This is a project developed by Douglas Self and published in the Elektor in October and November 2010.

The idea is rather unusual: "An interesting power amplifier can be made by connecting enough 5532s in parallel, how about 32 for a start? This may sound like a radical course of action, but it actually works very well, making it possible to build a very simple amplifier that retains not only the excellent linearity but also the power-supply rejection and the inbuilt overload protection of the 5532, which reduces the external circuitry required to a minimum."

Elektor used to sell PCBs for this project, which I used. I modified the circuit slightly to incorporate an active volume control along the lines of Douglas Self's preamplifiers.

Here is the result:

My first solid state amplifier build, in 2009, was The Penultimate Zen, aka Zen Variations Part 4, aka ZV4. It was built with the parts I could find back then, without knowing where to look. It ended up big, heavy with two 400VA toroidal transformers, ugly inside with some electrical tape here and there, without proper grounding, etc., etc. I was as disappointed with the sound as my wife was disappointed with that heavy black metal brick gathering dust.

My first mod on ZV4 came after I turned my disappointment into finding better part suppliers. I returned to my ZV4 and replaced the output caps with something marginally more decent from eBay. The sound improved a notch. Aha!

My second mod was to rebuild the power supply. I threw away one of the toroids, added a softstart circuit from Hypex, and put in a CRC filter with some Mundorf HC caps. One power supply for two single ended channels forced me to learn about grounding and add input transformers and balanced...

I built this one a while ago but could not find time to post.

This is a Class B power amplifier that follows the Ultra-LD Mk.3 amplifier design published by the Australian Silicon Chip magazine in 2011, which in turn draws heavily from the concept of a Blameless amplifier devised by Douglas Self. I used a different power supply and speaker protection, and changed the grounding scheme vs what was published.

The distortion at 1 kHz is below my measurement capabilities, and the amplifier does sound very nice, although it seems to give more sibilance than my Class A amplifiers.

The acid test I use is the dual CD "The Very Best of Placido Domingo" album published by EMI Classics. Many amplifiers I have heard cannot deal well with Domingo's tenor, esp. on louder and higher tones. Class A amps, particularly those using simple internal structure like the Zen series, perform well in this test. This amplifier seems to do better than any other...

Big, hot, and heavy! Just completed it and adjusted for distortion, have not had yet a real chance to listen to it.

The build is in a 4U/400 case from modushop; each side has two 200mm heatsinks, each holding six MOSFETs (three complementary pairs) and a biasing circuit.

The construction is dual mono (separate transformers for each channel) with CRC filtered +/- 18V rails, quiescent current is 3A per channel. Eight Mundorf MLytic® HC High Current Power Caps and two 300VA toroids occupy the most of the chassis, while the actual electronics is mounted on the sides.

Now I need a balanced preamp. I contemplate building some variant of BoSoZ, but am also thinking about a SuperSymmetric balanced preamp using JFETs or tubes...

Just finished fine tuning my First Watt F5 clone.

I built it about a year ago out of parts that I had on hand back then, including a power supply reused from another project. The resulting sound was rather disappointing. Now I spent time matching transistors and tuning the feedback network. The amplifier now is very close to the performance described in the original service manual.

I am listening to F5 (the music is Tchaikovsky performed by Wiener Philharmoniker and von Karajan) and realizing that it is so good that if this was my first amplifier, I might not be interested in building anything else.

Update: after a year, this is still the best amplifier.

Update: I purchased the official F5 PCBs from the diyAudio store. I will rebuild the amplifier, hopefully with better FETs and other parts.

Now that Nelson Pass published the details of the JFET upgrade to First Watt F2 and that Newark sells power JFETs from Semisouth, there was no excuse not to upgrade my Zen V4 with the new active device.

I replaced Q1 (see the original schematic) with SJEP120R100A; increased R5 to 130 kohm to bias the JFET correctly with Vgs of approx. 1.5V; and reduced R8 to 22 ohm.

The measured result is a nice reduction in THD+N; attached are the graphs for THD+N @ 1W into 8 ohm vs frequency, before and after the upgrade. With IRFP044 as Q1, the distortion was mostly 2nd harmonic (at -65dB), plus some traces of the 3rd harmonic. With the JFET, it is still mostly 2nd harmonic, which is now at -80..85dB, with the 3rd below my measurement floor.

Subjectively, the improvement is remarkable! There is additional detail and depth of the scene.

In 2009, Penultimate Zen aka Zen V4 was my first solid state build. I used the parts that were available, did not know much about making it work well, and did not have test instruments beyond a cheap multimeter. Also, I did not have a preamplifier to work with it or speakers sensitive enough to enjoy its 25W. As a result, the amplifier turned out not quite satisfactory and for two years has been gathering dust and my wife's complaints. I was thinking about scrapping it and reusing the enclosure for something more useful. Instead, I reworked it.

I removed one of the two power transformers (moving from dual mono to stereo configuration), replaced my original dodgy Chinese PSU filter caps with Mundorf M-Lytic HC, replaced internal low voltage wiring with thick speaker wire, reconfigured grounding (so it is finally safe!) and installed balanced inputs with input transformers. What a change in sound! With my Exposure CD player, B1 clone preamp and Heresy III speakers, the amp...

Unsatisfied with the sound of my gainclone amplifier, I re-used the enclosure and the power supply for a gainclone along the lines suggested by Bob Cordell, whose implementation of an LM3886 based amp was praised by at least one member on the NJ audio society.

I skipped both the Klever Klipper and the toroidal air core output inductor, and kept only 10,000 uF per rail in the PSU. The schematic can be found in Chapter 27 of Cordell's Designing Audio Power Amplifiers. The PCB was designed to re-use the existing mounting holes of the ChipAmp's PCB.

The result? Better than with a plain vanilla chip amp, but still not good enough for music. Perhaps I should not have limited myself to re-use of the PSU et al. but should have taken all the details of my implementation seriously.

To try out Class D without working too much, I picked two ready-made Class D amplifier modules: one is the $45 2*100 Watt Class-D Audio Amplifier Board from Sure Electronics with the optional $10 volume control board, the other is the $199 IRAUDAMP7S from International Rectifier.

The Sure Electronics module is based on Tripath's TK2050 chipset. Powered from a 150W 24V SMPS from Mean Well, it predictably puts out about 22W RMS to a 8 ohm load. (The declared 2*100W requires a 30V supply and a 4 ohm load.) The sound and the measurements are decidedly mediocre, although at the $45 price, the module still may be a good value. Also, the volume control board feels odd, as the knob only adjusts volume after a push; another push disconnects it again.

The IRF module requires dual rail supply, for which I chose an SMPS400A180 by hypex. With +/-40V rails, the amplifier delivers 100W RMS into a 8 ohm load. The measurements are good, and the sound is very interesting for...

I wanted to build a Class D amp based on TI's PurePath chips - e.g. TAS5086 PWM processor and a few TAS5162 power stages - until I read Bruno Putzeys' "The Truth About Digital (Class D) Amplifiers". He makes a convincing argument that digitally controlled class D is a dead end street. I also noticed that very few people around here have been writing about PurePath. Shall I look at the UcD patent and application note, as well as ready made Hypex modules, instead?