An Ultra Low Wattage, High Gain Tube Amplifier




Introduction: An Ultra Low Wattage, High Gain Tube Amplifier

About: Started playing guitar but developed a serious taste for pedal and amplifier building. Interested in smaller builds, for portability and lower volume settings, mainly using vacuum tubes and older technologies …

For bedroom rockers like me, there is nothing worse than noise complaints. On the other hand, it is a shame to have a 50W amplifier hooked to a load dissipating almost everything in heat. Therefore I tried to build a high gain preamp, based on a famous mesa amplifier using some subminiature tubes for ultra low output.

Step 1: Overview, Tools and Materials

This instructables will be structures as:

  1. Circuit overview: The amplifier
  2. Circuit overview: The SMPS
  3. Parts list
  4. Thermal transfer
  5. Masking
  6. Etching
  7. Finishing
  8. Adding sockets
  9. Assembling the boards
  10. Adjusting the trimpots
  11. Mounting everything inside the enclosure
  12. Final result and Soundcheck

There are some tools required to build this amplifier:

  • Hand drill, with different drill bits (in case you want to drill the PCB with a hand drill you need a 0.8-1 mm drill bit, not normally found in kits).
  • Soldering iron
  • Clothes iron
  • Multimeter
  • Sanding files
  • Access to a toner printer
  • Plastic box for etching

And some materials

  • Sanding paper (200, 400, 600, 1200)
  • Spray paint (black, clear)
  • PCB Coating spray
  • Ferric Chloride Etching Solution
  • Solder

Step 2: Circuit Overview: the Amplifier

Subminiature tubes for batteries

For this project I used 5678 and 5672 tubes. They were used in portable battery radios, where filament current was a problem. This tubes only require 50mA for their filaments, making them way more efficient than the 12AX7. This keeps the current consumption low, requiring a smaller power supply. In this case I wanted to power them with a 9v 1A power supply, as commonly used with guitar pedals.

The 5678 tube has a mu of roughly 23, which makes it a low gain tube in comparison with the 12AX7, but maybe with some tweaks even this could be enough. High gain amplifiers are known to have a lot of filtering between stages, where almost the majority of the signal is shorted to ground. There may be some air to play with.

The 5672, on the other hand, has a mu of 10, but was mostly used as a power tube in hearing aid devices, and was already used in some other subminiature amplifiers (Murder one and Vibratone, from Frequencycentral). It can produce up to 65mW clean...ish. Don't be scared with the low wattage, it's still pretty loud when distorted! The datasheet specifies a 20k output transformer for this tube.

As in previous builds, the 22921 reverb transformer will be used.


One of the difficulties is to bias these tubes without using different batteries, since they have direct heated cathodes. I did not want to make this more complicated, so I had to use a fixed bias configuration. This, on the other hand, allowed the use of the filaments in series, reducing the total filament consumption. With 6 tubes, each dropping 1.25V, I got pretty close to the 9V of the power supply, it just required a small resistor, which also improved the bias of the first stage. This means the total filament current is only 50mA!

Pretty good for a pedal power supply.

For it to work, some stages have a trimpot to adjust the desired bias. The bias is calculated as the difference between the voltage at the negative side of the filament (f-)and the grid of the tube. The trimpot adjusts the DC voltage at the grid of the tube, allowing the different bias configurations and is bypassed by a large capacitor, working as a short to ground for the signal.

The third stage, for example, is biased close to the cut-off point of the tube at -1.8V, achieved as the difference between f- (pin 3) at roughly 3.75V and the grid, at 1.95V. This stage emulates the cold clipping stage found in high gain amplifiers, such as the soldano or the dual rectifier. The 12AX7 in a dual rectifier uses a 39k resistor to achieve this. The other stages are almost center biased, at approximately 1.25V.

Step 3: Circuit Overview: the SMPS

High voltage supply

Regarding the plate voltage, these tubes run ideally with plate voltages at 67.5V, but also worked with 90V or 45V batteries. Those batteries were huge! They are also difficult to come by and expensive. That's why I opted for a switched mode power supply (SMPS) instead. With the SMPS I can boost the 9V to 70V and add some massive filtering before the output transformer.

The circuit used in this instructables is based on the 555 chip, successfully used in previous builds.

Step 4: Parts List

Here you have a summary of the necessary parts:


C1 22nF/100V__________R1 1M_______________V1 5678
C2 2.2nF/50V__________ R2 33k_______________V2 5678
C3 10uF/100V__________R3 220k______________V3 5678
C4 47nF/100V__________R4 2.2M______________V4 5678
C5 22pF/50V___________R5 520k______________V5 5678
C6 1nF/100V___________R6 470k______________V6 5672
C7 10uF/100V__________R7 22k_______________TREBBLE 250k Linear 9 mm
C8 22nF/100V__________R8 100k______________MID 50k Linear 9 mm
C9 10uF/100V__________R9 220k______________BASS 250k Linear 9 mm
C10 100nF/100V________R10 470k_____________GAIN 250k Log/Audio 9 mm
C11 22nF/100V_________R11 80k______________ PRESENCE 100k Linear 9 mm
C12 470pF/50V_________R12 100k_____________VOLUME 1M Log/Audio 9 mm
C13 10nF/50V__________R13 15k______________B1 10k trimpot
C14 22nF/50V__________R14 330k_____________B2 50k trimpot
C15 680pF/50V_________R15 220k_____________B4 50k trimpot
C16 2.2nF/50V_________ R16 100k_____________SW1 micro DPDT
C17 30pF/50V__________R17 80k______________J1 6.35 mm Mono jack
C18 220uF/16V_________R18 50k______________J2 DC Jack
C19 220uF/16V_________R19 470k_____________J3 6.35 mm Mono-switched jack
C20 220uF/16V_________R20 50k______________SW2 SPDT
C21 220uF/16V_________R21 100k_____________LED 3 mm
C22 100uF/16V_________R22 22k______________3 mm LED holder
C23 100uF/16V_________R23 15R/25R
C24 220uF/16V_________R24 15k
C25 10uF/100V_________R25 100R
C26 10uF/100V_________R26 1.8k
C27 220uF/16V_________R27 1k
C28 100uF/16V_________R28 10k
C29 47nF/100V_________R29 2.7k (LED resistor, adjust for brightness)
C30 22nF/100V_________R30 1.5k

Special attention to the capacitor voltage rating. The high voltage circuit requires 100V capacitors, the signal path after the coupling capacitors can use lower values, in this case I used 50V or 100V since the film capacitors have the same pin spacing. The filaments need to be decoupled, but since the highest voltage on the filaments is 9V a 16V eletrolytic capacitor is on the safe side and way smaller than a 100V one. Resistors can be of the 1/4W type.

555 SMPS

C1 330uF/16V__________R1 56k______________IC1 LM555N
C2 2.2nF/50V__________ R2 10k______________L1 100uH/3A
C3 100pF/50V__________R3 1k_______________Q1 IRF644
C4 4.7uF/250V_________ R4 470R____________ VR1 1k
R5 150k_______________D1 UF4004 or ES2G (ultra fast)
R9 2.2k

Attention to the switching diode! It must be of the ultra fast type, otherwise it won't work. For the SMPS low ESR capacitors are also desired. In case a normal 4.7uF/250V capacitor is used an additional ceramic capacitor of 100nF in parallel helps to bypass the high frequency switching.

These are the easier parts to find and can be obtained from any eletronic parts store. Now, the tricky parts are:

OT 3.5W, 22k:8ohm transformer (022921 or 125A25B) Banzai, Tubesandmore

L1 100uH/3A inductor Ebay, just don't buy the toroidal shaped. You also find it at Mouser/Digikey/Farnell.

Don't forget to buy:

  • A copper clad board, 10x10 mm will do for both boards
  • 2x 40 pin sip sockets for the tubes
  • A 1590B enclosure
  • Some 3 mm screws and nuts
  • Rubber feet
  • 5 mm rubber wire grommets
  • Six 10 mm knobs

Step 5: Thermal Transfer

To prepare the PCB and the enclosure I use a process based on toner transfer. The toner protects the surface from the etchant, and as a result after the etching bath we have the PCB with the copper tracks or a beautiful enclosure. The process of transfering the toner and preparing for etching consists of:

  • Print the layout/image with a toner printer using glossy paper.
  • Sand the surface of the enclosure and of the copper board using sanding paper with grit 200 to 400.
  • Fix the printed image to the PCB/enclosure using tape.
  • Apply heat and pressure with the clothes iron for about 10 minutes. Make some extra movement with the tip of the iron at the edges, those are the tricky places where the toner won't stick.
  • When the paper is looking yellowish trow it in a plastic container filled with water to cool it down, and let the water soak into the paper.
  • Remove the paper carefully. It's better when it comes off in layers, instead of removing everything in a single attempt.

The drill template helps to identify the positioning of the components, you just need to add your own art, and you're good to go.

Step 6: Masking

For the enclosure, mask larger areas with nail polish. Since the reaction with aluminum is much stronger than with copper, there could be some pitting in larger areas.

Giving an extra protection guarantees that there will be no marks to ruin the enclosure.

Step 7: Etching

For the etching process I like to use a plastic container with etchant and one with water to rinse between steps.

First, some safety tips:

  • use rubber gloves to protect your hands
  • work on a non-metallic surface
  • Use a well ventilated room and avoid breathing the resulting fumes
  • Use some paper to protect your workbench from possible spills

Here I only show the etching of the enclosure, but the PCB was etched in the same solution. The only difference is that for the PCB I just waited for about an hour until all the unprotected copper was gone. With the aluminum there must be some extra care, since we only want to etch the outside of the box.

For the enclosure I shake the box in the etching mixture for about 30 seconds, until it gets warm due to the reaction an rinse it in the water. I repeat this step another 20 times, or until the etch is about 0.5 mm deep.

When the etch is deep enough wash the enclosure with water and soap to rinse off all the remaining etchant. With the box cleaned sand the toner and the nail polish off. For the nail polish you can save some sanding paper by using acetone, but remember to keep the room well ventilated!

Step 8: Finishing

In this step I used the 400 grit sanding paper to achieve a clean surface, like in the third picture. This is clean enough for the drilling step. I drilled all the different sized holes, and used the files to make the holes for the tubes sockets. The PCB must be drilled too, I a 0.8 mm drill bit for the components and 1-1.4 mm for the wire holes. In this build I also used a 1.3 mm drill for the tube sockets.

With the drilling and filing done I give the box a black coat of spray paint and let it dry for 24h. It will give a better constrast between the etch and the enclosure. Obviously, the next step is to sand it off. This time I go from 400 to the finest grit. I change the sandng paper when one grit removed the lines of the previous one. Sanding in different dirrections makes it easier to identify when all the previous marks are gone. With the enclosure shining I apply 3 layers of the clear coat and wait until it dries for another 24h. The PCB can be protected from corrosion by using a protective coating. As you can see in the last two figures I like to have a dark green coating. This coating requires longer times to dry. I waited 5 days to avoid having finger prints on the board while soldering the components.

Step 9: Adding Sockets

Soldering the Sockets

According to the layout, the tubes are mounted at the copper side of the board. This way the board can come closer to the enclosure and profit from some extra shielding against nasty high frequency EMI coming from the SMPS. But using the copper side of the board to solder components has some disadvantages, such as the copper becoming loose from the board. To avoid this, instead of soldering the tube sockets, I made larger holes where the sockets could be pressed in. The pressure of a slighlty smaller hole and some solder on both sides should solve the problem. For this I used the machined style pin sockets, without the plastic structure, forced the metal pin in the hole and soldered on both sides (on the components side it looks like a blob of solder, but it helps to keep the pin stuck), as shown in the first 3 pictures. The 4th and 5th pictures show all the sockets and jumpers installed.

Soldering another set of sockets, this time with the plastic structure, to the tubes improves the connection to the board and makes it more stable. The original pins of the tubes are very thin, which can lead to some bad contact or even falling off the sockets. By soldering them to sockets we solve this problem, since now they have a tight fit. I think they should have come with proper pins on the first place, like the larger tubes!

Step 10: ​Assembling the Boards

To solder the components I started with the resistors, and moved to the larger parts. The electrolytics are soldered at the end, since they are the highest components on the board.

With the board ready it's time to add the wires. There are a lot of external connections here, from the tonestack to the high voltage and filament cables. For the signal wires I used shielded cable, shielding the ground mesh at the panel side, closer to the input.

Critical wires are around the first stage, coming from the input jack, and going to the gain potentiometer. Before we can build everything inside the box we need to test it, so that we still have access to the copper side of the board for some debugging, if it's necessary.

For the high voltage filtering I added another RC filter in a smaller board, mounted perpendicularly to the main board, as seen in the picture. This way the ground, high voltage and transformer connections are easier to acccess with the board mounted to the enclosure and can be soldered afterwards.

Building the tonestack

Although I was going to test the board outside the enclosure I already built the tonestack in the box. This way all the potentiometers are fixed and properly grounded. Testing the circuit with ungrounded potentiometers (at least the outside shield) can result in horrible noises. Again, for longer connections I used a shielded cable, grounded near to the input jack.

Unfortunately in this build the potentiometers are really close together, making it difficult to use a board with the components. In this case I used a point-to-point approach for this part of the circuit. Another problem was that I only had a PCB style 9 mm 50K potentiometer, so that I had to anchor it to the neighbouring potentiometers (panel mount style).

Now is also a good time to install the on/off switch and the LED with the 2.7k resistor.

As a result of two rows of potentiometers I had to file the inside wall of the lid, as shown in the picture, so that the box would close.

Step 11: Adjusting the Trimpots

Adjusting the 555 SMPS

If the SMPS is not working there is no high voltage and the circuit won't work correctly. To test the SMPS just connect it to the 9V power jack and check the voltage reading at the output. It should be around 70V, otherwise it needs to be adjusted with the trimpot. If the output voltage is 9V there is a problem with the board. Check for a bad mosfet or 555. If the trimpot does not work verify the feedback circuit around the smaller transistor. An advantage of this SMPS is the low count of parts, so it is a little easier to identify any mistakes or faulty components.

Adjusting the mainboard trimpots

During the testing stage is a good time to adjust the bias with the trimpots. It can be done later, but if the tone is to dark or to bright it is easier to make changes now.

The first trimpot controls the bias of the second, third and output stages and is therefore the most important. I adjusted this trimpot by measuring the bias of the third stage, the cold clipper. If the bias is too high the stage will be completely in cut-off, giving a raw, cold, spongy distortion. If it is biased hotter the output stage will be too hot, adding some power stage distortion, and running the tube closer to the max. plate dissipation. In this case, the lower side of the master volume should be connected to the negative side of the first stage, so that the bias is still around 5.9V. In my case it sounded better when the output stage was running at 5.7V instead of 6.4V.

Just measure the bias at the third stage (middle tube in the back row) and verify that it is around 1.95V The second trimpot just needs to be adjusted to taste, or nearly center biased at 1.2V (measured between pins 3 and 4). Similarly the third trimpot is also adjusted to approx. 1V.

The voltage readings at the tube's pins 1(plate) to 5 (filament) are:

V1: <8.8V><41.6V><1.21V><0.00V><2.50V>






Note that the filaments in the 5672 are backwards than in the 5678, so that the tubes can't be swapped. Another important aspect to consider is the tube manufacturer. I found out that the tung-sol tubes sounded better in the first positions, than the raytheon tubes. Checking it with an oscilloscope it was visible that the tung-sol tubes had more gain than the raytheon tubes I had.

Now is also the time to test the circuit and see how it sounds, if it is too bass heavy I suggest changing the 47nF capacitor between second and third stage to 10nF, that will filter some bass out from the initial stages and improve the sound. If it got too thin, just increase this capacitor to 22nF and so on.

Step 12: Mounting Everything Inside the Enclosure

I started adding the screws for the mainboard. On the inside I added the rubber wire grommets, to give some clearance between board and enclosure and also to dampen some vibration. By running the first stage in pentode mode this could help if the tube gets microphonic. Then I added the board and screwed it down with the nuts, connected the tonestack, inserted the input jack and soldered the remaining wires.

With the mainboard in position I added the output transformer, adjusted the lenght of the wires and inserted the output jack and power jack.

At this point I saw that my SMPS board would not fit in the desired position (at the lateral wall, with the components perpendicular to this wall) because I added the power jack on the wrong side of the output jack... To fix this I sawed the SMPS board at the input side, removing the inductor and capacitor, and soldered the piece back to the board rotated by 90 degrees, as shown in the picture. I tested the SMPS again to see if it still was working, and finished by connecting the high voltage to the main board, through the RC filter board.

Step 13: Soundcheck

Now just plug the amplifier to your favorite 8 ohms cabinet (in my case a 1x10" with a celestion greenback) and use your pedal power supply to play at non-deafening levels!

By the way, if you like the sound of your amp feedbacking when you stop playing at the end of a sound, wait for the middle part of the video, it feedbacks quite easily when sitting in front of the cab.

Pocket Sized Contest

Second Prize in the
Pocket Sized Contest

1 Person Made This Project!


  • Teach With Tinkercad Contest

    Teach With Tinkercad Contest
  • Fandom Contest

    Fandom Contest
  • Back to School: Student Design Challenge

    Back to School: Student Design Challenge



23 days ago

Nice design and build!

Is there a standard way to list the Bill of Materials (BoM) so that one can just submit it to, say, Mouser, and get everything without having to enter each part number separately?


2 years ago

Hi Thomas,

Could you please share brd file to be able to etch pcb? Thanks in advance!


Reply 2 years ago

Thanks a lot, it is a wonderful project! I will share my results if I can finish it :)


Reply 6 months ago

Hi Thomas, thanks a lot for sharing this project. I finally completed mine, I put an amateur name on it also ;) I designed and printed knobs, grill and upper corner protections. Sounding good but I will dig it little bit more to achieve better sounding as much as possible. Thanks again.

WhatsApp Image 2022-02-26 at 14.13.04.jpeg

1 year ago

Very nice


2 years ago

Hi Thomas,
I'm also interested in low-wattage amps using submini tubes. I used a uTracer to make some traces of the 5672 with a Vs of 90, in case you ever want to use a slightly higher B+ and attached them below. I've also attached the traces I made with Vs = 67.5 for reference since they show a bit more detail than the ones found in the datasheets. In particular, there is a tetrode-like effect happening when the Va is between 10 and 20.


Reply 2 years ago

Hi, Thank you! These plots will be really useful. I am kind of interested how I could push this tubes to the limit in a push-pull configuration. At the same time I will be working on a better spice model for them and at least your curves have a better resolution than the datasheet.


Reply 2 years ago

No problem! If you need traces at any other screen voltages, just let me know. I should note that these were done with the heater at 1.3v, with the cathode referenced from filament negative.
The main problem I see with P-P is finding an appropriate OT. Looking at the curves for Vs 90, I’d want something between 27k and 40k, which is just beyond what’s commonly available. I suppose you could do the “8 ohm speaker on the 4 ohm tap” trick, but that tetrode-like area makes it a bit tricky.


Question 2 years ago on Introduction

Also, how would I add a Little Bear E1 circuit for simulated speaker output for running into a PA?


Answer 2 years ago

I would add it after the output transformer. You would have to have some fake speaker, so that the transformer still sees a 8 ohms load and the circuit would go in parallel. You would basically waste the power on a resistor and use only a little fit to drive the little bear.


Question 2 years ago on Introduction

Does this have a phase inverter? Also, what happens if you add another 5672 output tube? Thanks.


Answer 2 years ago

There is no phase inverter, it is a single ended stage. You could add another 5672 in parallel for twice the power, but you would have to change the output transformer to 10k and adjust the filament series string accordingly.


3 years ago

That sounds really beautyful! It reminds me a bit of the AX84-project, a similar design. Keep on rocking! And keep good hand made music alive! If I hadn't an AX-84 with a single tube output-stage, this should be 'my living-/bed-room-amp' ! Good luck in the 'pocket-sized contest' !


3 years ago

I would like to buy a kit or the circuit board. Any chance? Price? Buy the one you made? :-)


Reply 3 years ago

Sorry, it's not for sale. Since I etch my PCBs I'm not confortable in selling them, maybe you know someone nearby that will make the PCB for you.


Reply 3 years ago

Thank you for the
reply. J I can appreciate your
situation. I want you to know that your prodject is EXTREEMLY INTERESTING and a
WONDERFUL IDEA! The mount ing of the
tubes is brilliant! ;-) You should consider the PC board option, if there are
enough responses like myself. $30.00 T0 $50.00 WOULD BE

Keep up the great
work!!! KUDOS! J



3 years ago

Why is a toroidal transformer not suitable?


Reply 3 years ago

If you use my layout, size is the first concern.

I used toroidal inductors in some builds and for this particular application they get really hot, and the voltage drops.

Some say that it occurs because the toroidal inductor has no air gap, and saturates more easily with DC current, which would also explain the heat and loss of efficiency I observed.


3 years ago

That's a cool project! Great work!