What is your opinion about the new technologies in electronics like multilayer boards and surface-mount components ... will they help to improve sound paths and quality when compared to what was available in 60’s and 70’s?
There is no general answer. There is Yes and No. The only way to find out is evaluation. For us this always implies a twofold approach: Measure AND Listen. We use very low impedance circuitry with lots of buffer stages in order to minimize noise and components losses. So amplifier stages are a crucial issue. Our experiences concerning active components like modern audio OpAmps are very positive whereas it is not always the manufacturer’s flagships which show the best results.
So in AM1 or EQ1 we do not use discrete amplifier design. I often hear the opinion that high quality circuits would have to be built “discrete”. I do not agree. It was true in the past, but not today. I am not saying that integrated circuits are better, but with a discrete design it is extremely complex and expensive to get the same performance. Today I would rather use discrete circuit design to achieve a certain sound, and a certain sound means distortion. But our philosophy with AM1 and EQ1 is preserving the character of the audio material. To reach that goal there is no need for discrete amplifier stages. There are brilliant integrated circuits available. They provide perfect compensation techniques for temperature effects which can hardly be achieved by a discrete design. They contain exact, laser-trimmed elements. Because of their minimal mechanical measures they are easier to shield against stray pick-up of electric or magnetic fields. My conclusion: best quality, more reliable and smaller form factor.
And since the latest generation of integrated analog switches the use of mechanical switches or relays for audio signals is no longer needed, except for some special applications like hard bypassing, microphone signal routing or high current outputs. Our experience with passive components is different. SMD resistors, as long as they are high quality thin film types are as good as conventional parts. But for audio capacitors we have not yet found SMD equivalents. So we use the well-proven conventional types. Multilayer boards suit well to our low impedance circuit design, therefore we use them thoroughly. In combination with SMD components they allow space-saving design which is also less sensitive concerning stray pick-up.
So I conclude: Yes, with the new technologies we can improve sound paths and quality a lot. But let us be accurate about the term “quality”. The sound of the 60’s and 70’s was to some extent characterized by circuitry and components which produced musical distortion, which we also perceive as quality. This aesthetic quality is not the issue of this article. We are lucky that there is a lot of vintage equipment available as well as the knowledge. So we can choose whether we want this special sound character or not.
Can you comment on your EQ circuit design in relation to common designs?
High and Low EQs are based on a Wien-Bridge filter element. A special circuitry provides shelf as well as bell characteristic. The circuit design is rather simple, it works with only a few active elements, but requires precise component matching and multiple individual trimpot adjustment.
The Mid EQs are based on a modified State-Variable filter element. It provides bell characteristic with independently controllable gain, frequency and Q (quality factor). It has higher component count but is easier to handle concerning component matching.
This is the reason why we tried to rely exclusively on this design at some of our products years ago. But as both EQ types have their own special character, we finally decided to use both for best performance and versatility.
The High and Low EQs are very natural and strong sounding, in shelf mode they have very smooth slopes. The High Shelf has a particularly airy character.
The Mid EQs are more the precise, surgical type, though they can be quite smooth with the Q set to the lowest value.
Both Mid bands have a switchable frequency range so they can overlap or even interchange with High or Low EQs - which opens up a whole multitude of options. Imagine using both High EQ and High Mid EQ with “frequency x 3” in the high band. High EQ contributes its soft slope and High Mid EQ its surgical potential with narrow band - resulting in creating your own filter type for emphasizing certain formants or removing annoying frequencies.
On the Low Cut we decided for a very musical 12dB/oct slope.
Do AM1 and EQ1 impart a character on the signal or are they “clean” devices?
Our amplifier stages preserve the original signal very precisely. But of course, once you use the EQ stages, they show their character. For some people “clean” can have a negative meaning, such as not musical, clinically cold, etc. This is not the case. But if you divide analog units or EQs in “clean” and “dirty” sounding, we are more on the clean side. We do not intentionally add distortion to the signal.
Just check it out yourself.
Typical feedback from our customers is that they can use it with virtually any audio material. The result is always highly musical, whether they perform small corrections or explicit treatment. Engineers told me they used it to recall tracks back to life after they got stuck in a digital audio work station.
Others – who see the use of EQs rather critically and if, they only use it for subtracting certain frequencies – say that AM1/EQ1 belong to the very rare devices which allow adding gain to frequency bands, always enhancing the character of the audio material.
Do you use any transformers or inductors in the signal path?
We very much like transformers and inductors. They can add great colors to the sound. These colors, even if they are rather subtle, are massive compared to what our amplifier stages do. So we prefer not to integrate these components into our devices. I would use a separate unit with transformers in case I really want this special sound.
Do you use Class A amplifier stages?
Let me preface: Here we are talking about circuitry from the Pre-Amplifier range. Not about Power Amplifiers.
A common idea is that Class A is a positive quality criteria and an amplifier stage with feedback a negative one. I disagree. This is not the matter. The only thing that matters is that the developer makes a perfect job on every single detail.
Pre-Amplifier stages with pure Class A – without feedback – do not suit our circuit design. We use many amplifier stages throughout the unit, because our circuits are extremely low impedance designs. For that, Class A is not practical. Besides others one simple reason is that the power consumption and the generated heat could not be handled within this form factor. But that does not mean we are compromising. Amplifier stages using a feedback loop can be designed in a way that they do not show any disadvantages compared with Class A. From my experience they can even more equal the original signal than a Class A design. All amplifier stages we use are audiophile designs which do have not the slightest crossover distortion, which was in the past the main reason to choose Class A. And we also are well aware that some Class A designs stand for a certain sound character or even sound coloring. Here we follow the same philosophy as concerning transformers: we very much like them, but want to be free to choose having them in the signal-path or not.
We systematically define the selection criteria for our OpAmps. We do not only check the datasheets or compare measuring results. But we carefully listen to all our amplifier stages throughout the development process. We analyze their distortion spectrum for harmonic or non-harmonic distortion. We make sure, that our amplifier stages work extremely precise AND musical – in case of doubt, we decide for the musical side.
In my view, some of the Class A hype is based on a very liberal interpretation. Many units that claim to use Class A actually use Class AB with a slightly higher quiescent current. Just think: A single Class A amplifier stage without transformer strictly built to the rules with a drive capability of +24dBu to 600 Ohms consumes nearly the same amount of power which is supplied by one slot of a 500 series rack. This makes me wonder … For the sake of completeness: Using a transformer, the Class A energy balance is much better. And in the past one or two active stages per slot were enough.
Do you follow general design rules different from other manufacturers?
I do not know about other manufacturers. But I can comment some of my own. Neither best electronic components nor top circuit design are enough for a good piece of equipment: They have to be implemented carefully in the best possible way. Each part has to be placed at its best position in the PCB design. Every minimal factor concerning grounding - I mean the internal grounding, the Common Rail - has massive impact on the sound result.
Designing the Common Rail has been my main issue right from the beginning of my design work. It is not enough having a massive Common plane or wiring. Signal currents which flow into the Common rail will always leave their tracks, will always have an effect throughout the unit. You need sophisticated techniques to compensate these effects or to avoid them. There is an expression in German: Stromlose Masse. I do not know whether there is an English equivalent. It means to avoid any currents going into the Common rail. Though it is not possible to achieve it to 100%, it is our design target as far as possible. The Common Rail is the reference for everything. It needs to be totally quiet.
We consequently use low impedance design. It does not only stand for minimizing stray pick-up or crosstalk, but also for minimizing odd effects on passive components or PCB layouts like parasitic capacitance or inductance.
Each circuitry needs its dedicated design. It is not guaranteed to improve a circuit by substituting e.g. an OPAMP by a better one. You have to be concerned about lots of things, e.g. the design of the supply rails or capacitance of PCB tracks. So a component "upgrade" can sometimes lead to a massive downgrade in performance.
Why do you prefer a linear power supply?
On the switched PSU which is used for audio circuits switching frequency is far higher than the audio-band. And ideally any interference of switching frequency to the DC power rails should be filtered out. But reality is not “ideal” in a technical way. In reality, when comparing DC generated from a linear PSU with DC from a switched PSU you can well measure the difference. Linear PSU can be just about perfectly “clean”. Line frequency, 50 or 60 Hz, and line interference, which can go up to radio frequencies, can be nearly perfectly filtered out or regulated by active voltage controllers. A switched PSU as clean as a linear PSU has not yet come my way. And though the interference frequencies which can be measured post filtering are beyond the audio-band, there is a difference in practice when you listen to an audio system, supplied by either linear or switched PSU. So for the time being I clearly prefer a linear PSU. Maybe in the future we may come across a convincing solution or do some research ourselves. I could imagine a combination of extremely quiet switched PSU followed by massive filtering and linear post regulation. We shall see.
One other issue is that the DC output should be as low impedance as possible. Only this can prevent interference between different devices supply by the same PSU. A powerful voltage regulator is a must, as well as short cables to the load.
AM1 has dedicated voltage regulators on each channel module and multiple on the master module. EQ1 has its own internal voltage regulators. For best results even in case the external supply is not perfect. But they will surely benefit from a clean supply. In most cases interference are not limited to supply rails but also affect grounding and common rails, which of course will affect audio signals in one way or another.