Project History

In the beginning...

I did not set out to design a board for DIY headphone amplifier power supplies. From the start, I primarily recommended commercial power supplies for the various amplifier designs I talk about on elsewhere this site. I’d only occasionally recommend a DIY’d power supply, and that mainly just to save money over the commercial offerings.

Years after I started in this hobby, I finally bought some board etching equipment. After making some simple boards to learn the ropes, I decided to make one of the STEPS’ predecessors. The initial impetus for designing this power supply was so I could test the 18×AAA cell configuration of the PPA battery board, which requires 32 V for the charging voltage. It’s essentially impossible to find a commercial supply over 30 V but under 48 V. 30 V just wouldn’t cut it, and 48 V would kill the battery charging circuitry. Therefore, I had no choice but to DIY my own power supply. At this point, I couldn’t see this as a practical DIY project for others. The parts cost was about twice that of the commercial supply I recommend most often for lower voltages, plus you had to build the supply, and at the end of it all you only gain a little more battery voltage. I figured most PPA battery board users would just use fewer cells rather than tolerate all this.

Since it was still just a personal project at this point, it wasn’t until months later that I took the next step: I made another hand-etched PCB, this time a crude predecessor of my LNMP measurement preamp. Its purpose was to let me measure the residual noise and ripple coming from linear power supplies. A good linear supply has an output noise well below below the noise floor of affordable commercial test equipment, so the LNMP was necessary to be able to measure this noise. I gathered up all the linear-regulated power supplies I had sitting around my house, including my recent STEPS predecessor, and tested them with this new measurement preamp. To my surprise, I found significant performance differences. The STEPS noise was over 12 dB down from an expensive, well-regarded audiophile power supply. (The most recent iteration of those tests is documented here.)

Just that one result eliminated my worries about practicality: it was possible to do better than some commercial supplies, for less money. Furthermore, its audiophile class performance made it useful for all kinds of amplifiers, not just for PPAs with battery boards. Thus reassured, I turned the STEPS into a production PCB project.

Revision History

STEPS v1.0 is the result of combing through many 3-terminal regulator power supply designs, and assembling what I believe to be the most sensible set of concepts that remain self-consistent. The STEPS is not distinguished by any single idea; they are all old ideas, taken from elsewhere.

Version 1.1 was a short prototype run to test improvements to STEPS v1.0, adding the following features:

• Made line filter choke smaller to keep primary side farther from secondary side.
• Enlarged main filter caps to allow 18 mm diameter caps.
• AC input wire pads are larger, allowing 18 gauge wire
• Board size tweaked slightly to allow it to fit directly in Hammond 1455N12 cases.
• On-board 5×20 mm fuse clips.

In addition, the board is a bit cheaper now, and some of the parts changes above allow still more savings. You can save a good $10 relative to a similarly-configured STEPS v1.0.

Version 1.2 is essentially the production version of 1.1, with some minor changes:

• Regulator moved inward 25 mils, to get it a bit farther from the case wall in the standard Hammond case.
• Fuse clips moved farther apart. In v1.1, you had to use fuse clips without "ears", or straighten out the ears with pliers.
• Added 5 VA, 7 VA and 10 VA Amveco transformer pads.
• Added thermals to several of the larger pads.
• C4 labels were missing. Fixed.