Thrax Audio Introduction & Tech pt.1

As the 20 year anniversary since the start of this project approaches I am going to do a series of posts exhaling the reasoning behind the development of each of the Thrax Audio products, the technology used, the challenges and the end result with its strengths and weaknesses.

It all started in 2006 with the idea of building a reference remote controlled vacuum tube pre-amplifier with modern features not available at the time from any commercial product.

The project was cost no object and it started with research of the current state of the art offerings, or at least the most prized examples at the time.

During the evaluation many problems in the commercial products where identified. Let's start with the input:

1. Shared ground among all inputs and outputs leading to ground current flowing between sources or sources and power amplifier. This raises the noise floor and creates inter-modulation products.

2. Only single ended inputs and outputs. On some units the balanced outputs where simulated (just a resistor to ground to emulate balanced transmission reducing ground noise at receiver over a limited bandwidth)

Let us stop here for a second. What can we do to resolve those issues? Is there a solution to this?

Having experience outside the realm of high end audio, namely in the recording industry and studio technology the solution for me was obvious. Use transformers.

In science there are a lot of assumptions and virtual concepts that ignore physical reality. I don’t want to go in that rabbit hole but let me attempt to illustrate reality in our audio world. Time for a most fundamental of explanations:

1. The concept of Earth or Ground in Audio and electronics. Imagine Water-world… absolutely calm sea everywhere. There is ground below common to all the oceans.

2. Now imagine our pre-amplifier and our source being boats floating on the calm sea. The boats are anchored to the bottom of the sea (illustrating the connection to earth), so they stay in place and don’t drift. The deck of the boat is your reference point, anything you do is referenced to it, jump run, stack stuff. This is your ground plane it is not the bottom of the sea! For us and the gain stage all signal is relative to that ground plane. This is the reference for all that happens on the boat. Any change in the input relative to the ground plane is considered signal and is amplified at the output which is also relative to the same ground plane. So if we connect two boats and they don’t move at all relative to each other all works fine.

This is what Science assumes. That is what most designers assume. But reality is far from it.

In a real environment each boat moves ever so slightly depending on the waves and wind, so you get a slightly different reading every time, meaning the movement of the boats relative to each other is superimposed on the signal. A moving reference is no reference at all.

But what about ground!! Well it is at the bottom of the sea.

The line (rope) between the boat and the anchor is soft and allows for movement. This rope will represent the impedance in the ground connection of the two units. The lower the impedance the stiffer the rope. When the rope is soft the boat moves freely around but is still attached to the ground, if we go to the extreme and we replace the rope with rigid pillars nailed to the boat we will stop the movement converting the boat to a pier. So this will translate to very low impedance ground interface. What we just did is reference our ground plane to an external factor we assume is stable (the bottom of the sea) Then the signal readout from our source will be perfect and steady if the signal boat is also rigidly connected to the pier, but it is not and there is always some small movement.

Any waves or movement on the boat will create enormous forces, those forces represent the current passing trough the ground connection. Not elegant but workable solution. As long as the two decks are not moving (much) relative to each other this can be “assumed” as common reference.

A brute force approach. So far so good, but this creates another set of problems.

What happens if we add a third and a fourth boat (different sources, power amplifiers). They all move relative to each other. Forcing them with rigid connections will reduce this, but how long will it take before the structure collapses? You have all seen those star grounding arrangements in some of the “high end” gear.

Rocking one boat will make the whole structure move ever so slightly. Leading to the conclusion that one device can move the ground plane of another if they are connected with low impedance kind of defeating the whole exercise.

But what would rock the boats to create those problems? Well think of it that way.

Audio equipment is affected by electromagnetic waves. So same as in water the waves travel in all directions and carry energy. They are at different frequencies and amplitudes and produce different artifacts. As an example AM radio in the old days could be heard from most equipment dealing with low level signals. Today you have all heard a mobile phone transmission burst interfering with audio gear. The principle is the same. Now instead of those kilowatt AM transmitters we have thousands of small ones transmitting in bursts. Mobile phones, Wi-Fi, dimmers, Motor controls, Switch mode power supplies and so on. This is called EMI/RFI interference. It is an invisible contamination (electronic fog). Any electromagnetic wave translates to a wave in our virtual sea rocking the boats in it.

Then there is magnetic interference from power lines, Gaped transformers and motors and varying magnetic field inducing small currents in the low impedance loops. (hum on your MC cartridge?). Think of those as wind.

To solve this fundamental problem of all unbalanced sources we developed a system that switches the ground plane. This is equivalent to the case above with a single boat tied to the pier. Keeping an undisturbed common reference.

This approach is isolating each unbalanced device from the others preventing any current passing between them and cleaning up our reference plane.

As this problem is not new, greater minds have come up with a solution called balanced transmission and differential input. This is the standard used in all studio equipment and professional installations. Think of the last concert you visited with dozens of microphones, guitars, synthesizers, amplifiers and intercoms. All connected to the same mixing desk at great distances and no hum, no buzz!

Obvious solution I think. But it was not popular in the 50’s and 60’s so “modern” tube amplifiers copy the old designs without thinking. As we showed in the previous example even a single source can create problems by dragging around the ground plane (reference for the input) of the pre-amplifier. To prevent this from happening the second unique feature of Dionysos, namely transformer coupling of the input comes to save the day by detaching the input from the pre-amplifier ground completely. Leaving it as a floating pier in the sea.

How does the transformer do that? The input that measures against the boat deck as a reference just moves to the source boat and measures against the same reference as the output of the source eliminating any relative movement from the equation and then transmits the result via magnetic flux to the pre-amplifier input completely detaching the two. There is no common part involved in the signal transmission. Now you don’t care about any relative movement between the boats at all (ground planes). This is galled galvanic isolation and it happens due to the transformation of one type of energy into another and back.

Do not confuse XLR or balanced inputs with transformers as the electronically balanced circuits are not galvanically isolated. Every modern piece of equipment that has balanced inputs and outputs on XLR are usually electronically balanced circuits. They emulate the operation of the transformer to some extent. By not requiring a common reference and being able to seance the difference between its inputs ignoring the common “shacking and movement” of the ground plane. This is known as Common Mode Rejection Ratio and again is very dependent on implementation!! But works.

Going back to the example above this will mean the input of the pre-amplifier is now relative to the source ground plane only!! This way the effect of relative movement of the boat and pier is gone. Yes GONE! The signal will always be picked up in its entirety and relative to its own ground plane. Now you have the lowest level detail unmasked and fed directly to the gain stage.

The transformers are by nature differential and balanced, so Dionysos features fully balanced and galvanic isolated inputs. For convenience and functionality along the modern XLR inputs we retained a number of RCA inputs featuring a ground lift option making them differential when appropriate.

Now lets dive deeper. Problems are always a lot more complex and of course there is a noise source or a source of ground instability built in every unit and it is called a power supply. It can be best ilustrated with the vibration of the boat engine. If all we work with is referenced to the boat deck then we are kind of ok as the disturbance is COMMON to all inputs and outputs but again whit-out galvanic isolation this vibration will go to the other boats and their vibrations will come to ours. So we use a transformer on the output as well. Galvanically isolating all connected equipment from the Dionysos gain stage and its power supply. This leaves undisturbed the source and power amplifier ground planes accounting for the exceptional clarity and transparency characteristic of the Thrax sound.

Some will argue transformers are bad and sound colored. I will do a separate post about that. Just remember that the reference music you use to test your gear has passed trough dozens of transformers before being cut on a record or digitized for streaming. So I leave this discussion for another time.

This was the reasoning for the choice of a transformer isolated input and output on the Dionysos. Next one will be about the attenuation choice.