The DIY Chronicles, Part One

Editor's Note: There is a large contingent of Stereophile readers who design and build their own equipment—the DIY (do it yourself) crowd. Hervé Delétraz from Switzerland has been e-mailing us photos and stories over the last year about his own ambitious DIY amplifier design, so we asked him if he'd be willing to share a chronicle of his progress, starting from the beginning. This is the first in a six-part series written by Mr. Delétraz.

In the Beginning
I have been a music-lover for more than 24 years now. In fact, I began as a sound-lover. In the '70s, the better the specs were, the better the sound was supposed to be. But I soon discovered, as many others did, that low THD had very little relation with the sonic result. At that time, the term "high end" was not used yet—at least not in Europe—and hi-fi (we pronounced it "hee-fee") was dominated by Japanese machines.

In the mid-'70s, a Finnish guy named Matti Otala took another sound approach, trying to offer better sound with solid-state devices. The negative feedback was his target. Instead of applying a global negative-feedback loop, he "simply" used several small, local loops. I put "simply" in quotes because, while this approach is universally known today and seems trivial, Otala thought about it 30 years ago. I guess it really was a challenge in those days to make a working circuit based on this approach. Matti Otala then designed amplifiers for two European brands, Bang & Olufsen and ReVox [The Harman/Kardon integrated amplifiers of the late 1970s and the Citation XX power amplifier, all of which used Otala's circuits, are probably best-known to American readers.—JA.]. I bought a ReVox amplifier at the end of '70s and kept it for years and years—the sound of this machine was simply timeless.

In 1984, during my last year in the Geneva engineering school, my lab teacher asked me if I would be interested in designing an amplifier. By this time I was more than ever in love with sound reproduction, and I naturally answered with a very enthusiastic "Yes!" The challenge was a bit demanding, however. My professor told me that the amp would have to be digital (class-D, or PWM), and that I would have six weeks to build the prototype, write my paper, and introduce it to the final expert jury. Fortunately, some electronics parts had already been designed in a past diploma session, so I wouldn't have to design all the stuff from scratch.

At first I thought it was a nice project. At that time, class-D was not so popular in the hi-fi industry, and no chips were available to directly convert from analog to 1-bit digital (PWM being roughly equivalent to a 1-bit digital stream). The theory of class-D operation had been well known since as far back as 1958, but the technology wasn't yet ready.

After six weeks of hard work, the final result was pretty encouraging. The design was completely free of negative feedback (a first for class-D), and had 10-nanosecond rise/fall times and a 300kHz sample rate. The jury was impressed enough with the sound to give me my diploma with a big smile.

The amp impressed me so much that I decided to build a commercial class-D amplifier, but with outstanding figures and sonic virtues. Then the long, funny, fully involving nightmare began. All of my work had to be done in my free, or "hobby," time, which explains why it took me so long to come up with the final result. Maybe it was better that way; all those years of thinking allowed me to build the device I will soon introduce to you.

In 1984, no chip offered what I was looking for; I was obliged to build all the stuff with discrete components and ICs. My problem at that time was that I wanted about a 1MHz sample rate, which implied very quick analog rise/fall times for a digital application. The jitter (yes, I met jitter in 1984, thank you) was also problematic, even if, in my case, it generated "only" noise. After many years, several prototypes, some fireworks, and nothing really good, I decided to switch (no pun intended) to pure analog, where my experience with high-speed switching circuits might help me . . .

Readers can contact Hervé Delétraz via e-mail at