The “Selfless” Amplifier

This is another very long-running project that I’ve just completed recently. (It’s my first big one of the Web 2.0 era, so you can read my blog posts about it.)

The finished amp

The story began immediately after I built my Alexander current-feedback amp, the “Ice Block“. I got a copy of Douglas Self’s “Audio Power Amplifier Design Handbook”, which I’d somehow managed to avoid reading up until then. Self persuaded me that I’d fallen for the MOSFET propaganda and not given BJTs a fair trial, and I decided I must build one of his “Blameless” amps.


Several years passed, and several half-hearted attempts at it failed to get off the ground. I couldn’t decide whether it should be a hi-fi amp, a bass amp, or how big it should be. Once I bought a 1kVA transformer, and built an enormous output stage with eight TO3 devices mounted on an old thyristor heatsink. It languished and I eventually took it apart without ever applying power.

Two moves of house and two job changes later, I decided to quit polyprojecting around and get it done. I made up my mind that it should be a modestly-sized hi-fi amp of around 100-120W into 4 ohms. And, in the 10 years I spent dithering, ON Semi had brought out their new “ThermalTrak” power transistors with integrated temperature sensing diodes, so I decided to use those in my output stages.

Changing from TO-3 to TO-247 transistors meant that none of my junk box heatsinks would fit. As I had more money and less patience than 10 years ago, I bought a ready-made power amp chassis from Tang Hill of Taiwan. This cost about Β£180 by the time VAT and shipping were factored in. It was labelled as an “Art & Power P-100” or similar, and was probably surplus from a batch they made for some other customer. In any case, a quick rub with petrol and WD-40 removed the screenprinted lettering from the front panel.

Along similar lines, instead of messing about with stripboard and the like, I designed and ordered a set of PCBs. The output stage ones are 4 layer, because I felt this would be a really neat way of minimizing loop area, and so radiation of halfwave currents, which is one of Self’s “Eight Distortions”. I laid them out as if I was making some high-frequency inverter for a Tesla coil. The 4 layer boards were expensive, but still cheaper than, say, a cryo treated power cord. πŸ™‚

On the subject of tweaks, I made some choices that audiophiles might find, well, interesting. Several internal connections are made with the green Phoenix Contact plugs, rather than being, say, lovingly soldered with silver-bearing solder. This means that the amp comes apart easily for service, troubleshooting and tweaking. In a slight concession to audiophilia, though, I used RG213 coax and gold-plated SMB connectors for the unbalanced input and feedback paths. Not to mention some nice looking speaker binding posts, salvaged from a pair of Tannoy speakers I found by the kerb with blown drivers.

The finished amp has dual-slope SOA, overtemperature and DC offset protection, anti-thump, soft-start, balanced inputs, and most importantly, a big shiny Β£16 volume knob obtained from Hi-Fi Collective.

The only annoying thing about it is the power transformer, a 500VA toroidal unit obtained from Farnell. It hums a bit acoustically, and leaks flux into the right channel power amp board, causing it to buzz. This is only just audible by sticking your ear up against the right speaker, and certainly doesn’t interfere with my listening enjoyment. But it would utterly wreck a THD measurement made to Douglas Self’s standards. So the solution is to use the left channel for THD measurements. πŸ™‚

And what, you might ask, does it sound like? As far as I can tell, it doesn’t sound like anything. However, it does make my Crown SXA sound like a crusty old heap of junk. For the time being it’s retired.

(PS: it also trashed the Ice Block in a listening test and that is now retired too. I think it must have something wrong with it.)

Driver board schematic

Output stage schematic

Preamp/protection schematic

Eagle schematic and board files for all boards

12 Replies to “The “Selfless” Amplifier”

  1. It sounds way better than the Alexander. As a diehard objectivist, I was expecting them to sound the same.

    The Blameless has what I can only describe as a very delicate and precise upper mids and treble. Cymbals and hi-hats sound nice and natural. The Alexander sounds kind of confused and mushy by comparison, and the stereo image is a bit scrambled. Cranking the idle current up to MOSFET-roasting levels helps, but it’s still not as good. This is borne out by measurements where the Alexander shows about 10dB more distortion.

    The Blameless also wins in the bass department, making the Alexander sound somewhat thin. I’ll have to go back and check the low-end rolloffs of the two circuits.

    I could probably improve the Alexander, but it would be easier just to gut it and install Blameless guts. πŸ™‚

  2. Hi Mason, links to the schematics are in the post above. But you may want to check out the advice in Bob Cordell’s new book- he uses the ThermalTraks with a slightly different bias circuit to mine and claims it to be a bit better. (ask him what the mysterious R3A is πŸ™‚ )

  3. Oh, I forgot to mention: My circuit just puts two ThermalTrak diodes (one NPN, one PNP) in series with an ordinary Vbe multiplier, and this goes in thermal contact with the driver transistors. Cordell’s circuit is very similar.

    My schematic shows R9 as “select on test” but I got best results with it left open circuit. I also used a hefty 10mA bias current for the VAS, because I thought it would increase the diode voltage and help them track better.

    I used the remaining two Thermaltrak diodes for a thermal shutdown circuit, calibrated by dunking a spare output device in boiling water. πŸ™‚

  4. Ah, thanks! I would definitely like to see your Eagle files for personal DIY use. I’m still debating between Class AB or Class D for my four channel amplifier. It’s for an active crossover speaker (Linkwitz’s Pluto design) system. Right now I’m leaning towards Class D for this project because of size/thermal considerations.

  5. Hi Steve,

    amazing how I managed to visit your teslacoil page dozens of times without recognizing this blog. Thanks for pointing me to your posts about your amp on 4HV.
    Your “Blameless” looks truly inspiring. It’s always nice to see someone building stereo equipment without buying “audiofool” stuff.

    May I ask you where the speaker protection relay we have been talking about earlier is located and if you used just an ordinary power relay? Is it the grey box roughly in the middle of the output stage PCB?
    And, as you mention the transformer leaking flux, have you tried rotating it as Douglas Self suggests? Dose the rotation angle in fact make a significant difference?

    Your blog (and Douglas Self) now made me thinking about building a power amp again, and I would really like to have a look at your Eagle files, if you would be so kind. And, once you tested the idea of a MOSFET based speaker protection further, may we hope to see some more information about it?

    Thanks again for this great blog!

  6. Hi Timo
    Thanks for the compliment! πŸ™‚ I’ve added the Eagle files for all of the boards used in this project.

    The speaker relay is an ordinary power relay. It’s not on any of the PCBs, it is mounted just above the fuses, next to another relay for the soft start.

    I haven’t tried rotating the transformer. The wires are too short and I can’t be bothered lengthening them! πŸ™‚

    Since a few people seem to be interested, I’ll do a blog on the MOSFET speaker relay soon.

  7. Hi Steve,

    Thank you so much for your writing, I truly enjoyed it! It also inspired me to maybe repeat your design. Do you offer your pcbs/kits for sale with instructions? It’ll be greatly appreciated. Thank you again for sharing.

Leave a Reply

Your email address will not be published. Required fields are marked *