High Quality Audio Mixer - Stage 1

Rod Elliott (ESP)

Introduction This is part one of probably the most ambitious project so far, in that it can be huge (there is no real reason that it could not be built up as a 36-8-2 (i.e. 36 input channels, 4 "sub" stereo output channels and a main stereo output, not including auxiliary sends). The picture on the left is a representation of how a single mic / line channel might look, with the sub-master (or group) switching included. There are modules planned for every input type imaginable, including:

Standard mic / line module with 48V phantom power (electronically balanced or using transformers) Stereo phono module with RIAA equalisation Stereo Line module Talkback mic / headphone amp module

All input modules have 3-band EQ, with the mid frequency variable from 500Hz to 2kHz to allow maximum flexibility. Inputs all have variable gain, and each input module has a peak level LED, PFL (Pre-Fade Listen) button, channel insert and auxiliary sends. Optionally (for a four output mixer), there is a simple selector switch to select the master bus (A or B), and all channels have a pan control for stereo positioning. Stereo input modules have a balance control instead. The maximum gain of the mic / line unit has been set at 46dB, which is more than enough for most music recording. This allows for microphone levels down to 5mV, which is almost always exceeded by most musical instruments with a typical low impedance microphone. In many cases (especially with vocalists), the mic output level can easily reach 250mV, and I have measured the output of a low impedance mic at about 1 Volt with loud singers! Where extra gain is needed, this is easily accommodated. Each of the input preamplifier units can easily be changed to give more gain, but noise must be considered the greatest enemy of any recording or live music mixer. As gain is increased, noise increases as well. The nominal operating level of all modules is 0dBv (1 Volt RMS), and the peak LED will operate at +6dBv. This allows 15dB headroom, which is more than adequate provided the mixer is not operated with all the peak LEDs continuously on! Please note that although presented in project form, this is not necessarily a "real" project. Rather, it is a gathering of ideas with a common theme, and

circuits for audio mixing consoles are few and far between on the web from what I have seen. The Project Continues . . . The output module is also configurable, allowing a selection of options. As can be imagined, this project will take some time to complete, so will be presented in stages. This is the first in a series, and provides all the information for building the microphone / line input module. Following is the master module, and this way you can start to build a complete unit, and add the other modules later. The next is the power supplies, which has much higher current capacity than one for an audio preamp and includes the optional 48 Volt phantom feed supply. This does not require particularly high current, but needs to be very well smoothed, to prevent hum from being introduced into the low level microphone inputs. The phantom feed supply is designed to handle a maximum of about 10 microphones (or direct injection boxes) at any one time. Block Diagram of Mic/Line Stage Figure 1 shows the block diagram of the mic / line input stage with a single stereo output. This is the option most likely to be built, as it is simple to use and is relatively cheap - but only compared to more esoteric options. This is not a cheap project to build, but will provide a standard of performance that is very hard to beat with equivalent commercial offerings. Although I have designed this mixer with two auxiliary sends, there is no real reason that more could not be provided. Even with the two pots, a switch can be used to select one of several buses for each control. Generally, the ability to select either of two buses for each auxiliary send is enough, but more can be used if desired. I will leave it up to the individual constructor to determine the ideal combination. Remember that for each bus, there must be a mixing module (these can be as simple or as complex as you like), and the necessary output connectors for each. This all starts to add up (rather quickly, too), so you do need to consider the final cost.

Figure 1 - Mic / Line Module Block Diagram

Each mic/line channel has 7 pots, up to 5 switches, one push-button and a slide pot for the fader. In addition, there is a Cannon XLR input connector, a stereo jack for channel insert and the overload LED. Naturally, there is also the electronics to make it all work. An ambitious project indeed, but one that I hope will be popular despite all of this. The mic/line input module is the most important part of the system, since it is this unit that determines the functionality of the entire mixer. Accordingly, this provides input selection, gain control, phantom power, phase reversal, tone controls, auxiliary sends, pre-fade listen, pan-pot and overload indication. (If you say that really quickly, it still sounds like a lot of stuff!) Mic / Line Input Section This is common to all of the following amplifier units, and provides the following functionality:

Standard Cannon XLR female input connector Phantom power switching (optional 48V feed for powered microphones and DI boxes) 20dB pad for high level line inputs Phase reversal switch for correcting out of phase microphone or line inputs DC blocking capacitors (can be omitted if phantom powering is not used) Protection zener diodes (can be omitted if phantom powering is not used, but I recommend they be used anyway)

Figure 2 - Mic / Line Inputs And Switching

The 48 Volt phantom feed can be omitted if you are quite sure that you will never need it, but it is needed if powered microphones are contemplated, and can also be used to power direct injection (DI) boxes so that batteries do not have to be used. The 20dB pad is needed if high level line inputs are going to be used. All of the amplifier options have a minimum gain of 3 (close enough to 10dB), so if a line input of 0dBv (i.e. 1 Volt) is applied the gain control will be set near its minimum gain setting. For higher levels, the pad reduces the input signal, allowing up to +20dBv (10V RMS) to be applied without overload. It has always been something of a convention to use a 20dB pad, but in my experience these are a bit of a pain. The pad tends to reduce (or increase) the signal by just that little bit too much, so I thought briefly about a 10dB pad instead. Feel free to modify the circuit if you prefer a 10dB pad. The phase reversal switch is used where two microphones are used in close proximity, and cause phase cancellation because of their relative distance from the source. This can result in a "hollow" and often unpleasant sound especially when miking a drum kit or piano. The phase switch allows you to select the best setting to get the sound you want. The capacitors and zener diodes protect the following amplifier from transients when a mic is plugged in or removed with phantom power applied. These are recommended even if the phantom feed option is not used, especially with the electronically balanced circuits. They are not essential with the transformer input, but will do no harm.
Click on the PCB image to see Project 96 - a different version of the phantom feed and distribution (and a power supply) - PCBs are available.

Mic / Line Input Module (Transformer Input)

There are three options for the input, using a transformer to balance the input, or using either of two electronic balancing circuits. My preference is (and always has been) for the transformer, as the common mode performance is far better than the electronic method, and also provides much better radio frequency interference suppression. However, transformers are expensive, so both methods are shown, but the noise performance will suffer and interference will also be worse if the transformer is not used. In reality, these differences may not be an issue, and the use of a cheap ($20 - $50) transformer may (will?) degrade performance far more than the use of an electronically balanced circuit.

Figure 3 - Transformer Input Mic/Line Input Circuit

The suggested transformer is a Jensen JT-16-A. This is a 1:2 step up transformer, which provides a useful gain of 6dB, and more closely matches the impedance to obtain the optimum noise figure from the input amplifier (an NE5534A single opamp). It is more than probable that most constructors will be stopped by the transformer (it stopped me, so I did most of my preliminary testing using a 1:1 transformer that I had to hand), but if you can get them this is the best option. Be warned, good transformers are expensive, so if you can't afford good ones use the next circuit instead. Cheap transformers will degrade the sound to an unacceptable degree, and should be avoided. Also note that as shown, this circuit has a maximum gain of 40dB. To achieve the 46dB gain mentioned above, change the 50k pot to 100k. Mic / Line Input Module (Electronically Balanced Input - Version 1) The electronically balanced input stage needs some fairly radical protection from the switching transients produced when the 48 Volt phantom supply is turned on or off. This is most easily accomplished using the zener diodes as shown in Figure 2. The input circuit is a modified version of what is commonly called an "instrumentation" amplifier, and provides far better impedance balance than the simpler version commonly used. The impedance balance is very important in this application, because if the impedances are not equal for the inverting and non-inverting inputs, the noise rejection suffers badly.

The gain is set using the 10k pot VR1. The capacitor C4 helps prevent the circuit from oscillating (these opamps have a very wide bandwidth), and also act as an RF stopper, along with the two 1k resistors. This combination gives a worst case upper -3dB frequency of about 25kHz, and from tests I conducted is extremely effective.

Figure 4 - Electronically Balanced Mic / Line Input Module

I would rather have liked to have used a "bootstrapped" input circuit, but it is patented (by someone else), so was not used in this design. The alternative shown is only marginally worse than the bootstrapped design, and should give a fairly good account of itself, even under adverse noise conditions. There is one major benefit of the arrangement shown, in that it behaves just like a transformer for unbalanced inputs (well, almost exactly). If the source is connected to only one input, the output voltage is negligible, and it requires that the unused input connection is grounded. Apart from anything else, this indicates that the external noise contribution is far lower than would be achieved with the simple balanced input circuit (U3) alone. The input stage shown is variable gain, without any fuss or difficulty. This circuit has much better impedance balance than the circuits you normally see, because the centre tap of the two input resistors is not connected directly to ground. This allows the input circuit to "float" above ground, and improves common mode noise rejection for inputs that are not perfectly balanced (which is most of them, due to imperfect leads etc.). The balancing shown is so good that it is necessary to disconnect the phantom supply completely, or it will degrade the performance for noise rejection quite badly. The same applies to the transformer input, by the way. In tests I conducted, the unbalanced signal rejection was better than 30dB. This is not magnificent, but it is much better than the 6dB from the conventional single opamp version. Maintaining a good signal to noise ratio is very hard without the transformer, but using either the NE5534A or the dual version (NE5532 or [and IMO only if

you really have no choice] LM833) is a good start. With an input noise figure of less than 5 nV / (see Noise Figure and Other Stuff) for an explanation), these are probably the best choice for a reasonable price, and they are readily available - this is a bonus, as I hate to use devices that are hard to get. There are (allegedly) better opamps, but they may not easy to get or may be inappropriate for high quality audio, and will be more expensive as well. The NE5534 is a very good opamp, and is always a safe bet. Make sure that you use the NE5532A for the input amps, as these have a better noise figure than the standard version. Mic / Line Input Module (Electronically Balanced Input - Version 2) This version is very similar to the one above (although it might not look like it), but uses an Analog Devices SSM2017 microphone input amplifier. It is electrically almost identical to the "version 1" amplifier, but uses a single device. It is a far cheaper option if you can get the SSM2017 devices, but I figured that most constructors will have trouble finding them (hence version 1). Since this article was originally published, the SSM2017 has been declared "obsolete", and is no longer available. Texas Instruments make a pin compatible replacement, called the INA217, and its performance is said to be as good or better than the original.

Figure 5 - Electronically Balanced Mic / Line Input Module (Alternate Circuit)

As you can see, this is a very simple circuit indeed. I cannot vouch for its performance, since I have been unable to get the SSM2017, so the circuit is basically directly from the manufacturer's data sheet. The only modification is the input grounding, which is the same as shown for version 1. The SSM2017 has been around for many years, so I must conclude that it is probably fairly good. Noise performance is something of an unknown, because of the rather obscure way it has been specified in the data sheet. It is alleged that the noise figure is 950pV / (yes, that is pico-Volts), but this is only for a gain of 1000 (60dB). At lower gains, the noise figure climbs, and is 1.95nV / at a gain of 40dB and 11.83nV / at a gain of 20dB.

Regardless, it can be assumed that the noise is very low, but I have also read reviews of products using it that claim it is "ordinary". Exactly whether this is good or bad is unclear, but I would expect that you will not be disappointed if you can get hold of the ICs.
Click on the PCB image to see Project 66 - yet another version of the mic/ line preamp - this one has a PCB available.

Tone Control Module The tone controls are unusual in this design, because I wanted to have something a little more flexible than the standard 3-band EQ commonly used. As a result, there are two "gyrators" or simulated inductors, and one of these is made variable to allow the midrange control to be swept from 500Hz to 2000Hz. This is expected to cover the range where most nuisance peaks and dips will be found, and will make accurate equalisation far easier than is the case with a fixed control. Bass and treble are conventional fixed frequencies, but as can be seen are also connected unconventionally. The channel insert jack allows a signal to be routed via any external device compressor, graphic or parametric equaliser, or any of the multitude of effects that are now available. This is post EQ (after the tone controls), which is good in some respects (the external unit's noise is not increased by applying treble boost, for example), but is a problem if the channel insert is used to inject a signal into the mixer. This is not the way any signal should be sent to the mixer, and I prefer it the way I have designed the circuit (but I suppose I would).

Figure 6 - Tone Control Module

The mid frequency control can be modified to change the range, so by multiplying the value of C3 by 4, the frequencies are halved (so it will have a range of 250Hz to 1kHz) and dividing by 4 will double the frequency. Likewise, the bass and treble controls can also be modified to change the turnover points from 300Hz and 2.7kHz respectively. With the values as shown (the mid frequency is set to 1kHz), the tone control characteristics are shown in Figure 6a. As can be seen, there is plenty of variation, and bass and treble controls are deliberately moved away from the centre frequency band. Conventional tone controls tend to be centred around 1kHz, but the idea of providing bass or treble boost and cut from this frequency has always seemed a trifle ridiculous to me.

Figure 6a - Tone Control Response Curves

The +/- 3dB frequencies for the bass control are about 300Hz, and 2.7kHz for the treble. The midrange control is variable, and its Q ("Quality factor") increases as the maximum boost or cut is approached. At low settings, the Q is quite low, so the control is "self adjusting" to some degree. High Q values are normally not needed at moderate levels of boost or cut, but if there is a real problem frequency (the rim of some snare drums springs to mind), it can be notched out very effectively to get that "fat" sound without the hollow ring. I have not included high or low pass filters in this design, but they can easily be added. Virtually any simple opamp filter can be used, or even a passive first order (6dB / octave) could be included if you wanted to. Fader And Auxiliary Sends The fader is connected to the "tip" connection of the insert jack, so a signal can be directly inserted. When not in use, the switching jack connects the output from the tone controls directly to the fader. It is important to use the best quality faders that you can afford (or can find - they are pretty thin on the ground, I'm afraid). The Aux 1 send can be switched pre or post-fade, while Aux 2 is post fade only. An additional switch can be used to allow pre and post fade for this as well, if desired. The PFL (Pre-Fade Listen) push button is designed to allow you to listen to the signal, even if the fader is fully off, and over-rides the main headphone signal. This is done in the master module, and the PFL switch is connected to a bus.

Figure 7 - Fader, Channel Insert and Aux Sends

When more than one PFL switch is pressed, there is still a minimal drop in level, but at less than 0.1dB this will not be audible. To remove the problem altogether, use a buffer stage before the PFL button. Optionally, the main output can be switched to one of a number of stereo buses, allowing the mixer to be used with group masters, all feeding the main stereo send. This is commonly used to allow sub-master control for a group of microphones, such as drums, back-up vocals, horn or string sections, etc. Although this complicates the construction of the mixer (yes, even more), it allows an entire section to be raised or lowered in the mix with a single fader. Without this, if for example, the percussion section needed to be a little louder for one song (or one section), it would be necessary to operate perhaps four or more channel faders simultaneously. This can be done, but it is far easier if the sub-masters are used. A single fader adjusts the entire section, maintaining the balance of the mix exactly as it was originally set up. See Figure 9 for the details. The mic / line module can also be used as a recording mixer, by taking a send to the tape machine from the channel insert or an extra connector could be used. If you use the channel insert, use stereo plugs with tip and ring shorted (so the mix is available in the studio monitors). For recording use, it would be a good idea to have an additional (switchable) input for tape playback, wired

in parallel with the XLR. If balanced outputs are required for the recorder, then you will have to wait for the master module, since this has a floating balanced output circuit. Peak Detector The peak detector is the best method of ensuring that all signals are well below clipping level. In this design, the detector operates when a signal is greater than 2V RMS, and is not polarity sensitive so it will indicate when either a positive or negative peak exceeds the threshold. The dual opamp is a very basic (and cheap) 1458 type, as it is more than adequate for this application.

Figure 8 - Peak Level Detector

The peak detector can be a simpler affair than that shown, but most of the simple ones only sense one polarity. This is not suitable in my opinion, because audio signals can be extremely asymmetrical, so one side can be clipping and you would not know it without listening carefully. Note that the circuit is designed to ensure that no LED switching currents flow to the signal earth. These currents are "dirty", in that they contain fast switching times and the resulting transients. It is important to ensure that the earth connection used is separate from the signal earth, and must not run in parallel with the mixing buses or analogue earth (ground). Note that the peak detector is after the tone controls. I have seen many circuits where the detector is before the EQ, and it is entirely possible to have a signal that just flashes the LED, but goes into clipping when EQ is applied. This arrangement will hopefully help the user to avoid any such problem. Although C3 is shown as 10uF, it can be as small as 100nF. Using a smaller cap allows shorter transients to be captured, but also reduces the display time. I suggest that you experiment with the value to find something that suits you. My personal choice would be to use 1uF.

The detection threshold is set by the resistor string R1, R2 and R3. As shown, it is 2.75V peak (close enough to 2V RMS). Increase the threshold by increasing R2, and vice versa. The maximum I recommend is to increase R2 to 22k, giving a threshold voltage of 5V - equivalent to +16dBm. The closer you get to the clipping threshold (+22dBm), the greater the risk of momentary overloads causing clipping. Output Switching For Sub-Masters The use of groups or sub-masters makes a live (i.e. stage) mix much easier. It is generally not needed for studio work, but can still be very useful for the mixdown.

Figure 9 - Switching for Sub-Master Groups

A method of switching to multiple buses is shown in Figure 9, and uses a dual-gang rotary switch to select the required bus. As shown, you would be able to select sub-masters 1 to 4, the master mixing bus or Off, giving six positions in all. A 5 position switch would leave out the "Off" option - this is the version shown on the panel artwork above. This (or an expanded version) provides the maximum possible flexibility for the final mix. Still To Come In following articles, I will show the master module, power supplies and additional input modules. This is by far the most complex single module, so it gets easier after this one.

High Quality Audio Mixer - Stage 2

Rod Elliott (ESP)

Introduction In this article, the Mixing Modules, Line Output stages and the start of the power supplies are described. There are two main types of mixing module - a stereo unit to accept the outputs of the Mic/Line modules, and a mono version for the Auxiliary sends that includes an Aux return as well (these are typically used for effects, or as a simple foldback mix). The number of modules depends entirely on the final configuration you are aiming for. A typical unit might have four stereo sub-groups, one master mixer and two auxiliary send mixers. The master modules do not have tone controls, as it is anticipated that the final mix will be sent to an outboard graphic equaliser or parametric equaliser for room acoustic tuning. Simple tone controls in the master sections are worse than useless. Description A block diagram of a "typical" configuration is shown below. This is based on my original premise of four sub-groups, two auxiliary sends, and a master. All sub-groups outputs are re-mixed to the master mix bus, and optionally can be connected to balanced line output stages as well. The aux sends are not shown, but work in much the same fashion (except they are mono, not stereo).

Figure 1 - "Typical" Configuration

As shown, channels 1 & 2 are switched to Mix Bus 1, channel 3 to bus 2, channels 4 & 5 to bus 4, and channel "n" to the master bus. The outputs of each sub-master (or group) all connect to the master bus. All buses are stereo, but this is not shown for clarity. This is to demonstrate the flexibility of

using the multiple sub master groups, a 6 channel mixer with 4 groups and a master would not be worth the effort, but when you have a 36 channel unit this will change ! Mixing Modules The mixer modules are "virtual earth" types, meaning that the actual mixing bus carries signal current, but that little or no voltage is measurable. This is the most common type of mixer, as it means that there is no interaction between the various input level controls (the faders). The input capacitors should be high quality tantalum types, as these are capable of operating safely with no bias (or a slight reverse bias). The Gain control is used to trim the Master Module gain in the same way as the input modules. This allows the operator to get all faders in a position where gain can be increased or decreased by the desired amount, and returned to a known starting position. A setting of -6dB on all faders is a good starting point, and this allows the operator to be able to change the level of any channel or group, and set the level back to the "standard" starting position again. An output is provided for peak detectors (the circuit for these is in Part 1) their use is recommended, but not mandatory. Even better is to use good quality meters (to be described in a later article) for all Sub-Groups and Masters.

Figure 2 - Stereo Mixing Module (Groups & Master)

You will also need mixers for the auxiliary sends from each channel. Maybe you will decide to have more than 2, but in any case you will require as many auxiliary mixers as you have sends. These are virtually identical to the stereo version, but of course are mono. The circuit is shown in Figure 3, and is a dual mixing module.

Figure 3 - Mono (Auxiliary) Mixing Modules

Line Output Module The line output stages (like the mic / line inputs) can be either transformer of electronically balanced. Both types are described so you can choose the one you prefer (or can afford - good transformers are expensive). For recording work, the line output stages can also be connected to as many of the mic/line input modules as needed to suit your recorder. Typically, line out modules would be connected to the output of the tone control section for each channel. The fader is used only for mixdown, and to set the correct mixing level during recording. Some engineers prefer to record flat (no equalisation) to ensure that nothing is lost in the original. Adding a switch to connect the line out pre- or post-equalisation allows for all possibilities.

Figure 4 - Active Line Output Module

The signal is reduced by 6dB as it is applied to the line out module, otherwise the level would be double that expected due to the balanced circuit. The transformer version (below) is unity gain, since transformers do not have this problem - they have others instead.

Figure 5 - Transformer Line Output Module

There are a great many different circuits for balanced line drivers, but the simple approach is often the best. The simplest is to use a transformer, but unfortunately, if you want good quality you will pay for it.
Click on the PCB image to look at Project 87 - Balanced Line Driver & Receiver PCBs are available.

Power Supply There are two sections to the power supply, the main +/-15 Volt supplies and the 48V phantom feed supply. The main supplies require considerable

current capacity, due to the large amount of electronics involved, so conventional 3-terminal regulators will not work unassisted. The 48V phantom supply needs to be as quiet as possible, as any noise will appear at microphone inputs. In theory, the noise is simply cancelled because it is common mode, but I believe that a quiet power supply is worth the effort.

Figure 6 - Power Supply Transformers and Filters

The transformer and bridge rectifiers are shown in Figure 6 for the two supplies. The size of the transformer for the phantom power supply depends on the number of phantom powered devices you want to operate. Since each will take a current of (typically) 7.0mA or 14mA worst case, we need to design for the worst. If 10 phantom powered devices were to be used at once, we will need a supply capable of 140mA at 48V. Before regulation, we will need about 60V minimum, and I suggest a 20VA transformer which will be more than enough. For the +/- 15V supplies, I have used an estimate of 100mA per module (this is a bit of overkill, but it is far better to have it and not need it than to need it and not have it). Allowing for a 36 into 4 into 1 mixer with 4 Aux sends and 20 line outputs, this means a total current of about 4A. To be on the safe side, a supply capable of 5A is the design goal, so using a 20-0-20 transformer (40V centre tapped) at 1.4 times the DC current means that we will need a 150VA transformer as a minimum. (In case you missed it, the AC current into a capacitor input filter is 1.4 times the DC current for a full wave bridge rectifier.) The minimum power supply components are listed below, feel free to increase the ratings on transformers and capacitors, and adjust the fuse rating accordingly.

Transformer - 50V secondary, 20VA Transformer - 40V centre tapped, 150VA Electrolytic Capacitors, 4 x 4700uF 50V Bridge Rectifier - 5A 200V Bridge Rectifier - 25A 200V

Mains Fuse - 1A slow blow

I stronly recommend that the power supply be built in a separate box, so that no transformer hum is injected into the circuitry. The virtual earth mixing buses are particularly succeptible to magnetic flux leakage from transformers, and a separate supply will eliminate this problem, and helps to keep the weight down as well. The power supplies will require generous heatsinks, so a small enclosure is not an option. Still To Come

Power Supply regulators Monitoring amplifiers Pre Fade Listen circuitry Meter bridge General layout of the complete unit. Phono and auxiliary input modules

Elliott Sound Products High Quality Audio Mixer - Stage 3

Rod Elliott (ESP)

Project 30c

Introduction Stage 3 shows the power supplies for the mixer. The supply needs to be quite substantial, due to the high current drain of a complete 36-4-2 mixer. I will have to leave it up to individual constructors to determine if they really need this much power, or can survive on a lesser version, but if the full power supply is built, it will allow for expansion later. The phantom supply is designed to provide 150mA at 48V, which should be more than enough for any application. Also included is the headphone power amp (two needed for stereo), with the mixing and selection coming in the next section. 15 Volt Power Supplies The mixer requires +/- 15V at up to 4A for a fully configured system. Since this is well outside the capability of a 3 terminal regulator, I have used booster transistors (a quite common thing to do) to increase the available current.

The regulation is completely controlled by the 3-terminal regulator, and the transistor simply boosts the current. Dissipation in the transistor will be quite high, so good heatsinking is essential. Using a 20V (AC) transformer gives about 28V before regulation, allowing plenty of margin for low supply conditions, but increases the transistor dissipation. At 4 Amps output, worst case transistor power will be over 40W, so the heatsink will need to be rated at less than 1C /W to ensure that the transistors remain well within their safe operating area.

Figure 1 - Transformers and Rectifiers

The transformers, rectifiers and capacitor banks in Figure 6 are also shown in the previous section, and the circuit is included here for completeness. As indicated in the last section, T1 needs to be at least 150VA, and T2 can be a smaller (and cheaper) 20VA. The bridge rectifiers need to be more substantial than you might think, because of the sustained current. I suggest that a 25A bridge is used for the +/- 15V supplies, and a 5A bridge for the 48V supply.

Figure 2 - 15V Regulators

Figure 2 shows the regulators for the +/- 15V supplies. DC from the rectifier is supplied to the 3 terminal regulator IC via a 33 Ohm resistor. When the current exceeds about 20mA, the power transistor will turn on, and the IC will ensure that the DC output is kept exactly to the specified voltage. This ensures that the regulator IC is operating at a low power (requiring only a small PCB mount heatsink), and will remain cool. Typically at an output current of 4A, the transistors will require about 200mA of base current, which must be passed by the regulator. The maximum regulator power is therefore about 2.6W - a 10C/W heatsink is therefore sufficient (but only just - use a bigger one if possible). Make sure that each regulator is on its own heatsink, and uses no insulating washer for maximum heat transfer. Use heatsink compound between regulator and heatsink. Do not mount the regulators on the main heatsinks - these will operate at a higher temperature than the small individual heatsinks. The power transistors will be operating at a sustained high power level, and must have substantial heatsinks. Dissipation will be in the order of 25W each at 4A, which means that the temperature must be kept below 100 degrees. Using the arrangement shown with parallel transistors, this is easily achieved, and will require a heatsink with 1.0 degree C/W for each transistor pair. By this means, a smaller heatsink may be able to be used than would otherwise be the case, since the dissipation of each transistor is reduced and the

junction to heatsink thermal resistance is also reduced. The heatsink will still run at quite a high temperature, and should be mounted where it cannot be touched, but still has good airflow. I recommend either the biggest heatsink you can accommodate, and/or use a fan to assist cooling. The transistors shown are the minimum specification you can use - preferably, use higher power transistors. The diodes (1N4001 or equivalent) around the circuit ensure that disconnection of the DC input will not damage the regulator. Note that the +ve and -ve regulators have different pinouts - do not get them mixed up, or they will be damaged or destroyed when power is applied. Please check the datasheet to make sure that you get the IC connections right! Note also that there is no easy way to provide short circuit protection to this configuration, and care is needed to make certain that a short cannot occur. For this reason, I suggest that the transformers, rectifiers and capacitors are housed in a separate case (typically floor mounting), and that the DC output be fused as shown in Figure 1. The regulators can be housed in the main mixer, keeping all input wiring well away from mix buses. The DC input will have ripple at high current, and if allowed near a bus will inject hum into the system. The 100uF capacitors should keep the circuit stable at all operating levels, and can be bypassed with 100nF caps if desired. 48 Volt Phantom Supply Although high voltage 3 terminal regulators are available, they are not readily available to most constructors. I have therefore elected to use a discrete design, which although quite simple will give very good results. The regulation does not need to be any better than about 1 to 2% from no-load to full-load, but noise (ripple) must be kept to a minimum.

Figure 3 - 48V Phantom Supply

At full load, the power in the series pass transistor (Q1) will be only about 1.8W, so a massive heatsink will not be necessary. A small 10 degree/W sink

will ensure that the maximum temperature rise is less than 20 degrees Celcius, which is a generous safety margin. Q2 should also have a heatsink, as its dissipation will be about 400mW at no load. A simple flag heatsink will suffice. Using a resistor filter is awful for regulation, but gives good ripple rejection, and is simple and cheap. A 22 Ohm resistor and 2200uF capacitor will have a very profound effect on ripple applied to the regulator, which simplifies its design. This simple addition reduces ripple to about 1/30th of that without the filter, so is well worth the small extra outlay for the resistor and cap. The circuit shown has better than 80dB of ripple rejection, so the output will typically have less than 20uV of hum. As this is applied as common mode to the microphone leads, the overall rejection will be found to be more than adequate. The circuit in Figure 3 has a regulation of better than 2% from noload to full-load, which is quite acceptable. When finished, adjust the trimpot (R6) to give an output of 48.5V with no load. This will fall under full load conditions (140mA) to about 47.8V or so, indicating a regulation of about 1.4%.
Click on the PCB image to see Project 96 - a (slightly) different version of the phantom feed power supply. PCBs are available.

Headphone Amplifiers The headphone amps are based on the LM1875 power opamp. This is capable of 25W, but in this application is deliberately restricted to 0.5W into 8 Ohm 'phones. This is more than enough, but at least ensures that the amp will never clip at any listening level that is even remotely sensible.

Figure 4 - Headphone Power Amp

The units can be assembled on Veroboard (but see below first - there is a PCB available for this power amp). The 100nF caps in the supply lines need to be as close to the power pins as possible. Some care is needed with the

layout to keep input circuits well away from the output, as these devices have wide bandwidth and will oscillate if construction is unsuitable. The 56 Ohm output resistor should be at least 1W, and limits the power to the headphones. Two circuits will be needed for stereo 'phones, and the ICs should be mounted on a suitable heatsink. Make sure that the power rails for these amps are taken straight back to the regulator outputs, and not connected to the main supply buses for the mixer modules. Relatively high peak currents with unpleasant waveforms may otherwise introduce distortion into the completed unit. The Pre-Fade Listen and headphone selection and mixing circuitry will be published in the next installment.
Click on the PCB image to see Project 72 - almost identical to the above, and PCBs are available.

Still To Come Headphone selection and pre-fade listen amps. Also the talkback mic input, phono and auxiliary input modules, metering circuits, and a few miscellaneous other bits. There will also be an overall layout diagram, and a picture of what the panel of a complete system might look like. This in itself should be interesting, as I have no idea yet how I am going to make such a large drawing (and keep the file size down to something sensible). High Quality Audio Mixer - Stage 4
Rod Elliott (ESP)

Stage 4 - PFL, Monitoring, Talkback Mic & Auxiliary Modules Introduction This installment describes the pre-fade listen and other headphone mixing and switching, as well as the talkback mic amp, phono and auxiliary input modules. The pre fade listen (PFL) is designed to override the main signal, so although an individual channel is heard, it is not in complete isolation. This is useful, because the recording engineer hears some of the main signal so not only is the channel heard, but it is in context. This is adjustable. The talkback microphone is comparatively simple, as it does not require phantom feed, phase reversal or attenuator pads. It still has adjustable gain, and uses the same mic preamp as used in the channel modules. The phono and auxiliary modules are similar to the main channels, but are stereo, and have no mic preamp. The phono preamp is based on one of my

other projects (surprise), and can be added to as many channels as desired or not used at all. Monitor Switching The headphone monitor switching is fairly simple, and basically selects any of the output buses as the source. If the bus is mono (such as auxiliary sends), the mono signal is sent to both channels of the headphone amp. The switch is used to select the Aux Sends (1 or 2 - more if installed), SubMasters 1 to 4 or the Master bus. While the selector is used for long-term monitoring of a source, it can only be over-ridden using a PFL (also called 'solo') button anywhere on the desk when switched to PFL. The selector switching is shown below - like all switching, it is very simple. The inputs come from the mixing amplifier for each source indicated except for the PFL (Pre Fade Listen) which comes directly off the PFL bus.

Figure 1 - Monitor Switching

PFL and bus monitoring is completely conventional, except that a small amount of the final mix can be injected into the PFL bus to allow each channel to be heard in context rather than complete isolation. This may be preset to any level desired (including off). If my original configuration is used (as shown above), you will need a dual gang 8 position rotary switch for selection. If more aux sends (or sub masters) are used, the switch becomes bigger - the largest switch you will be able to get easily is a 12 position. An alternative is to use interlocked push-button switches, but these are fairly expensive and may take up too much space on the panel.

Pre Fade Listen The PFL bus normally (almost) floating - i.e. not connected to anything. It is only connected to the input channel(s) where the PFL button is pressed. One variation is that it can have a selectable amount of the main (Master) mix present at all times, but only in mono. When a PFL button is pressed, this signal is reduced so the channel signal is heard in context, and the combined signal is applied to the headphones. It is possible to use more than one PFL button at a time, so that two or more channels can be monitored at once.

Figure 2 - Pre Fade Listen Bus

The PFL switches are shown as a reference only. As can be seen from each module, the PFL button connects the channel to the bus via a 1k resistor. Since the feed from the Master outputs (via VR1) uses 22k resistors, the PFL signal will be 10 times that of the Master 'background' signal with the level pot set to maximum. If you don't want to use this, then the PFL bus is simply connected to the monitor switch as shown above. Talkback Microphone The talkback mic can be switched to any of the mix buses, and also has its own separate output. This output is always active (as long as the talkback switch is activated), regardless of the bus select switch setting. The output may be balanced, using the selected balanced output driver from Stage 2. There are no tone controls for the talkback mic, but these can be added if you want.

The connection to the mix buses is a fixed 50/50 split - there is no pan control. The signal is sent to both channels of the selected bus whenever it is activated.

Figure 3 - Talkback Microphone

The high pass filter may be omitted if you don't want to use it, and it can be fixed at 80Hz by removing the dual-gang pot. The series resistors must remain, but are changed to 56k, and the pot is simply shorted out. If you want, you can use the filter with any (or all) of the mic/line channels. At maximum pot resistance, the -3dB frequency is 15Hz - increase C1 and C2 to lower this (double the capacitance is the frequency. The full range using a 100k pot as shown gives a minimum of 15Hz and a maximum of 160Hz. Auxiliary Input Module The auxiliary input module is line level only, and is unbalanced. This is intended for inputs from CD players, phono cartridges, tape machines and effects returns. Each is stereo, and you can use (or not) the phono circuit as needed. The tone control circuit is a stereo version of the main mic / line channel controls - it is not shown below. The tone controls (if desired) are inserted at the points marked 'TONE'. The fader is a simple stereo affair, and there is a balance control rather than a pan pot. The line input stage is a simple variable gain amplifier, having a gain range from -10dB up to a maximum of 20dB. This will be more than sufficient for typical sources.

Figure 3 - Auxiliary Module

The outputs are connected to the mix buses using the switching shown in Stage 1, Figure 9. This may be simplified (or direct connected to the master bus) if multiple bus selection is not required. Phono Preamp The phono preamp may be any of those on the ESP website, but for the best performance overall, I suggest Project 06. This is a very high quality phono preamp. It is unlikely that more than one phono stage will be needed (if that), but for the cost (which is minimal) it is worth including. The phono preamp should be switch selected on an Auxiliary module (as shown in Figure 4), and connected to the input of the amplifier shown. In this way, the module is not dedicated to a single task. It is also possible to have the phono stage as a separate unit (but inside the mixer chassis), and connected to the desired aux channel using a patch lead.
Click on the PCB image to see Project 06 - the recommended phono preamp to use with the line input module - PCBs are available.