gate with inputs tied to act as an inverter. Note the voltage divider providing plate voltage/2 on the screen grid:
cdn.hackaday.io/images/5242921577408800565.jpeg
However, someone commented:
"I cut my teeth on early vacuum-tube amplifiers and logic. A couple of comments here... In your pentode NAND gate, any grid can provide control but the effectiveness decreases as the grids progress from nearest to the cathode out... I think your NAND circuit would be more effective if you used the control and screen, or screen and suppressor grids as your inputs."
Of course, the screen/suppressor grid combination for input is not possible when the suppressor grid is internally connected to the cathode. In his 6EH7/EF183 pentode, it isn't.
On another point, here are some inexpensive and apparently fairly common tube types to try if you feel like it might be useful to do so. I realize you prefer to use the 6AU6/6CB6 types you're using because they're so incredibly common. Just a WAG, but some kind of triode/pentode cascade connection of these might give a sharper hi/lo transition:
6KA8/6KT8 High-Mu Triode with High-Gm Sharp-Cutoff Pentode
frank.pocnet.net/sheets/191/6/6KA8.pdf
frank.pocnet.net/sheets/135/6/6KT8.pdf
...or any of the many other high-mu triode high-Gm pentode types. They must have been very commonly used because there are so many types. It appears they were used in both B&W and color tube-based TVs. Another type:
6CM8 High-Mu Triode with High-Gm Sharp-Cutoff Pentode
frank.pocnet.net/sheets/127/6/6CM8.pdf
frank.pocnet.net/sheets/137/6/6CM8.pdf
In the Vacuum Tubes: Episode 8 Shootout, I see you already used something I was going to suggest - a resistor from the output to one of the grids to provide feedback.
Thanks for the absolutely awesome comment!
I have actually tried to use the control grid and screen grid as dual inputs for a NAND gate, but I never had all that much luck in getting it to work reliably. The screen grid is so much more loosely wound, that the inputs have to be treated fairly differently. I should probably give it another shot one of these days as it’s been a while since I last tried it. Now that I’m thinking about it, it might actually work a bit better to tie the suppressor grid and screen grid together, so that combined they can provide a little more control over the flow of electrons.
I do actually have about 30 each of the 6KT8 as well as the 6GH8, which are both pretty similar triode pentodes. Since they’re quite a bit more common and affordable than the 6DJ8, I really want to build some flip flops out of these someday. The pentode on the 6GH8, when triode strapped, performs almost identically to the triode half, so they should be able to do pretty well! It’s just another one of those projects I’ve been meaning to get around to. So much stuff to do, so little time, haha.
That looks SO much like my original Zenith/Heathkit 25MHz scope from my beginning learning days about 30-32 years ago! 😍😍😍
I love my little BK Precision 20 MHz scope! I use it all the time. The only thing I wish it had is a storage function, but finding an analog scope with a storage function is not easy. I think Tektronix were one of the only ones that were really working on that. So, whenever I need a proper storage function, I pull out the modern digital scope. But for 90% of the time, these old analog scopes are fantastic!
+24V and -12V? Nothing to see here. Lol.
Now i know what to do with my JAN 6418 tubes ..and i just got the SPICE model for it so i will first do a simulation
The 6418 is a really cool sub-mini pentode and should work excellently at low voltage!
Let me know how your project turns out!
Hey I was thinking and maybe you could use the tubes from AC/DC radios, because they are made for a voltage of 110v on plates, another valves for low voltage are the sub-miniature ones, made for military equipment, I saw the specs and they are rated on nice low voltages like a minimum of 45v in the case of a 5678 for example, and the last option and more difficult to find are the real 12v tubes for use in car radios from the battery, they use the "space-charge" tecnology like 12k5 or 12u7
Thanks for checking the video out!
Interestingly, the 6AU6 I'm using here is designed for around 100V on the plate (though the maximum is up to 300V+). But, in my experience, most tubes work really, really well at lower voltages. It's just hard to predict how they will react without actually testing them as most datasheets don't have any info for low voltage operation.
Still, I would love to get my hands on some sub-miniature tubes and even space charge tubes one of these days to see just how different they are to their high voltage counterparts when running at the crazy low voltages I'm at! The 12U7 you mentioned in particular is on my list of tubes to buy, because it would be awesome to compare against the 12AU7.
The tube car radios used a vibrator device, transformer etc. to get a B+ voltage in the mid-100 volts.
Loving this stuff! Any idea what you'd be able to do with low plate voltage (12V) automobile tubes?
www.junkbox.com/electronics/lowvoltagetubes.shtml
Thank you so much!
I genuinely need to get my hands on some space charge automotive tubes one of these days and do some proper comparisons! For example, the 12U7 I believe is the 12V version of the 12AU7, but I'm still not entirely sure on what's different inside them. I have some 12AU7s floating around, so it would be really good to get the two side by side and see just how differently they perform at low voltage. I'm really curious if they transition from saturation to cutoff any differently than the high voltage versions!
Hmm, I think it's time to start trawling eBay!
@@UsagiElectric I have been trying to replicate your results with the set reset flip flop using the 12U7 in place of the 6DJ8 since I either don't have or cant find a 6DJ8 in my stash of tubes. The 12U7 doesn't have the gain the 6DJ8 has. I can't seem to get it to work, I tried tweaking the resistor values but no luck. The only success I had with dual triodes was with the 6SN7 but it needed plus and minus 150 volts to work correctly. the only suitable indicators in this case are neon bulbs. I have only been able to get pentodes to work with low voltage so I see why you use the 6AU6 quite a bit. For giggles I pulled out the 12U7 and replaced it with a couple of NPN transistors and that worked without any tweaks. In fact I found I didn't even need the negative supply. If you would like some 12 volt car radio tubes to experiment with I would be happy to send you some maybe you could find a way to get them to work. They work great for building low voltage radio circuits and flea power guitar amps which is what I have used them for.
@@dans8287 Years ago I did low plate voltage (12 & 24V) plate curves on a large number of different dual triode tubes. The 6DJ8 and 6KN8 had twice the "current gain" of any other tube which I guess indicates why they are so expensive on eBay. The next best at half the current gain were 5BK7A, 6BS8, and 6BC8, with the 5BK7A's being the cheapest because of their modern(ish) logic voltage 5V heater. I've got a bunch of different automotive tubes including pentodes, but never tested them for relative current gain, a whole bunch of different subminiature tube types, and even a 6DS4 hi mu triode Nuvistor, but as usual, I've already got several other electronic projects I bounce around between in progress.
@@dans8287 Thanks for checking the video out!
I actually have a 12AU7 in my collection (which should be the HV equivalent to the 12U7), so I decided to pop it out and see if I could get it working.
It took a bit of trial and error, but I did finally get it working as an SR Flip Flop. I had to go real big on the resistor values, which means the output off the plate is extremely high impedance (33k ohm at 24V is just 0.7mA). Which means that for indiciators, there probably isn’t enough juice there to drive an LED without causing the flip flop to fall into instability. But, a cathode follower buffer will solve that problem with ease!
Here’s the values I ended up using: i.postimg.cc/B6xC1Jw1/Capture.jpg
This actually brings up an excellent conundrum for me. That worked far better than I thought it would, so I may have to rethink quite a lot of my designs, haha.
If you want, come join us over on the Discord, we often share little projects like this and it’s much easier to chat in real time than through the TH-cam comments (discord.gg/p7UsfHD)!
@@winstonsmith478 Yeah, the 6DJ8 is really an amazing tube, it performs so dramatically different than every other dual triode I've tried!
One of my favorites! Thank you 🙂
Thanks so much for checking it out!
I'm not quite done with this OpAmp though, I've been working hard to get it precise enough for proper applications!
The fact is, it works. Brilliant.
I'm actually surprised it worked as well as it did, given how much capability was left on the table. I've been refining it even further working towards a more compact version for the 555 kit, and it's crazy how much more capability I've been able to extract from the design by just refining resistor values!
@@UsagiElectric Yeah, Trial and error goes a long way.
Ooh, Lalá, Mr. analog computer!! 👌🏻⚡🤖
Haha, this OpAmp would make a terrible analog computer! (That's not gonna stop me from daydreaming about it though!)
@@UsagiElectric one could get the latest hi-tech wombo-combo gaming PC, but that would be missing the whole point here, right? 👀
Love the stuff you create and the approach it has. Great and entertaining videos!!
@@hmbrt12 Thank you so much for the awesome feedback!
I mean, I do have a fairly hardcore gaming PC sitting next to me, but it rarely brings me as much joy as getting these tubes to react how I want them to. There's just something that's so much fun about getting this ancient tech to do things it was never intended to do!
@@UsagiElectric exactly that's what we are talking about 💡🤖💡 Ancient tech ain't dead, man!!
Keep the good stuff coming please!