The video above shows some testing and tuning of the AAE EQ-DU & REVERB, a valve audio processor currently being designed and built by Theo Argiriadis, with input from Thom McIntosh. One channel of two had been built for this session. The spring reverb unit can be seen sitting atop the eq/distortion unit. Drums are recordings from a 9090 analogue drum synthesiser along with 808 and 606 samples. Analogue synth is from a x0xbox. No audio processing was added in post-production.
Delve into the EQ-DU’s uniquely flexible sound shaping design in a Q & A with Theo and Thom…
Project Q & A with Theo and Thom
Thom, what were your requirements for the project?
Overall, I was after what I identify as a specifically valve sound quality. Some vintage valve gear I've used has it, and that’s what I wanted. In the studio I work mainly with electronic sound sources, such as synths, samplers, drum machines, DAW etc. Valve amplifiers, EQs and mechanical reverbs offer unique ways of processing audio that can imbue what I find to be very desirable sonic characteristics. These characteristics were clearly recognisable in examples of AAE's hardware I listened to online. Next, I was intrigued by the control AAE units give over the sound being processed, such as in the EQ design and spring reverb driver. It’s like they have qualities of being instruments in themselves, and that quality lends itself to a way of working in the studio which has always excited me.
Preparing a new case - ensuring plenty of air flow keeps internal components cool for longevity and reliability.
My basic requirements when I approached AAE were for a two channel unit I could use at any step - recording, mixing or pre-mastering. It needed to be flexible, dynamic and low noise enough to support subtle through to extreme application. For example, it might be used for severe distortion and filtering when recording, but then later across a final mix for subtle introduction of warmth, tone shaping, and ambience.
During our initial design discussions Theo's deep knowledge and practical experience in valve audio design was both fascinating and inspiring. I got more and more interested in the wide range of effect that valves are capable of having on an audio signal. We talked a lot about features that would exploit this characteristic, such as triode, pentode, on and off load operation, a combination of active and passive EQ components and so on. The final design added a continuous bias control and feedback circuit for the same reason.
In addition, I wanted the order of the Distortion and EQ modules to be swappable because in the past I've found this can make a big difference to the achievable range of tones. Finally, I wanted a filter section to sculpt the distortion output, and make it more flexible for mixing with the non-distorted signal.
As the project developed and Theo planned out the signal flow, I realised send and return points at stages along the audio path would be useful for integration of the various components into a studio environment. This enables external processing and parallel mixing of signal feeds through or from the unit as necessary.
Overall the design has become quite modular. Theo described it as something akin to an analogue audio computer, which I really like the idea of. After a couple of sessions testing and tuning the first channel it's clear there's huge scope for sound shaping in there, all built on a foundation of fantastic valve sound quality.
Exactly what we were after!
Theo, can you describe the project from a technical perspective?
This is a dual mono project that combines sections and circuits that I've designed and built in the past. It's effectively a valve based analogue sound processor - a combination of a distortion unit (DU), various frequency boosters/equalizers (EQ's), valve spring Reverb (REV), Return (RET) system plus various inserts (send/return sections) per channel.
The inputs accept a wide range of line level signal from 0.316Vrms (-10dBV old unbalanced standard) to 1.23Vrms (+4dBm balanced standard) or even higher depending upon how the input stage gain switches and level controls are set. The main inputs and outputs are balanced using XLR type sockets, along with optional unbalanced jack sockets.
Bespoke analogue - completed wiring on the underside of one channel of the EQ/DU module.
The unit is able to be adjusted to produce clean, mild or even heavily distorted sound.
The type of harmonics generated can be even and low in order, or odd if so desired, for a harsher sound. This harmonic distortion can be combined with adjustment and boosting of certain frequencies through the various EQ's, and it can be very effectively used to shape and animate the sound qualities captured in digital recordings.
On the other hand, the unit can also be used as an extreme sound effect; for instance by introducing the Positive Feedback and OFF LOAD mode features in the DU section.
Next are the various modes that the valve spring driver operates in. These can produce anything from traditional mild reverb to serious overdrive spring distortion.
Finally, the various frequency shaped signals - dry, reverberated, with or without overdrive distortion (it all depends on how each section is set) plus the original signal are mixed at the output stage and a variety of different sounds can be created.
It's important to emphasise that just like with any analogue processor, experimentation by the user is necessary!
To achieve the required sound quality and sonic flexibility within a valve based design, can you give some more detail on the design approach you took?
Let's start with the 'DU post EQ mode' which is when the input stage drives the EQ driver:
Both of these stages are triode valve designs - 12ay7 and 12au7 respectively.
The 12ay7 is a very versatile input stage valve, because of the range of signals it can accept. It has a slightly lower gain and plate resistance than the commonly used 12ax7 (UK ECC83), so it can accept higher level signals and can drive a lower resistance potentiometer (the DRIVE) control. Its input gain can also be adjusted through the gain switch by introducing local negative feedback, if the desire is to reduce the distortion on high input signals. This tube was widely used in some of the early Fender guitar amps of the "Tweed" era.
The 12au7 (UK ECC82) is the best choice to follow, due to its even lower amplification and plate resistance, accepting the high level signals coming out of the 12ay7 stage through the DRIVE control. Its low plate resistance (especially when it works in parallel mode) makes it possible to drive (and interact with) the passive EQ.
The EQ DRIVER and the EQ MIXER 12au7 tube grids contain diode/resistor networks that limit and control grid current clipping to keep the overall "tone" smooth (subsequent distortion can be supplied by the DU, of course).
The way this tube is biased is important for creating the "second harmonic sound", and for being able to benefit from its grid's ability to accept high signals before it runs into the grid current region, thus avoiding the blocking distortion effect. However, if the DRIVE control is set high enough this valve will produce grid current distortion... it's all a matter of choice.
Totally tubular - 12ay7 and 12au7 valves installed in one channel of the EQ/DU module.
Alternatively, on "low gain " mode it becomes very clean through local negative feedback again. The 12au7 can be found in both pre-amps and phase inverters in Hi-Fi power amps, making it a very versatile device too.
The gain of the 12ay7 can be adjusted by a 3-way switch, and the 12au7 stage input is controlled by the EQ DRIVE control, as well as having its own 2-way gain switch.
These control and gain switch settings are very important for the initial sound processing. For example, with a 0 dB (0.775vrms) input signal, even if the gain switch is set high, as long as the EQ DRIVE control is set somewhere around half way, the distortion will be low, mainly even and of the second harmonic type.
As a result the sound is clean but with a colouration which may also be expressed as "warmth". This is because the triodes are operating under conditions which aim to achieve a high level clean signal at the plate of the second (12au7) valve.
One of these operating conditions is the relatively high power supply voltage (400V) - another is the choice of the load that each one of these valves "see". Carefully selected for second harmonic distortion, the amount of load is proportional to the amplitude (strength) of the input signal and choice of control settings. For these reasons the two stages cannot produce very high levels of distortion.
The passive EQ section loads the EQ driver and plays a serious role in the sound here - firstly due to the loading effect and secondly it emphasises certain harmonics, reduces others and gives a certain "character" to the sound.
The EQ interacts with the stages before and after it. If you turn up the EQ DRIVE control and set both switches high to overdrive the EQ DRIVER, you can use the EQ to boost or attenuate harmonics that have been generated by this stage.
Similarly, you can make those two initial stages clean by the use of the control switches, and then use the EQ to shape the signal before it enters the following stages (ie DU, REV DRIVER etc), setting the gain(s) on these stages to overdrive one or more of them.
The high signal solid state buffers that pick up the signal from the EQ also work at 400V. In order not to affect the signal, these circuits must never reach clipping point and are very transparent.
How does a Vacuum Tube (Valve) work?
A vacuum tube, or valve is a form of electronic amplifier that has been in use for over a hundred years.
Watch this video to find out more about the inner workings of valves being discussed in this Q & A.
The Distortion Unit (DU) features two valve operating modes - Triode and Pentode - what are the differences between them?
The purpose of all the options and controls is to access the wide range of wave shaping, or distortion effects which valves have a unique ability to produce with audio signals.
At low input signals TRIODE mode produces predominantly second harmonic distortion which is more musical. Drive and gain controls must be set low to drive the DU softly if this kind of sound colouration is desired; and to avoid clipping. As the drive level is turned up, more distortion will be generated, first low even, then low odd (like third) and a smaller amount of higher orders both even and odd, until eventually soft clipping occurs.
In PENTODE mode distortion starts earlier, as far as the drive level is concerned. PENTODE output will produce more third and higher order odd distortion products, in addition to second and higher even products. By overdriving the power stage in higher drive settings and with gain switched high, harsh edgy sound and a boost in high mid and treble occurs.
Distortion and harmonics - what are they?
Distortion is the alteration of an original input soundwave by an amplifier's circuitry, and can be introduced by harmonics. Harmonics are related to the original input soundwave which is being amplified. The fundamental frequency of the input soundwave is the First order harmonic. The Second order harmonic has two oscillations within the same time period that the fundamental frequency has one oscillation, and the Third order harmonic has three oscillations. Fourth, Fifth and higher order harmonics continue this pattern. Even numbered harmonics tend toward a triangle wave shape, which is similar to the fundamental wave shape. Odd harmonics tend toward a square wave shape, which is significantly different to the fundamental. This is why odd harmonics tend towards a more 'strident' or 'sharper' sounding overtone than even harmonics, which tend towards 'sweeter' and 'warmer'.
Triode and Pentode modes can also be switched between Low and High loading operation. What effect does this have?
The loading affects the behaviour of the valves and associated circuitry, giving access to a greater variety of tonal colouration through shaping of the audio signal. It’s important to note that the 'load - mode' rotary switch is simultaneously switching parts of circuits on both primary and secondary coils of the DU, so it affects them both.
The terms Low and High relate to the load that the DU output "sees", or is subject to.
In TRIODE mode, 'low load' is essential for a mild and low distortion tone, especially if the DRIVE control is also set low.
In PENTODE 'low load' mode, the valves' screen voltage is also switched to a lower value and as a result greater distortion, and more odd harmonics are produced.
'High load' in TRIODE mode creates greater even harmonic distortion.
'High load' in PENTODE mode generates less distortion (compared to 'low load') in this particular circuit. For low level drive signals the second order dominates, followed by the third. As the drive signal increases, so does the output (plate) signal, and it will contain higher orders of odd distortion and some even.
In this DU design I've chosen the high loading to be near the 'critical load' value. This was the norm that audio designers tried to achieve in the 1950s and 60s when they were designing valve circuits for anything from table radios, to low power guitar amps and Hi-Fi's. The term 'critical load' refers to the load value that audio engineers derived from testing and extensive mathematical and graphic analysis.
Pentodes are more complicated than Triodes due to their extra electrode, known as the 'screen', and the voltage of this screen is also related to the critical load value.
So why use the critical load value? Pentodes are far more power efficient at critical load, so this choice results in a good compromise between low distortion and power. When this approach is combined with NFB (negative feedback) it reduces the residual distortion which increases at higher output signal levels, and it reduces output transformer core saturation distortion. However, low distortion was not of primary concern in guitar amps (and small PA's) so these employed very little NFB, like the Fender Champ with its classic sound.
In this DU design, power is not an issue because it's not being used to drive a speaker. But, if you want to emulate this sound in one of the DU modes then you go down the 'critical load' route. NFB is not applied because NFB will result in a less characterful sound, subjectively it "deadens" the sound. If you want lower distortion you can simply reduce the DRIVE and increase the OUTPUT LEVEL.
So to summarise
TRIODE LOW = low distortion
TRIODE HIGH = more distortion primarily second on low signals
PENTODE LOW = high distortion, plenty of third and odd products at higher signals
PENTODE HIGH = less distortion mainly second with a bit of third, products increase at higher signal levels due to the absence of NFB.
Loading and resistance - how are they related?
High load refers to the higher amount of current flowing from the output amplifier of the distortion unit to its internal load. When set to 'High load', this internal load has a lower resistance. Lower resistance in a circuit enables a higher current to flow through it. At the extreme, an infinitely low resistance (nearly zero Ohms) would result in a very high current flowing through it, effectively being a short circuit. Low load is the opposite - a lower current resulting from a higher resistance. An infinitely high resistance (i.e. an open circuit) results in zero loading (see OFF LOAD switch).
Does the DU create any forms of distortion in addition to what we've covered so far?
Yes, it does. In addition to the ordinary 'triode or pentode distortion' we've just been discussing, there is what we call 'grid distortion', This is also where the BIAS control plays a significant role.
Grid current distortion/clipping is a very important phenomenon in all types of tube overdrive sound, especially in AAE DU's. It only applies to vacuum tube circuits and there has been a lot written about it online in recent years.
Distortion vs Clipping - what's the difference?
Distortion is the alteration of the original input soundwave shape by an amplifier's circuitry. Clipping is a type of distortion. Clipping occurs specifically when the soundwave shape is altered because its maximum amplitude is constricted by an amplifier's circuitry.
The PCL86 valve's power section can be switched into triode mode through the MODE switch, but it is worth noting that historically this section was typically used as a pentode.
Any pentode can be converted into a triode by connecting its screen grid to its plate, and that is exactly what the MODE switch does.
In a valve, the first (and in triodes, the only) element that controls the flow of electrons is the 'control grid'. The DC voltage on this grid is negative with respect to the cathode, but the BIAS control makes the current flowing through this tube adjustable in this DU.
Under normal operating conditions this grid draws no current and its input impedance is infinite. (In "normal" HI-FI and clean sounding studio equipment the grids of the tubes used are not meant to enter their grid current region, and so long as this holds true their grid current region can be ignored.)
For this reason, I use the term "ordinary triode or pentode distortion" to refer to the type of distortion that a triode or pentode produces had the grid current distortion never existed.
I also include "cut off distortion" under the term ordinary distortion. Cut off clipping can be very soft at low bias settings, and it flattens the top of the waveform gradually, a distinct tube phenomenon.
Cut off clipping or distortion occurs because the plate (tube electrode) cannot go higher than a voltage value related to the DC power supply voltage, and it is caused by a high negative going signal in the control grid. Therefore it is the opposite of grid current clipping.
Grid current clipping is caused by the drastic changes that occur when the DU drive signal reaches a value high enough so that its positive going cycle shifts the grid from its negative bias (with respect to the cathode) to zero volts and beyond into the positive grid region. The negative grid bias value prior to the high positive cycle is set by the BIAS control.
The grid now acts as an anode (know as plate in the USA), that means it can also pull electrons i.e. it starts conducting very quickly, and the incoming signal "sees" a diode made out of the grid to cathode "junction".
In plain English, hard clipping occurs during this high positive cycle similar to a "diode clipper" in a fuzz-box type of circuit.
Control grid distortion due to clipping contains a large amount of high order harmonics due to the sharp transition from a smooth curve (the positive going sine cycle) to the "flat" clipping top.
The two videos below investigate the DU 'ordinary' and 'grid current' distortions, and how loading and bias create different types of distortion.
Note: In these videos the phase of the waveform is reversed by the unit's circuitry, so the positive or top cycle we are referring to in the page text appears as the negative or bottom cycle in the videos, and vice versa.
So the BIAS control plays a big part in the way distortion is introduced?
Yes, for example if you want to generate as little distortion as possible, or just mild coloration, you must use triode mode low loading (triode low distortion). Also, you must set the BIAS control as high as possible because this is where (in high plate currents) the triode behaves in its most linear manner.
Then the sound will be as clean as possible, until grid clipping occurs and that clipping is hard, especially at high bias control settings. So if you want to keep the sound clean you must be aware of the grid current issue, and turn the DRIVE control down as soon as it appears.
Both ordinary triode/pentode distortion and grid current distortion can cause the waveform to become asymmetric, and waveform asymmetry will produce second and even harmonic products.
The BIAS control gives great flexibility in altering the waveform. When the bias control is turned anticlockwise the "ordinary" distortion dominates, and the grid distortion comes later. The bias now is higher, the grid more negative, and the triode produces more even harmonic distortion (ordinary type).
Another way to avoid grid distortion is to select TRIODE HIGH loading where the cut off distortion (which is softer) occurs first. However, take into account that in this mode second harmonic distortion will be greater even at lower signals levels, well before the grid distortion effect takes place. In other words, the sound will be a bit "dirtier" at average signal levels but without the sudden harshness of grid distortion dominating high level signals.
Of course you can play with the BIAS and the DRIVE controls to alter the signal in any way possible and get anything from completely asymmetrical to symmetrical clipping, where the top cycle clip is due to grid distortion and the bottom cycle clip is due to "ordinary" triode distortion.
OK, we've talked about High and Low loading, but what about the Off Load switch?
Right, from our earlier discussion we can see the Distortion Unit (DU) output transformer has an important role in the signature of the overall tone being produced. It is possible to design the DU without a transformer, for example by getting the Input stage to overdrive the EQ driver; however this won't emulate the power overdrive of a single ended amp in triode/pentode mode (low, high and off load) and also the transformer inductive tone quality will be missing. Remember valve audio amplifiers need transformers to drive speakers, and similarly, vintage valve audio outboard such as EQs, compressors, etc. needed transformers to drive their outputs. These transformers created a characteristic "tone".
OFF LOAD mode is related to the same concepts found in High and Low Load operation discussed previously (and in the sidebar 'Loading and resistance - how are they related?').
To explain further, the output transformer has a 'primary coil' and a 'secondary coil'. The term 'side' refers to the chain of circuits that are connected to or driven by the primary or secondary coils of the transformer. Without getting into technical details, the electromagnetic link between these two coils of the transformer is what makes a transformer function the way it does.
In OFF LOAD mode the load that is connected to the secondary coil of the transformer is removed (reduced to zero) by disconnecting the load resistors through the OFF LOAD switch.
This significantly changes both the signal and the DC current flow through the primary coil which is an inductor, and in turn alters the characteristics of the amplified sound wave, introducing a specific type of distortion that is extreme, harsh and shrill.
The effect is strongest in Pentode mode, and more subtle in Triode mode. The BIAS control has a significant influence, so experimenting with that and the DRIVE control is necessary to find what settings sound best.
In typical valve circuits, removing the load on the secondary coil of the transformer will be catastrophic, causing high voltage spikes that damage the valve and transformer. However, in this AAE design, DU supply voltages are reduced by high wattage resistors in the OFF LOAD mode so they pose no risk to the circuitry and components.
This is an AAE invention so to speak, as it has never been done in audio before. However back in the day of TV tube engineering, they used a similar method to generate the vertical and horizontal fly-backs. The inductor in that case was the vertical scan coil and horizontal output transformer/scan coil
The video below investigates DU Off Load distortion, experimenting with Bias, Drive and Bass EQ settings
To be continued…