I've been interested in DIY audio, and after reading a blog post about someone building a small power amplifier with transistors, I wanted to try it myself. My reference was Chapter 4, "Designing a Mini Power Amplifier," of "Teihon Transistor Kairo no Sekkei" (The Definitive Transistor Circuit Design). It walks through a 0.5 W amplifier very carefully, from choosing component values all the way to evaluation, and I really recommend it to anyone about to build one.
The design specs are a voltage gain of 10x (20 dB), at least 0.5 W into an 8 Ω load, a frequency response of 20 Hz to 20 kHz, and distortion under 1%. The circuit consists of a common-emitter voltage amplification stage and a class AB push-pull current amplification stage. I learned a lot from the parts about adding diodes to eliminate crossover distortion in class AB, and thermally coupling the transistors to the heatsink to prevent thermal runaway.
I don't own an audio analyzer, so I filled the gap with LTspice, a free circuit simulator. Measuring the gain from the input and output waveforms in a transient analysis gave about 10x (20 dB), almost exactly as designed. I then checked the harmonic content with an FFT and calculated THD (distortion) with the .four command, getting about 2%, which closely matched the figure in the book.
I designed a PCB for the first time in EasyEDA and ordered it from JLCPCB. With a coupon it came to about 500 yen, surprisingly cheap, and shipping from China took about a week. I was nervous about whether it would actually work, but it produced sound properly. Distortion is low, and it plays as pleasantly as any ordinary speaker setup. Next, I'd like to take on power supply circuits and digital circuits too.
Hackaday features a nice repair-and-rebuild story. A builder known as Boz was given a broken Pioneer surround-sound receiver. It sat on a shelf for a few years until he had the idea of building a jukebox for his ripped CDs and bringing the receiver back to life with a new main board.
Modern home entertainment amplifiers are full of digital functions, and when those fail they are often proprietary and hard to fix. This one turned out to be surprisingly modular, though. The front panel with its VFD display and driver still worked, and so did the class AB amplifier and its power supply. With the service manual in hand, reverse engineering was straightforward, so he removed the main board and replaced it.
He reused the original connectors and a few other parts and designed a PCB that carries a Raspberry Pi Pico and a DAC. With MMBASIC firmware the result looks as if it had been made this way from the beginning, and the look of the board inside, apart from a couple of bodges, adds to that.
Hackaday notes that this one is special in terms of quality and that revival projects like this are usually seen with gaming hardware.
I like that the working analog power stage was kept and only the dead digital part was replaced. It is a practical idea for anyone who has an old receiver in a closet with good amplifier parts and a broken brain.
A project on Elektor Labs caught my eye as a turntable fan: SchickeRIAA, a phono preamp for MC cartridges in a compact through-hole build. It is based on an older design by John Linsley Hood, in which the RIAA equalization is not a passive network but is moved into the negative feedback loop of the amplifier stage.
The input is a THAT 1510 instrumentation amplifier with a linear gain of 40 dB. It also inverts the phase, so the whole chain ends up with no phase inversion. The equalizer stage that follows uses an OPA227, which the author chose for its low 1/f noise and very high open-loop gain. At 20 kHz the gain is 1, so no extra pole is needed for high-frequency correction. The lowest frequencies, such as the rumble from warped records, are already filtered out before the high-gain RIAA stage.
The power supply is especially interesting for home builders: there is an external transformer, a Delon voltage doubler on the board, and a regulator. An earlier version used a 317/337 regulator pair with a zener reference, while the current one uses an actively feedback-controlled regulator and a modified constant-current source built from two bipolar transistors.
I like that an old concept has not simply been copied but carefully rethought with modern components. The project was published in July 2026 and updated again in early August, and the schematics are on the project page. The final capacitor values are not given in the text, so it is worth checking the documentation and the discussion there.
Tube audio has a reputation problem. Too often it gets taken over by people who care more about bragging rights than about the fun of building the thing, so it was refreshing to read about a project that goes the other way.
Bettina Neumryr specialises in building circuits from old electronics magazines, and her latest is a tube amplifier from 1955. Hackaday calls it a fantastic antidote to the gold-plated silliness of audiophile tube amps, and I can see why.
The design is fairly straightforward: a preamplifier, then a two-tube transformerless splitter driving a push-pull output stage. She walks through the circuit a little, including the feedback path to the cathode of the first splitter tube. The power supply is a classic tube-rectifier design with a hefty low-pass filter built around a giant choke.
My favourite detail is the chassis. Instead of paying silly money for a purpose-made tube chassis, she used a pair of baking trays, and apparently they look surprisingly good. Most parts came from the junk box, including the output transformer, which is where the trouble started.
After a few mishaps and a bit of unintended oscillation, she ended up with an amp that shows promise but has noticeable distortion, which Hackaday puts down to that transformer. A better transformer should bring the figure into an acceptable range. There is a video of the whole build if you want the details.
I like this because it treats the amp as an electronics project first and a status object second. A reminder that the junk box plus patience can go a long way, and that high voltages deserve respect while you do it.
Elektor has released a USB stick with the work of Ton Giesberts, for years one of the best-known designers in the magazine. For anyone who likes building audio amplifiers, it is an interesting collection: hundreds of projects, articles, and design notes from several decades.
Audio is the common thread. It includes the compact audio power amplifier from 2013 (Elektor 201309) and the Fortissimo-100 high-end amplifier, which appeared almost ten years later. The latter has a fully symmetrical output stage. The Q-Watt is also there, a semi-discrete amplifier that delivers more than 200 W into 4 ohms with very low distortion.
But the stick goes beyond amplifiers. It contains a buffer board for the Raspberry Pi 400 that protects the GPIOs and allows connection to 3.3 V and 5 V logic, a USB-to-multiple-protocol serial converter, a DDS function generator, and an AC/DC power meter with galvanic isolation and Arduino. There is also a mobile charging station built on AA NiMH batteries. In an interview, Giesberts looks back on his career and says that older designs are worth revisiting, with some improvements and modern components.
The recommended price is 49.95 euros, with a sale price of 39.95 euros. A similar stick, The Elektor Audio Collection 2026, costs the same. Personally, I find Giesberts's last point the most appealing: a good analog design does not get old, it just waits for a new builder.
For speaker builders, Voice Coil's Test Bench column by Vance Dickason has a new entry on the Dayton Audio E150MR-4, a 5.5-inch carbon fiber midrange from the brand's Epique line, which targets high-end sound at a moderate price.
Some of the construction details are worth noting. The cone is a solid, curvilinear molded carbon fiber piece that looks like the old woven cones, with a solid aluminum phase plug and a positive-roll NBR surround. A 3-inch flat cloth spider provides the rest of the compliance. The voice coil is two layers of round copper wire on a 30 mm titanium former. Titanium has become popular for formers because it is more rigid than aluminum, resists deforming at high excursions and is paramagnetic.
The frame is a proprietary six-spoke cast aluminum design with narrow, tapered 8.5 to 10 mm spokes to reduce reflections. It is open below the spider shelf, and since the driver has no pole vent, air in the gap is channeled out over the front plate, which cools the motor. The ferrite motor uses a T-yoke with a copper cap shorting ring and an aluminum sleeve on the pole. Xmax is 2.7 mm and the RMS power handling is rated at 60 W, depending on the high-pass frequency you use.
The article goes on to the measured results, which I recommend checking if you are planning a three-way design.
Midrange drivers are often what decides how a DIY speaker sounds, so a measurement-based look like this is useful before spending money.
Elektor has picked up the T-Reg in its circuit series, a high-voltage regulator for tube amplifiers. The project comes from Jan Didden and first appeared in March 2009; CJ Abate reintroduces it on May 19, 2026.
The purpose is quickly explained: hum, residual ripple and noise on the anode voltage degrade the sound. The regulator delivers a clean, stable high voltage and is suitable for tube preamplifiers as well as power amplifiers. The output voltage is adjustable up to 500 V, and the pass element can be either a vacuum tube or a MOSFET.
What I find interesting is the turn-on protection. A CMOS 555 timer (IC1) pulls down the LED in IC3 with a delay after power-on; the delay is set by R8 and C3 and, with the given values, is about 30 seconds. IC3 contains an LED that drives an optotriac; when it lights, the optotriac fires thyristor TH1, which applies the rectified high voltage to the pass tube. The control and reference circuit is powered from the tubes' heater voltage via bridge rectifier B1 and capacitor C2.
The editors point out that the article is more than 15 years old and some components may no longer be available. It is mainly meant as inspiration for a new design. The text lacks a complete parts list, and the schematic is included as an image. Anyone working with high voltage should of course be especially careful.
A Leslie speaker is one of the most recognizable sounds in music, known for a tremolo and vibrato that come from physical rotation. Hackaday covered a project by Levi Graves who built a modern recreation, since original Leslies can cost thousands of dollars.
The original worked with two speakers. The treble speaker fed a pair of horns on a rotating platform, with only one horn producing sound and the other serving as counterweight. The bottom speaker faced a rotating drum. The musician chose the speed of both, which created the characteristic movement in the sound.
Levi uses an original Leslie drum, but the sound comes from a 100-watt mystery speaker, and everything is packaged neatly in a speaker enclosure. He uses a single-speed Leslie motor, plus a custom-built foot switch that gives finer control over the rotation speed of the drum.
The article reminds us that today's musicians can model such sounds in a computer, while in the 1930s you had to go to great lengths to get new sounds. Even so, the physical manipulation of sound waves gives a feeling that is hard to copy another way, which is part of why Leslie speakers stayed popular alongside Hammond organs.
Personally I find it appealing that a project like this combines salvaged parts with a custom cabinet and electronics. It is a good example of DIY audio where the mechanics matter as much as the amplifier. Note that the article is from July 2025.
Here is a small speaker project I enjoyed reading: a mini folded horn in the Fostex style, built around a 2-inch full-range driver that costs about two dollars. The author, mircemk, published it on Elektor Labs in September 2026 and updated it shortly after.
The starting point is modest. The driver has no datasheet, so there are no Thiele-Small parameters. The author measured the resonance frequency himself, about 170 Hz, using an audio generator, a series resistor of 10 to 47 ohms, and a multimeter, with the driver hanging in free air. It is a simple method that any hobbyist can repeat.
The speaker is a rear-loaded folded horn. Behind the driver, a first semi-open chamber is filled with damping material, and then an exponentially expanding duct follows. Its dimensions depend mostly on the resonance frequency. The calculations are deliberately rough, and the exact dimensions appear only in the drawing attached to the project.
The author is honest about the result. It is not a high-fidelity speaker but an experiment. Listening against a NAD 801, he notes a noticeable boost between 150 and 300 Hz, a sharper sound on voices, and more lower-midrange energy than a small sealed speaker. There is no real subwoofer extension, and the response measurement, done with pink noise and a small microphone, is very approximate.
I find this approach refreshing: little budget, home measurements, and conclusions without exaggeration.
audioXpress has republished a DIY project article by Tim Mellow and Peter Farrow, first printed in May 2024, that tackles an old problem in tube amp building. Part 1 introduces the Zero Output Impedance (ZOI) amplifier concept and walks through the design of a single-ended input version.
The problem is this: low-feedback triode amps have been popular since the 1990s for their open soundstage, natural tonal color, clean transients and soft overload. But feedback is also what normally brings the output impedance down so that the changing impedance of a real loudspeaker does not alter the sound. Speakers are not a pure resistance. There is a big impedance peak at the fundamental resonance, the coil inductance makes impedance climb with frequency, and drivers with different nominal impedances or steep crossovers cause more irregularities.
Without feedback, the result depends on the output device. Semiconductors, pentodes and beam tetrodes have a high output impedance and act almost like current sources, so output voltage follows the speaker impedance, which can make the bass boomy and poorly controlled. Triodes behave better, with an output impedance roughly a quarter to half of the nominal load, but still not zero.
The authors ask whether you can keep the low-feedback character and still get zero output impedance. They present two designs, one for single-ended input based on a topology from the 1950s, and one for balanced input that is a new configuration. Part 2 covers the balanced version, measurements and design calculations.
I like that it explains the why before the schematic. It is a good read if you build or plan to build a tube amp.
I came across a build write-up that shows nicely how a perfectly usable audio system can be made from spare parts. The author, writing on Habr, had kept two 50-watt Yamaha speakers on a shelf for years, and a board with an ESP32 lying in a desk drawer. He decided to combine them and build a Bluetooth player of his own.
The core of the device is an ESP32 WROOM-32 module, which handles Bluetooth and the peripherals, and a GY-PCM5102 DAC that receives audio over the I2S interface. Volume and pause are controlled by a rotary encoder, and an OLED screen shows a spectrum visualisation. He chose the XH-A232 amplifier simply because he already had it. Its peak output is 30 watts per channel against the speakers' nominal 50 watts, but he says that is plenty of volume for a single room.
The most interesting part for me is the small details that are rarely written about. A voltage divider sits between the ESP32 and the DAC, reducing the DAC output from 2.1 V to 0.27 V RMS, because otherwise the amplifier overloads and hisses. The ratio was found by experiment, and he warns that a different amplifier may need a different value.
Another problem was digital noise from powering the DAC and the ESP32 from a common supply. The fix was to power them from separate converters. Keeping the wires between the DAC and the amplifier short and twisted also helped, and two 0.1 µF ceramic capacitors after the DAC made a difference.
The software is built on PlatformIO with the arduino-audio-tools and ESP32-A2DP libraries, so there was no need to write the low-level A2DP audio transfer himself. He has published the code on GitHub. It is a good example of how cheap modules can become something useful, and how you can learn where noise comes from along the way.
