The Dream Machine: the Echorec 3°
by Phil Taylor
It’s interesting to consider what kind of Echorec delay machines would Binson be manufacturing if they were still in business today? This question has constantly fascinated me and I’ve spent many years researching and tinkering with these incredible machines with the ultimate aim of one day building my own souped-up version of the Echorec. For me Binson’s iconic Echorec 2° machines with all-tube electronics, pulsating green ‘magic eye‘, backlit black plexiglass fascia and hammered gold paintwork represent the pinnacle of their Echorec range. After this point Binson Echorec designs didn’t look so aesthetically pleasing and, in my opinion the transistorised electronics and additional heads did nothing to enhance the tone or usefulness. What if Dr Bini had instead taken an alternative product development path and gone on to make improvements to the tube electronics and mechanical aspects of the Echorec 2°, whilst retaining the original styling, to create a more reliable and better sounding machine, the Echorec 3° (third generation Echorec)?
This is my ultimate dream machine, an enhanced Echorec 2° – a tube/drum tube delay with several playback heads (multi-tap). The Echorec is one of the most inspiring and effects around. It makes it possible to play rhythmic patterns or chord swells, and then solo over them to come up with whole compositions without the need for the rest of the band around. Although I’m still perfecting the new Echorec 3° echo concept, in the process of designing it I’ve uncovered a great deal about the nuances and idiosyncrasies of the original Binson machines, and this has led to the creation a wealth of Echorec technical documentation and service information, which can be found on the Effectrode website for free. But there are still questions that need answering though, the big one being: just what is it that makes the Echorec sound so special and can it be replicated without the expense of the heads and Binson’s specially manufactured magnetic drum? Perhaps, although I am pretty skeptical about this being possible. Here’s what I’ve discovered so far…
Initial Musings on How to Create the Echorec 3°
The first step was to undertake an investigation of different types of delay-line technologies. The limited frequency response performance of the bucket-brigade device (BBD) integrated circuit (IC) ensured it dropped out of the race straight away. BBD technology is only good up around 3KHz, noisy and the delay time is limited to around 300ms – a ‘Super Binson’ would have to have a delay time in the order of 800 to 1000ms. Moving on to digital delay technology, there are plenty of easy to use cheap and cheerful ICs currently on the market including devices such as the Princeton Technology Corp PT2399 digital echo processor IC and the Spin Semiconductor FV-1 complete reverb solution in a single IC. However, these really are budget devices with onboard ADCs, DACs and support components, if this project was going to be digital then it would need to be done with a little more panache and power, after all, we want to do the Binson legacy justice – this Echorec 3° needs to be done properly if it’s going to be done at all.
The higher end DSP (digital signal processing) chips looked very promising with high sampling rates and massive number-crunching capability – if you were going to use a DSP then this surely had to be the way forward. However, even with all the horsepower of a DSP chip such as the Analog Devices SHARC DSP chip I wasn’t confident about quantifying and describing the audio performance (transfer function) of the Echorec as mathematical equations. Here’s why. After working with vacuum tubes since the 1990s I’ve come to realise there are many subtleties in he way an analogue tube circuit behaves. Just adding a capacitor of a few picofarads here or there or changing the value of a bias resistor by a few hundred ohms will have an obvious and often quite dramatic effect on the tonal character and response of the circuit. There are so many variables affecting the distortions, resonances, frequency response that I couldn’t even begin to quantify ‘tube sound‘ let alone reduce it to a set of equations.
Limitations of Digital Modelling
Introducing an electro-mechanical system, that is, tape heads and a magnetic drum, adds even more variables to the equation – a mind boggling number of them, which leads me on to what I consider is a fundamental limitation of modelling. Even if one assumes the DSP hardware is perfect, at the end of the day, the algorithm is only as good as the engineer’s understanding of the system he’s attempting to model. For example, let’s say an audio engineer wants to model the wow and flutter of an old Echorec machine that hasn’t been serviced in a while. What parts of the system is he actually going to attempt to quantify? Idler wheel deformation [see photo], or slippage because there’s some oil on it, mains power fluctuation affecting motor speed? All these parameters define the character of the wow and flutter on an Echorec and it will be uniquely different to wow and flutter in, say a Watkins Copicat tape delay. Our engineer might not be fully aware of all these factors or undertaken an in-depth analysis of the Echorec’s performance and perhaps decides that all he needs to do is modulate pitch sinusoidally. The result is a wow and flutter algorithm which may as well be the model of pitch de-tuning in a Copicat, an Echoplex or a Sony Walkman with stretched drive belt, for that matter. It’s certainly nothing like representative of what’s happening in a tired old Echorec.
The Binson Sound
I must admit to playing devil’s advocate to some extent with all this talk of modelling wow and flutter though. You see our engineer has incorrectly assumed that wow and flutter is something that defines the Binson sound, when in fact it’s just an artifact of a poorly maintained Echorec. The machines rolling off the Binson production line back in the 1960s were tip-top – mechanical parts, such as the magnetic drum were milled to micron tolerances, aligned and brand new – pitch fluctuation of the delay repeats was inaudible in a new Echorec. Binson were obsessive about the sound quality of their products. Surely that’s the sound we’d want to model, not the sound of some beat up old machine. Dr Bini designed and marketed the Echorec primarily as a hi-fidelity machine for use by vocalists and orchestras. Wow and flutter has nothing to do with the Binson sound any more than mushy sounding delay repeats do, which is another artifact of a poorly maintained machine. A new Echorec, where the memory system has been properly set up in the factory has a very good frequency response. For me, poor high frequency response and high wow and flutter are about as welcome in an Echorec as they would be on a Revox reel-to-reel tape recorder or a Garrard turntable.
A Better Approach?
Given my lack of faith in modelling, I wondered if there might be some other way of creating a drumless or solid-state Echorec memory system. But the memory system is the heart of the machine’s sound. What it does to the delayed signal is just beautiful, creating dimensional, ethereal repeats that swirl and degrade into a lush wall of sound and when pushed to the edge of self-oscillation the Echorec becomes a sound synthesiser as well an effects processor – a musical instrument in it’s own right. How can we replace that? One idea was to replicate what engineers call the ‘transfer function’ of the Echorec with analogue components, that is, inductors, capacitors, resistors, tubes and authentic Echorec heads and design custom digital circuitry to create the delay. The transfer function is the ratio of the output of a system to its input. So if we have an input function of X(s), and an output function Y(s), then the transfer function H(s) is given by the formula:
H(s) = Y(s) / X(s)
In the more specific case of the Echorec, the system is the drum, heads and tube signal path. Here the transfer function describes how the combination of high and low pass filter responses in the tube circuitry and the saturation distortion in the magnetic recording medium affects the tone of an instrument signal being fed into the Echorec.
I’m convinced that a properly designed solid-state memory system would get incredibly close to the authentic Echorec sound and, in fact be better than the original in several respects. The Echorec signal path would be duplicated using modern tubes, polyester capacitors and precision metal film resistors and the tube heaters powered with the D.C. and fully regulated and smoothed H.T. power would ensure hum and hiss were kept to an absolute minimum, all of which would help improve the audio quality whilst retaining the magic of the Binson sound. But the real advantage of the solid-state memory system would be zero wow and flutter (which is is what Dr Bini was aiming for) and much longer delays than the 300ms of the original Echorec – it would be capable of creating delays above that useful 600ms region and beyond and it would also be feasible to add tap-tempo too. A real dream machine.
Anyway, that’s the plan so far. I’m sure there will be much more to tell and plenty more spin-off Echorec documentation written as I continue my quest to penetrate the mysteries of these fabulous old machines…