Introduction: An Old Charger? No, It’s a RealTube18 All-Tube Guitar Headphone Amp and Pedal
What to do during a pandemic, with an obsolete Nickel-Cadmium battery charger, and 60+ year-old obsolete car radio vacuum tubes sitting around needing to be recycled? How about design and build a tube-only, low voltage, common tool battery-operated guitar headphone amp and distortion pedal? I had some time and more leftover parts, so I also built one inside a dead Milwaukee tools lithium ion battery charger. These are rewarding e-recycle projects.
Before I get into the nuts and bolts of this build, I realize that the readers of this will range from novice to experienced in the required skills and experience. This being the internet age (with a bunch of links at the end), I won’t pretend to be able to explain as well as the technical sites how tubes work, electrical theory, how batteries work, how batteries differ, how to test tube circuits with oscilloscopes, use power tools, how to solder, etc. There is so much good material out there, and better than I could write. 120 years of electrical design is too much too learn for any one person anyway. Lastly, I am writing my design thinking process here, so that you can see how I approached my choices, in hopes that you will feel emboldened to customize the design.
Many thoughts came to mind as I designed the RealTube18 headphone amp and guitar pedal circuit. The final product ended up a safe (20 volts dc max) and convenient way to experiment with vacuum tube circuits, and for a packrat like me, quite low cost due to all of the components I had socked away.
Rescue an old tool battery charger.
Find appropriate vacuum tubes that someone was kind enough not to throw away 60 years ago.
Assorted resistors, capacitors, sockets, wire, jacks, and potentiometers.
You'll need a large assortment of tools, ranging from drills and hand tools to soldering iron, breadboard, digital multimeter, and don't forget a battery that will fit in the old charger's battery socket.
Step 1: How I Chose What the Recycled Battery Charger Would Do
I wanted a simple tube amp design, no or few transistors or integrated circuits, and relatively few other components. In the end, the only semiconductors in the final design are the power and effect LED’s.
I wanted this to be low voltage, run off a tool battery, be safe to breadboard with exposed wires, no ac filament or plate voltage transformers required. Low voltage breadboard experimenting is a safe way to learn tube circuits, and, allows for fast component changes without soldering parts (until the final build). (Warning: the tubes still get too hot to touch.) I bought a couple 9-pin tube socket adapters online that plug straight into a breadboard. Low voltage (rated at least 25v) electrolytic capacitors are inexpensive and small, unlike the 400 or 600 volt-rated siblings required in power supplies of high voltage tube amps.
I desired zero ac electrical noise: by keeping to direct current from a battery, the only ac involved is the audio signal itself.
Tube sound: I was building this to create authentic tube harmonic distortion for guitar. I am fairly pleased with the result. This amp operates in the linear, low-distortion regime with the guitar volume knob low and the drive control low. Depending on guitar pickups, the distortion can go to extreme pretty quickly. Those who are extremely familiar with tube guitar amps will not be surprised that my choice of single-ended tetrode will not have the same sound profile as one with a beam power tube, nor the harmonics palate of a push-pull power stage. Still, I like the results for this project.
Affordable: I wanted to use as many components from my parts boxes as possible. I confess I employed several used parts, even electrolytic capacitors. If you are building for the long haul, once you settle on your design and are happy with the breadboard, I suggest new, good quality electrolytic capacitors-your future self will be happy to not replace capacitors in 5 to 10 years.
Step 2: Selecting the Low Voltage Vacuum Tubes
To affordably accomplish low voltage, genuine “tube sound”, I decided to use the low voltage tube type developed for automotive radio use from 1955 to 1962. There are two categories of these low voltage tubes: “space charge” and conventional. The space charge type basically use an extra current flowing through the tube to mimic the electron activity consistent with higher plate voltage operation. I was okay with either type, but low voltage conventional types do not require the extra current that space charge types do.
These low voltage tubes were created because the low-voltage power transistor had just been successfully developed, but high-frequency transistors were not yet available. Car radio manufacturers were seeking a solution to operating at 12volts, to do away with the need to generate high voltages for the standard vacuum tubes. It didn’t take long, however, before all of the tubes became outmoded, and the low-voltage tube type automobile radios existed only briefly. While these automotive tubes were designed to handle the rigors of bumpy roads, they lacked the design life cycle to improve performance as well as get rid of microphonics. With the volume up, for example, you can tap the circuit board and hear it in the headphones.
My single-ended headphone amp/guitar pedal would need two or even three triodes to get even enough drive signal, and then one power tetrode or pentode to drive the headphones.
Tube availability: low voltage tubes are no longer manufactured, so New Old Stock will be the only option.Vacuumtubes.net, and several other websites do a nice recycling job of rescuing these from the landfills by buying them in bulk at estate sales and from closing businesses. The tubes I chose represent both categories for tubes these days. The 12U7 is popular with guitar tube pedal crafters so prices are up. Oppositely, the 12J8 is used by very few crafters so prices are very low. Happily, at these low voltages, tube power dissipation is so low that the tubes last a very, very long time.
The tube heater filament was tricky. I wanted to use a 18-20volt tool battery and not waste money/space/power on separate heater filament power circuits. I set out to find a tube combination that allowed the filaments to be placed in series and/or parallel to operate within manufacturers’ tolerances at a total of 18 to 20 volts. More discussion on the winning arrangement later.
Tube types: I wanted a twin triode pre-amp feeding into a tetrode or pentode power amp, for classic single-ended Class A operation. A third triode could work if I needed the gain, but I ended up not needing that extra gain, so a tetrode/triode combo tube was not necessary, only a tetrode.
The list of dual triode, Low Voltage tubes is quite short. None of these tubes are true “space charge” type, as this technique is used to allow more current to flow in a power output tube as opposed to a voltage gain tube.
See image of low voltage, dual triode tubes. I am not sure how well these photos will upload, so resolution might make these difficult to read.
For the power tetrode, the 12J8, 12DK7, and 12EM6 all had decent power. The 12J8 tube has the highest power output of the non-space-charge type, and, has a 0.325 amp heater current at 12 volts.
See image of low voltage tetrode tubes.
I was looking for a dual triode tube that could work with the 12J8's 0.325 amp current. As luck would have it, the 12U7 tube has 0.3 amp heater current at 6 volts, when using the heater centertap.
So, one 12J8 heater at 12.6 volts in series with one 12U7 in split-filament configuration at 6.3volts want 12.6+6.3=18.9 volts total for the heaters, right around 0.3 amps. An 18 to 20 volt tool battery is a perfect match for this combination. Search the internet for “tube datasheet” to see manufacturers’ tolerances for operating parameters of tubes you are interested in. In testing, I found that a fully charged battery at 20 volts powering these filaments resulted in 11.8 volts to the 12J8 and 7.2 volts to the split 12U7 heater (14.4 volts non-split filament equivalent). These values are within the 10 to 16.9 volt specifications for these tubes, and ran at about .32 amps. I got very lucky with this combination.
Another note: the 12U7 is more or less a specially tweaked 12AU7 tube. The 12AU7 (European code is ECC82), designed way back, at least in 1946 and perhaps earlier, was meant for high voltage operation, and is again manufactured today, due to its excellent audio pre-amp performance.
For completeness, “Space Charge” types of power pentodes or tetrodes do not have a suitable current match to the 0.3 amps of the split heater operation of the 12U7. And, the total tube current draw is higher due to the space charge grid. So, 12J8 was my choice for power tube. If you are going in a different direction, then the higher plate currents might be more attractive to you. See the picture of the “space charge” power tubes that were made, for further reference.
So, for my project, the best match is the 12U7-12J8 pair. The 12J8 is rated to 20 mW audio output power, which is second only to the 12K5 at 40mW. But, since the plate voltage will be 18 to 20 volts, instead of 12.6 volts, the power output will be a bit higher, with my measured result around 40 mW—my actual power output got higher than this, but distortion was quite high. Note that some of the tubes’ screens and plates have 16 volt maximum ratings, but most are rated at 30 volts—the 12U7 and 12J8 both are rated to 30 volts.
Conveniently, replacing the single-ended 12J8 power stage with a push-pull pair of 12J8’s with 12U7 phase splitter, would result in two 12U7 and two 12J8 total—meaning the heaters would still be workable as one split filament 12U7 in series with one 12J8, just twice. So, a push-pull version of this amplifier is just as doable within my constraints. I might build a push-pull version at some point.
A quick note on tube brands: for New Old Stock tubes (made before 1980, basically), brands differed somewhat on quality, but for these tubes, I haven’t noticed a discernible (to me) difference in performance. Whether RCA, Sylvania, GE, etc. or, the re-branded tubes with automobile manufacturer’s names on them (FoMoCo, GM, etc.), they all should perform similarly, albeit they did not stay mainstream long enough to get fine-tuned.
Step 3: Choosing the Amp Enclosure
I wanted to use an enclosure that already had a battery connection for the desired battery type and could be used reasonably as a guitar pedal.
For the Ryobi version, I used an abandoned Ni-Cd charger that was buried in the garage, waiting for an e-recycle trip. After removing the unneeded internals (destined to be recycled into a dc power supply in another project), enough space remained to mount the necessary components. This is a very handy use for obsolete Ni-Cd chargers.
Similarly, for the Milwaukee M18 version, I bought a failed charger online and gutted the enclosure. Added step here: the charger I used does not have the positive battery terminal in the correct position, so a careful cut and epoxying of a terminal in the correct position are required. This is because the M18 charger was for a lithium ion battery, and required special charging connections.
When laying out the components and drilling holes, patience is a virtue. With plastic, go slowly to avoid cracks or errant locations. And cover most of the case with masking tape: this allows you to mark for drilling, and protects the case from more scratches. Spend time envisioning the location of all components before you make any holes. Clearance between components can’t be changed nicely once they are mounted.
To drill for the tubes, I used a forstner bit and piece of pre-drilled scrap wood as a guide, clamped to the box. A hole saw probably would have worked better.
To re-purpose any kind of enclosure, you will need a fair number of tools. If you are just gaining experience doing this kind of thing, I suggest practicing on a junk enclosure first—better still, if you can get two of the same old box, then you can have a back-up if the case breaks or you don’t like your placement.
Step 4: Choosing Components
Resistors: I have accumulated a zillion resistors over the years, many of them carbon composition type. Nowadays, I would not recommend carbon composition due to reliability. I used what I had on hand, though. Even though this is all low voltage, you might not be able to use the small 1/8 watt resistors everywhere—do the math to be sure you don't fry a resistor (power dissipated = current^2*resistance).
Capacitors: since this is below 25volts, every electrolytic can be rated for 25 volts, some lower. So, these are inexpensive compared to the capacitors I use in amps with 350volts B+. The coupling caps, with these high megohm grid resistors, can be smaller than 0.022 and 0.1 uF. However, I have a bunch of each value that are rated at 100v, so I used them. If you are going to buy a bag of them for this type of project, I suggest a pack of ten 0.05uF 100V rated, or 0.1uF if the tone control needs it—or assortment to experiment. The coupling caps mostly set your bass frequency response cutoff.
Output transformer: Typically, at high voltages and dc idle currents, the audio output transformer is big and heavy—and pricey. However, I used a 70 volt line transformer, which is fine for these low dc currents. These are lightweight and inexpensive. If you have a suitable audio output transformer sitting in a parts box, that should sound even better, but a 70v transformer will work. There is a lot of guidance on the net for choosing the correct taps for your project, but I chose the 2W tap to get roughly 2500 ohms load impedance shown to the 12J8 output.
Load: I designed this for parallel 16 ohm headphones/earbuds. Two 16 ohm in parallel is 8 ohm, which works well for the 70 volt line transformer 8 ohm output. But, I added a 1 ohm resistor in series to the headphone/dummy load as a voltage divider, providing a low guitar pedal output. This divider was determined experimentally, targeting a loud effect output voltage that is similar to the input voltage when bypassed to the output when the stompbox switch is pressed.
Step 5: Designing My Circuit
Any complex electronic circuit is made up of several, much simpler circuits. A sketch of my circuit is uploaded.
Guitar input: The guitar input terminates immediately to one end of the first pole of the two-pole-double-throw stompbox switch, and continues on to the first triode stage’s input capacitor. A single coil pickup puts out about a 0.07vac signal, while a humbucker can reach around 0.7 vac.
Pre-amp: To maximize amplification factor, grid-leak bias was chosen for the first triode of the 12U7. The coupling capacitor is needed for grid-leak bias operation. This capacitor also reduces risk during experimenting, making it impossible for an improper connection to backfeed any dc current into the input test source or guitar pickup. (I would prefer not to say why I point this out...) Anyway, the grid-leak resistor basically works on the principle that the cloud of electrons in the area of the hot cathode (what is really the “space charge” cloud) will offer a tiny electron flow through a resistor either connected to the cathode or connected to the B+ supply. Experimentally, a 5 megohm resistor connected to B+ sounded the best to me, and gave about -.5 volt bias (leakage current can reach as much as 10uA per the datasheet). With a humbucker pickup of 0.7vac, the -0.5v bias is a pretty good place to operate. Experiment with different values from 2 to 10 megohm to hear the difference, and see it on an oscilloscope. (An oscilloscope is pretty specialized, but really valuable if you wish to experiment with designs.)
A note about battery notation: the names “A,” “B,” and “C” for portable radio batteries were established over 100 years ago. Since my design doesn’t need a different voltage for the heaters, there is no “A” battery in this design. Everything operates from the plate voltage, i.e. “B” battery, so there is no “A+” connection. Also, I am biasing the grids with resistors, so there is no “C” battery.
Second audio stage: This is the second triode of the 12U7, fed from the output of the first stage. This stage is cathode-biased with an adequately bypassed 10K potentiometer. This pot is what I use as the “drive” control, to basically increase the amplification factor of this second stage, which will reduce the level of guitar input required to cause distortion. Note, with this design, if you dig into a humbucker with the guitar’s volume knob up, every stage saturates and sounds, well, not good, since all three stages are distorting. But, when you experiment between guitar volume, amp drive setting, and amp volume level, there are a lot of tones to found. This does not sound as good as a 6V6 tube to my ears, but fun nonetheless. For use as a pedal, an Automatic Gain Control circuit would be nice, but I don’t feel that ambitious for now.
The tone control is optional. And, you can experiment with any tone stack you want. Be aware that some tone control configurations can greatly attenuate your coupled signal.
Power stage: The 12J8 has two built-in diodes that I did not use. These were meant to detect (tune in) radio signals and then amplify them enough to drive a (newly invented then) power transistor. I tied the diode’s shared cathode and anodes to ground (- of battery), so that they would essentially be inert. Theoretically, one could tweak the capacitance between the tetrode section and diodes by changing the potential, but someone else can experiment with that...
The output signal goes first to the headphone jack, and then back to the circuit board's 1ohm resistor to pick off the pedal output signal. So, it is important to use this type of headphone jack, which has the interrupting contacts allowing for the onboard 16 ohm load resistors to be the load to the power tube if headphones are not plugged in.
The tetrode screen is connected to the same B+ power supply ladder node as the B+ for the first two stages-- I experimented with decoupling these (12U7 B+ from the 12J8 screen), but I did not see any advantage on the scope. You might want to decouple these with 200 ohm resistors in the B+ ladder and add a 25uF at each node.
Power supply capacitors: the B+ power supply node feeding the 12J8 has a 100uF capacitor, which is overkill, but I have the caps sitting around. The rest of the power supply ladder nodes can be 22uF or 47uF. These caps are not here for 60Hz noise filtering, just response. Lower capacitances in the power supply ladder might give you a little bit of the “sag” reminiscent of tube rectified amps—I did not experiment with that.
I used the second pole of the stompbox switch to send B+ to either the tube plates or the “bypassed” LED (not typically done on standard guitar pedals, but the Ryobi charger had a third LED). The heaters and “power” LED are run directly from the main power switch contact. There is not actually a benefit to removing power from the plates when the effect is bypassed, since a “standby” switch is really only meant to use on initial heat-up on high-voltage tubes, but I am looking to reduce battery-drain any way I can. The tubes take 25 seconds to get sounding normal, so I didn’t want to cycle those with the stompbox switch. Still, this single-ended design only draws a third of an amp, so a 4-amp-hour battery theoretically could drive this for 12 hours. I certainly have run many hours in testing before I needed to recharge the battery.
In hindsight, I probably should have inserted a fuse right on the B+ input terminal. This would decrease the chance of a fire in the event of some kind of unforeseen issue inside the enclosure. I recommend you fuse whatever you build, because the batteries can dump a lot of current into the circuit.
I used paper, experience, computer spreadsheet, multimeter, and oscilloscope to create and refine my design. For those spice simulation devotees out there, there is tremendous advantage to trying, virtually, all sorts of circuits on the computer. I understand, though, that tubes are not easy to model perfectly well (especially at low voltage with grid-leak bias), so when you get to actual component assembly, do not be too surprised if the behavior of the circuit deviates a bit from the simulation. I should think the notion of a heated cathode releasing electrons into a charged “cloud” billowing out in the direction of the grid, screen, and plate must be quite challenging to model—especially for tubes like the 12J8 which was not around for long enough for anyone to publish operating curve data.
Step 6: Making Your Own Design
I uploaded a bunch of pictures of the two build phase of both amps. I recorded a few guitar chords at four different settings to give an idea of the tones.
My design here is just an idea to show you that you can choose your own goal, your own tubes, your own form factor, and build it at safe voltages to learn about tubes. You could add an inexpensive, battery operated integrated circuit power amplifier and speaker to make a hybrid amp. You could make a true push-pull tube or transistor amp. You could use a different DC supply and run these tubes at 30 volts to get more power. You could use an ac-to-dc power supply instead of a battery. You could bias in linear operation regimes only and make an audiophile headphone amp. Different guitar effects could be built in. This could be packaged into a 19-inch rackmount version. Go for it. Rest easy knowing that whatever you feel like trying is just as valid as anyone else’s idea.
My only cautionary advice is to those of you who are relatively new to these subjects. Take small steps to keep from getting discouraged. Get a breadboard and a power supply and start learning how circuits work. Work with one tube or one transistor and see how it operates, before adding complexity. At low voltage, you can still smoke a 25 cent transistor, but you won’t damage a tube unless you get really far off, like connecting B+ to the control grid for a long time. Add complexity slowly. If you can get a digital multimeter, function generator (app on the phone) and an oscilloscope (either bench equipment or app/program on an old PC), you will then have all you need to learn a lot. This knowledge could springboard you into digital signal processing, or modifying your existing equipment, or repairing broken equipment.
Step 7: Acknowledgements
I won’t pretend to have invented all of the ideas presented here.
If you do an internet search for patents (2864026,2946015, 3017507, 10063194, to randomly name a few), or check out “sophtieamps” or "Frank's massive tube datasheet collection" or "NJ7P's tube manuals with theory" or "tubetheory" or "antiqueradios" or “diyaudio” or “space charge tubes” or “angelfire” or “radiomuseum” or literally thousands of other pages, you will find many guitar amps, guitar pedals, headphone amps, and general tube circuit guidance that contribute to my build, and yours. Thanks to all have come before, and best wishes to you future makers/recyclers.
Step 8: A (Very Technical, Sorry) Update to an Already Technical Project:
In the last several weeks, I made two tweaks to the design.
First, to optimize the tetrode's power output and sound quality, I set the screen voltage between 12.6 and 13.3 volts with a voltage divider. I experimentally setteld on a roughly 3K resistor from B+ to screen, and then 10K resistor to ground. I bypassed screen to cathode with a 1 or 2 uF cap. You may need to adjust the 3K higher, depending on your actual circuit to set this screen voltage. The current is a little under 2mA through the 3K. The screen is tied ac-wise now to the cathode with a 1uF bypass capacitor, to allow the screen to better do its job as the plate and cathode voltages swing. This screen voltage setter seems a good architecture for any low voltage tetrode, to maximize performance.
Second, I found that the Ryobi 18v lithium ion battery emits some kind of digital charger communication request every 15 seconds, causing a "tick" sound. It is a short ac blip atop the DC voltage. I added a filter ladder for it. If you can get a small (1 or more mH) inductor, you can add that to the power supply filter ladder. I did not see a need to run the heater current through the inductor.
A last note: the 10K potentiometer needs to be a good quality, since it can see several milliamps and any noise generated goes straight to the plate and impacts the sound.
If anyone who didn't want to start vacuum tube experimenting at high voltages, and instead tries something like this, please let me know.
Thanks for reading.
Participated in the
Recycled Speed Challenge