**Lab 02** A Tube Amplifier: We Have Gain ![Figure [full_setup]: A Two-Tube Headphone Amplifier](./resources/overall.png) # Overview In this lab we'll build a relatively simple three-stage audio amplifier using two vacuum tubes: * The [12AT7 High-Mu Double Triode](./resources/12AT7.pdf) * The [12K5 Space Charge Tetrode](./resources/12K5.pdf) The circuit can drive either a small speaker to a kinda sorta reasonable volume or a pair of headphones at a loud volume. With this circuit you should be able to: * Plug in the waveform generator from the lab scopes and crank it up quite a bit to get it into headphones (or even a small speaker) * Plug in that weird tiny guitar I found on the loading dock years ago; it should be sitting around lab. This could arguably be a good "tube headphone amplifier" * Hook up your phone and play music at it, though you'll see the amp is probably less powerful than the built-in class D amp in your phone. * Plug in the diode output of your crystal set from Lab 1. If your amplifier gets tuned just right, you should be able to put together the front-end of lab 1 and this amp from lab 2 to have a completely-old-technology radio receiver. With all that said, don't expect wonders from this circuit (relative to what you've come to expect from a junk box op amp of nowadays). We have to do some serious *work* just to get a few mA here. I've designed this circuit to most importantly be electrically safe for us. The entire thing works on 12V, which is very, very low for a tube circuit. Running tubes with a B-voltage this low is sometimes called "starvation mode," and their behavior will differ in some ways from what you'd expect in a more standard circuit where B+ is 300V or something like that. Of particular note is that it is very, very tough to get much power from these things in these operating conditions. (the Grateful Dead's [Wall of Sound](https://en.wikipedia.org/wiki/Wall_of_Sound_(Grateful_Dead)) which caused early-onset hearing loss in all the band members did not achieve this by running in starvation mode.) With that said, I think this circuit isn't __too__ different from some classic tube topologies, and it'll maybe serve as a good sandbox for how to work with these weird devices. I've included four explicit adjustment points (in the form of potentiometers) that I think are pretty useful to mess with (and are arguably needed since each tube is going to be different in its behavior since I'm buying these second hand and they're a bunch of different brands, ages, etc...) Beyond even those potentiometers, just like Lab 1, a lot of the values in this circuit are actually pretty flexible. I encourage you to turn on a oscilloscope, and get out multiple probes and study how the signals propagates and gets manipulated as it moves through the circuit. Mess with values. What happens if you make that 47K resistor 390K? What works? What doesn't? How does the *bias point* change? How does the shape of a pure sine wave input change? Experiment! !!! WARNING We are using vacuum tubes in this lab! I've seen how your generation treats its cellphones. You can't do that with tubes; they'll break so you have to respect them. We are using them at low voltages so the biggest danger they present to you will be the potential to burn you (particularly the 12K5 since that consumes about 6W) or to cut yourself if you break the glass and then you stab yourself or eat it or something. So just be smart and don't abuse these things. The tubes must be used in sockets. I've pre-wired a few (and can make more as needed. In general tube pinouts and numbering shown as oriented from the bottom (since usually when working on a tube circuit you'd be working "underneath" the tube in the old chassis. For clarification, I've included this diagram below: ![Figure [pinouts]: Miniature Tube pinouts relevant for lab](./resources/tube_pins.jpg) This lab will go through the two main parts and discuss each component like in lab 1. The "special" components are up front as before including a selection of tubes...some in boxes and some in a bag/bubble wrap. You can't breadboard tubes directly so you'll need to use an appropriately-sized socket (9 pin for the 12AT7, 7 pin for the 12K5). I've pre-soldered some wired sockets for you, but if we're out you may need to make more. I have enough raw materials for sure, I'm just not sure how much time I'll have to make more wired sockets. Passives like caps and resistors are in the drawers like usual. Lecture 4 also went through a lot of general tube-circuit design stuff so make sure to jump back and review that since I'm not retyping it all here. # The Schematic The circuit can be broken down into two parts: a preamplifier that amplifies the *voltage* of an incoming audio signal using both halves of the 12AT7 and then what could loosely be called a power amplifier built around the 12K5 tube. This entire circuit, is running on a very low voltage for tubes. You'll see in studying the 12AT7 that its data sheet doesn't even show the voltage we're working at since it is so low (more on that below). The 12K5 is actually an anomaly of a tube developed later on in history specifically with 12V car batteries so it isn't as bad in a 12V B-voltage circuit, however even then, it is a power-amp *driver* meaning it was really meant to drive early semiconductor power transistors which would then drive a speaker in a car audio system rather than a speaker directly. We're going to use it to drive some earbuds and it actually is more than capable of that. We'll go over each portion of the circuit below. !!! WARNING You MUST use the benchtop supply for this circuit. 9V from a battery is too low. I encourage you to use the built-in current limit in the benchtop supplies. Set it to 0.7A and set the voltage to 12V. What will happen is when you turn the power supply on, the tube filaments will pull an initial large amount of current, but if there is a current limit in place it will cap it and this will actually help preserve the tubes for longer. At steady state the entire circuit should pull about 0.65A (0.15A from the 12AT7 and 0.5A from the 12K5). The circuit should really not pull more than 0.7 A. If you're seeing that the voltage is not getting to 12V with the 0.7A limit on the current, then you must have a short or a bad tube or something and you should investigate. Oh also as a reminder: !!! NOTE As stated in lecture, tubes take time to warm up. Once you power your circuit, expect that it won't start working in any reasonable way for ~20 seconds or so. The 12K5 has to run really hot and it takes a bit longer than the 12AT7 so you might see intermediate signals from the preamp show up 5 or 10 seconds before you start to see/hear signals from the output audio stage. That is to be expected. ## Preamplifier ![Figure [preamp]: The PreAmplifier](./resources/preamp.png) OK, how does this preamp work. Remember its goal is to amplify the voltage, not specifically the power (even though those are interrelated). * **The 12AT7 Dual Triode** is the heart of this circuit. Its characteristics are roughly expressed in the characteristic plots below: ![Figure [12at71]: The 12AT7 plate characteristics when the grid voltage is positive (ideally a region we want to avoid particularly when our plate voltage is so low) ](./resources/12at7_1.png width=50%) ![Figure [12at72]: The 12AT7 plate characteristics when the grid voltage is negative (ideally where we want to be) ](./resources/12at7_2.png width=50%) These plots all give us a rough idea of what to expect, but the problem with them is that they're really "zoomed out" for use in a standard tube circuit with 300V B voltage. That's not us. Thankfully we're not the first people to use our tubes in starvation mode, so [the Valve Wizard](http://www.valvewizard.co.uk/Triodes_at_low_voltages_Blencowe.pdf) has already done some very low sweeps on some commonly used tubes. ![Figure [12at73]: The 12AT7 plate characteristics zoomed in so that we get a sense of how it looks in the low plate voltage range we're operating in ](./resources/12at7_3.png) * **Capacitor $C_1$** acts as a DC blocking cap that passes AC (audio) into what is ultimately the grid of the first triode. We want to block DC here to prevent whatever we're hooking up from screwing with the biasing of our triode. * **Resistor $R_1$** is a "grid stopper" resistor. What it does is form a low pass filter with the parasitic capacitance that forms between the grid and the plate and the grid and the anode. I picked 1K sorta randomly. Feel free to increase or decrease. * **Resistors $R_2$ and $R_3$** form a grid leak path which allows us to achive a negative bias on the grid. How in the F does connecting the grid to 12V make the grid a negative value? See lecture 4. This trimmer pot $R_2$ should be varied by you to find the right operating point. Each tube is going to have a different behavior so you should use an oscilloscope and measure the voltage at the grid so that it gets centered into the negative voltage range. Since the cathode is tied directly to ground in this first triode, we **need** the voltage of the grid to be negative. Once your amplifier is working overall, go back and see how screwing with the grid leak will wreck your gain. * **Resistor $R_4$** is the plate resistor for the first gain stage. This has a large influence on the gain of this stage. I chose 47K. You should use that value to draw a line on the IV characteristics of the tube to see roughly where we're at in terms of operating point. !!! NOTE Check out lecture 4 since we talk about choosing the biasing resistor and its subsequent impacts! * **Capacitor $C_2$** is another DC-blocking cap to isolate the two triode stages so their biases don't get messed up. * **Potentiometer $R_5$** is a variable voltage divider that taps off a certain amount of output from the first stage to feed into the second. As a result it acts like a volume control. It also acts a bit as a grid leak for the grid of the second triode stage. * **Resistor $R_6$** is another grid stopper, this time for the grid on the second triode. * **Resistor $R_7$** is a cathode bias resistor. Its presence will force a bit of voltage to develop across it and if the grid is roughly at 0V, the positive cathode will therefore make the relative voltage of the grid with respect to cathode negative and allow for operation in the proper regime. This pot is tunable since the bias currents will vary from tube to tube and you will need to fiddle with this. * **Capacitor $C_3$** is a bypass capacitor which basically shunts out resistor $R_7$ at audio frequencies allowing the gain to be higher than if it were only the resistor there. This avoids what is termed "cathode degeneration." * **Resistor $R_8$** is the gain resistor for this second stage. I might have made this one too high, but if you adjust your bias point it should be able to fit. * **Capacitor $C_4$** is another DC blocking cap for stage isolation. ## Power Amplifier ![Figure [poweramp]: The Power Amplifier](./resources/poweramp.png width=75%) For a final stage of this system, we'll use a [12K5](./resources/12K5.pdf), which is a space-charge tetrode. It is capable of putting out more power than the 12AT7 which is good since we need it to drive the speaker. Its job is not necessarily to achieve voltage amplification, but rather power amplfication! Lecture 4 goes over its operation. * **Resistor $R_9$** is a grid stopper into the 12K5. 100 ohms should be fine. * **Resistor $R_10$** is a cathode bias resistor. Note its value is much smaller than stage 2 since the quiescent current in this tube is much higher. In contrast to the cathode bias resistor ($R_7$) for the second triode, I found this one *really* mattered. So you'll want to watch its tuning! * **Capacitor $C_5$** is another bypass cap that will shunt out its neighbor the cathode bias resistor at audio frequencies preventing issues. * **Transformer $T_1$** is a transformer used to impedance match the load impedance of the speaker (8 ohms to 32 ohms depending) to the output impedance of this tube (roughly 600 or 700 Ohms). Remember we want to match impedances (in fact match them so they are complex conjugates, but let's assume they are both purely real here) in order to ensure maximum power transfer. There is a right way to put the transformer in and a wrong way. Both ways will likely "work" but one will definitely work much much much better. * **The speakers $S_1$ and $S_2$** can be many things. I have some TRS connectors and a bunch of those crappy earbuds they give you airplanes (don't worry these are all new). Each speaker in these earbuds is 32 Ohms so I suggest wiring them in parallel so you have a 16 Ohm load. Then channeled through the transformer we have, it'll be a reasonable couple kOhms. This won't be a perfect match for the 12K5's output, but it is better than driving directly by far (see some napkin numbers in lecture) !!! WARNING I failed to include a grid leak resistor in this stage. The behavior of the 12K5 is such that at the bias voltages we are targetting, the steady-state leakage current on the control grid should be really small (like single nA) and therefore even with parasitic resistance, it should be able to leak off and not accumulate. However, with that said, it might not be a bad idea to put something like a 470 K resistor to ground from the grid just to pin the grid more firmly to 0V and prevent any drifting of the bias point while you tune the cathode resistor here (having it tuned in a bad spot for a while can build up extra grid charge and then that will slowly leak out and then you'll find yourself having to modify the cathode resistor again. We don't want too high of a resistor here, but we also don't want too low of a resistor since that will pull down our input impedance. Since the output impedance of our previous stage is about $15000||220000$ Ohms or about 15K, something like 470K or 330K is big enough to keep $R_L\gg R_S$. # Putting It All Together and Testing This circuit is pretty easy to test in stages. Before playing audio or something, get out the scope in lab and turn on the function generator. Make a 400 Hz sine wave of amplitude 20mV and 0V offset. Attach it to the input of the whole amp. Using two oscilloscpe probes, measure the input and output signals. While watching them adjust the grid leak until you have maximum undistorted gain. When you are happy then move onto the next stage. Either bring in a third probe, or transfer your two forward. For stage 2, just crank $R_5$ to max volume. Compare the signal before $R_6$ to that below $R_8$. Study the gain and adjust the cathode bias resistor $R_7$ until your gain is as big as you can get it. Then hop forward to the next part. For stage 3 (the power amp), again compare input to output. Make sure you are driving the speaker now since not having that wired up will greatly impact how this tube is working. Carefully tune the 12K5's cathode bias resistor $R_10$...I found this one definitely has a real parabolic bad-good-bad pattern so keep an eye on it. Hopefully by this point you're getting audio? If so feel free to go back and mess with the resistors. $R_4$ and $R_8$ in particular can be messed with to mix up the gain. As you do this, you'll need to shift your bias points. Remember we really want to keep the shape and quality of the sine wave "clean" as we propagate it forward through each stage and amplify it as much as possible. The 6.2000 lab scopes have four channels. You may want to use all of them to get maximum info while tuning! ## Is it Working? Once you get things working, consider tring the guitar out. With headphones it actually sounds not too bad. If you're particularly adventurous you could try to adjust the bias points in the circuit to actually get some distortion. Apparently there are long-haired hippy freaks out there who think distorting their guitar is "music" instead of voting for Nixon. In all seriousness, tubes (triodes in particular) distort very gently as opposed to semiconductors which usually have very hard and sharp clipping. This means tubes usually introduce interesting and arguably musical distortions which is why people like using them for things. See how that affects the sound of the guitar. You can also try to use your crystal set with this circuit! From Lab 1, you will need to transfer everything from Figure 2 (including the $R_1$ and $C_2$ LPF) to your input. There *should* be enough gain to get decent pickup. If you can do this, you've just totally made a radio completely using nothing but pre-1960's technology. Go you. If it is working, feel free to take a video to commemorate. You can hang onto the amplifier or put it on top of a bench with your name and a "DONT'T TOUCH" note if you want..