I mentioned yesterday that preconceived notions and old biases die hard - and sometimes stay with us forever. It's something we're never going to fix as it is simply human nature. But what we can do is educate ourselves as to what caused the opinions to be generated in the first place and then see if they still exist.
One such cherished belief is that class D amplifiers are fatiguing over time. This was absolutely true at one point in the technology's development but I would suggest that, for some of the better designs out there, it is no longer so. Let's take a look at what caused these amplifiers to earn their reputation.
I would characterize, in broad strokes, four area where these amplifiers struggled when first launched:
- Input stage design
- Unforgiving nature of better linearity
- Power supply sensitivity
- Output filter impedance matches
As in any design there's hundreds of little issues that contribute to a product's sound quality but in my opinion these are the four areas that, when taken together, contributed to the well deserved reputation class D amps earned for themselves in their infancy. But just like CD's and digital audio, which were awful in the beginning, products evolve and grow to become great sounding if you give them enough time. Today let's start our focus on the input stage design.
There are four central components that make up a class D amplifier: the input amplification stage, the PWM modulator, the output power stage and its supply, the output filter.
An input amplification stage is rather simple. It is an analog amplifier much like a preamplifier - consisting of a high impedance input pair and a gain stage that amplifies the incoming signal to a larger output than what you put in. This stage is common to every form of power amplifier, from class A, A/B to D - the differences are in the amount of gain and the size of the signal. The importance of this stage is just as critical in a class D amp as it is in any amplifier when it comes to sound quality.
But that importance to sound quality is something the pioneering engineers of class D didn't pay much attention to and the results were hard, bright and fatiguing. Why would they ignore something so important and simple to design properly as the input stage?
They ignored it precisely because that stage is so simple to design. When you're designing a new technology the challenges are huge and you tend to prioritize your engineering efforts: get the tough stuff out of the way, finish up with the easy stuff last. This has a major impact on product development because engineering solutions that don't yet exist are rarely ever finished on time and the tendency for new technology engineers to focus so so hard on meeting deadlines, and getting their new technology working right, that they wrap up the "easy" details quickly - something I call "slap and ship". All the hard unknowns have been figures out, the project is late, over budget, and you have a manager standing over the design team saying "slap on the easy stuff and ship it!" The input stage is "easy".
We saw this in CD players of yore where the Sony and Phillips engineers put years of effort into the amazing concept of optical storage and digital audio retrieval - then slapped on a conventional output stage without much regard to how the whole thing sounded. Japanese receiver engineers focused all their efforts on building in features to their creations and then slapped on a power amp to tie it all together.
Class D amps were no different when input stages were slapped on at the last moment to finish the design and get it to the market - so much engineering time had been spent getting the tough stuff figured out.
The Slap and Ship mentality is so pervasive and ingrained in engineering circles that most never realize they're even doing it. In fact, so common is this mentality that much of the high end audio industry exists because of it. We, and several other high end companies, invented the outboard high end D to A converter because Sony and Phillips Slapped and Shipped their first player's output stages - and the results were so dreadful someone had to do something about them. But we had to stand on the shoulders of giants to be in a position to do this.
Tomorrow we'll cover what was wrong and what can be done with the input stage.
Paul McGowan
