Modular Audio Amplifier System design philosophy

Abstract

In another article I explain the technical concept behind the modular audio amplifier system. In this article, I like to explain my general design philosophy that is applied to every of the modules and the whole project in general. Every project needs a general design philosophy so that the direction is clear and things fit together in the end. I have my philosophy in mind and elaborating here helps me to focus and others to follow my ideas.

Modularity

modular audio ampolifier system logo
When it comes to audio, I observe countless preferences for all kinds of equipment and countless options to pair equipment with each other. Different signal sources like vinyl, tape or CD, different pre-amplifiers or sound conditioning, different power amplifiers using either tubes or solid state electronics, running with our without feedback and all kinds of different loudspeakers can be wired together. Almost each and every device can be paired with each other and may sound better or worse in different combinations. Modularity is key in order to experiment with different combinations. The modular audio amplifier system brings the same kind of modularity down to the level of the power amplifier interior. Connecting equipment with each other requires standardized interfaces (or adapters) and the same is true for modules inside the power amplifier. By experimenting with different modules and different wiring schemes like feedback loops, the optimum setup can be identified. Since the effort to change modules or even basic mode of operation is comparably low, the risk of being trapped inside confirmation bias is low as well. Imagine the amplifier designed and built, that has consumed so many resources, sounds like crap in the end. Who would not call it the best ever built instead of realizing failure and taking it apart again to start over?

Cost vs performance

engineering triangle
In engineering and many other fields, everything moves inside the magic triangle "cheap, fast, good" (pick any two of them). Usually cost and resources are minimized while maintaining acceptable performance. Especially for high volume products, limits are tested to see what can be removed, reduced and downgraded until performance becomes unacceptable. What is deemed acceptable depends on corporate philosophy and sometimes also standards. Corporate philosophy also determines what the product is optimized for.

In the crazy zones of engineering, where products are optimized for unusual performance requirements or applications, the product will be on the edge of the triangle with less focus on time or cost. This is true for HiFi as well as other fields like racing cars. It seems that there is no real standard for what is considered HiFi or high end and the products I have seen so far were all engineered towards different goals and show varying quality. HiFi or high end seem mostly a subjective attributes that are not measurable.

This translates HiFi and high end into deliberate over-engineering with the goal to achieve highest performance. While the high end products are designed to outperform the low cost optimized segment, chances are to find some gems in the mid price segment.

The difference between bare functionality and something that offers a lot more performance, is often subtle. A few inexpensive components can make a huge difference. But those components may be omitted in many products for cost saving or because they don't contribute to the performance metrics defined. The customer often cannot judge or does not honor higher quality at all - so why bother?

I clearly want to position this project in the DIY high end segment and this is why everything is engineered towards more performance. Of course, cost is still a constraint, but in case I can make a good trade-off, I favor performance. Everybody can buy something ready made from any price segment of products, enjoy and be happy. Or build something that is totally inferior to even the low end segment, but enjoy the building process and be happy, too. Something self-made is always something special, even if it is inferior. I have built a crappy and ugly amplifier already and now it is time for something better and I aim to beat the better segment of commercial products.

HiFi, high end and the modular audio amplifier concept

What is HiFi and High End st all? My impression is that both terms are not clearly defined. HiFi seems to be the upper segment of the normal audio product spectrum and High End is the extraordinary high cost and boutique segment above the HiFi segment. I don't know whether there is an official definition, but would say both segments are defined by following attributes:

  1. Exceptional performance
  2. Unique technology
  3. Extravagant appearance
  4. High retail price

Performance

While most most manufacturers of audio and HiFi equipment provide some performance metrics for their products, this seems less common in the high end segment. Instead, claims about exceptionally superior sonic properties are made. I guess this is having to do with some of the products not scoring high in the competition for numbers that are generally considered to be metrics for audio quality. I tend to begin to question performance metrics as well since this has not lead to anything meaningful in other segments of engineering as well. The output power of an amplifier is equally meaningful to sound quality as the number of megapixels is to the quality of a picture. There is some relation, but it is more complex and metrics need to be considered in relation to each other.

Technology

A product positioned in the high end segment needs to have something unique or at least claim it has. Some products really have some unique, advanced or high performance technology. But in general, this is the point where marketing nonsense begins. "Our amplifier purifies the electrons in order to create more space and life in the music reproduction." would be a fictional example of common marketing nonsense companies may use to convince the technical layman into spending a fortune for their products. My observation is that claims are typically phrased to empathize properties that cannot be proven at all.

Appearance

Most things are foremost judged by their appearance. The first impression makes a huge difference. And many companies in the high end audio segment really invest a lot in extravagant product design. I can't see anything wrong with this. Who would want to place anything butt ugly in the living room? When I visited the High End trade fair in 2018, I focused on high end amplifiers chassis design and share my impressions in an article here on my website. Some companies really impress with great design, other seem to care less. So it seems that this cannot be generalized.

Price

All of the a fore mentioned properties translate into extraordinary high retail price of products in the high end segment. Power amplifiers can easily cost 20.000€ or more with open end. Even cables may cost more than 100€ per meter, also with open end price range. The price is so high because the market is small and therefore R&D, boutique manufacturing, huge marketing campaigns and fancy exterior cost a lot. Sometimes value for the money is rather questionable, but I've also seem outstanding engineering and craftsmanship.

The modular audio amplifier in the high end segment

So how can I position my modular amplifier project in the high end segment?

I can spend a lot of time on R&D since the whole project does not need to be economic. I don't need to pay myself any salary. Commercially oriented companies investing that much inevitable end in the high retail price segment. I've been working for years on this project, although with very low intensity and prolonged breaks. During development I follow an iterative approach questioning previous results continuously and therefore perfecting each part even in case it means to dismiss a lot of previous work.

Pretty much everything is invented so far, so I won't invent something entirely new, but I investigate circuits not commonly used whether they could offer superior performance and just aren't used more often due to high complexity and cost. Paired with proven technology, this can result in interesting solutions that outperform more simplified implementations.

Matching some of the excellent chassis designs of commercial high end products is not possible for me, but considering many basic chassis designs I have seen at the High End trade fair, I probably don't need to hide my chassis. I don't have the resources to fund machining of something outstanding and unique, although I have ideas for outstanding chassis designs. The exterior has always been the most difficult part of the project for me.

I believe it is possible to build something with extremely good value for the money due to the fact that in DIY projects R&D and labor are free. So there is an option to build an amplifier worth 20k for a tenth or less worth of material cost.

After all, a unique hand build instrument is something that for its own scores pretty high on the extraordinary scale.

Optimization

As pointed out earlier, products can be optimized for many different goals and often it is possible to reach several at once. To some extend, numerical performance translates to subjective listening performance, but quite often, this is not related. And even the numerical performance metrics to consider are unclear. What measurement values to optimize for? THD? IMD? PSRR? Noise? Watts? Energy efficiency? All of them at once? And how does each of them relate to perceived sound quality?

Usually low distortion is achieved by large amount of global negative feedback. Global negative feedback is somewhat frowned upon in the high end segment. As I see the situation, GNFB has gained bad reputation in the past by applying it to circuits that were highly nonlinear to begin with. No surprise that the result was unpleasant and GNFB dismissed as a good concept by many audio enthusiasts. In fact, pretty much any amplifier claiming to use "zero" feedback, does actually use feedback, but locally instead of globally. Even emitter degeneration is feedback. As always, things cannot be generalized that easily. I believe that mindful use of feedback is the key.

Nelson Pass has written an excellent article about the matter of harmonic distortion, reduction of distortion by feedback, the side effects thereof and how the effects are perceived by the listener.

The ability to suppress power supply related noise on the audio signal should have topmost priority in my opinion. Harmonic distortion is related to the signal, while the power supply grunge is absolutely not. Therefore, I emphasize to clean the power supply as much as reasonably possible and improve power supply rejection wherever possible. A cascade of filters in the power supply chain adds only negligible cost, but for sure lowers power supply noise, which contributes to better performance both numerically and subjectively. Furthermore, the circuitry handling the audio signal needs to have inherently high PSRR on its own.

Clipping performance is something I care a lot about. Clipping should be handled gracefully. Clipping the audio signal does a lot of damage already. Adding injury by the amplifier freaking out during clipping is both undesirable and unnecessary. That means the amplifier needs to tolerate this condition without showing any signs of instability. Also, clipping should be soft since this is rated more pleasant subjectively. The front ends I have under development use sophisticated circuitry to handle clipping very well.

I prefer to keep the signal chain as short as possible. While the signal has suffered a lot of awful processing and serious damage thereof earlier, and it is somewhat unlikely that the last "inch" does the most damage, any unnecessary processing is best avoided.

The big benefit of the modular approach is low risk of investment and evaluation. In case any module fails to deliver the performance or sound expected, this single module can be replaced with a different one or redesigned for optimization without affecting any of the other modules. The resulting amplifier can be configured an many ways and be quickly reconfigured. This way, the amplifier can be optimized a lot easier than with the traditional approach to put as many functional groups on a single PCB as possible. I see the modular amplifier concept as a good way to customize and perfect the audio power amplifier.

Simulation

I find simulation very helpful designing electronic circuits. Based on simplifications and assumptions, predictions regarding performance can be made. It is important to be aware that those predictions are not accurate by nature and should be verified in real life. I find the inclusion of the PCB in simulation is extremely important. I see a trend to just omit doing so. It is pretty pointless to believe that a circuit wired together in the simulation environment by unrealistic zero resistance, zero inductance connections without any stray capacitance will work in real life just like simulated. I maintain two sets for critical circuits: One in an ideal environment and another one that aims to somehow reflect the actual PCB design of the circuit. While this is a lot of effort, it has revealed sensitive nodes I had not considered being sensitive before and allows to implement measures in order to mitigate the situation beforehand.

Measurements

Measuring audio performance is extremely difficult for the DIY audio builder since precise measuring equipment is prohibitively expensive and therefore financially far out of reach. Some people have shown good results improvising and building measurement equipment. Measurements are important to verify proper behavior and ensure no ill conditions like oscillation occur. Even applying a purely subjective attitude should not result in disregarding basic numerical performance parameters. Given the limitations, making numeric performance priority is not an option for me anyway.

Components

In the audio, and especially the high end segment, sometimes esoteric components are preferred due to their magic properties. I have way too much engineering background to believe in any magic and do not use any special magic components for my designs. I like components that deliver solid performance at reasonable cost. I like widespread packages that allow to interchange components easily. Second source is important as well as more economic options from China for non-critical components like heat sinks.

Capacitors in the audio signal path should be either film, mica, ceramics with NP0 / C0G characteristic or worst case audio grade bipolar aluminum electrolytic. I don't need any hand rolled caps in wooden can using bees wax as dielectric or any similar esoteric parts.

I dislike soldering wires to anything and avoid cable harness wherever possible. Instead, I prefer point to point connections using connectors or terminals. Small loop area is important for both immunity to pick up and emission of electromagnetic interference. Low voltage audio signal connectors should be gold over nickel plated by all means.

Surface mount components offer many benefits, but are difficult to process for the average DIY builder so I prefer through hole mounted components instead. THT components also offer more mechanical robustness due to built-in strain relief.

Interface

My preference for interfaces are symmetric XLR for the inputs and Speakon for the outputs.

While unbalanced RCA connectors are still de facto standard in most of the consumer audio segment, this interface is so much inferior and causes so many issues like hum from ground loops that I find it not worth considering for anything that aims for higher performance. Symmetric signal transmission is by far the best way to achieve excellent noise immunity and freedom from ground issues.

Binding posts for speakers are great because this is the most simple and universal interface, but have some disadvantages as well. For me the foremost issue here is that it is easy to accidentally cause a short circuit at the amplifier output, which may damage the costly output stage of the amplifier. Another risk is high AC voltage present for the operator to touch. While this is only a minor risk due to the low voltage of typical domestic low power amplifiers, it becomes a serious safety concern for high power amplifiers. Lastly, connectors are convenient in handling and provide repeatable contact force and contact resistance.