Mixer Bus Systems
In all but the smallest mixers there is a need to connect together all the modules so they have access to the mixing buses, power-supply rails, and logic and control lines.
There are three basic ways of connecting together the modules. The smallest mixers are usually constructed on a single large PCB lying parallel to a one-piece front panel, and here ‘modules’ means repeated circuitry rather than physically-separate modules. The all-embracing PCB minimizes the money spent on connectors, and the time plugging them in during assembly, but there are obvious limitations to the size of mixer you can build in this way. A definite problem is the need to run summing buses laterally, as this results in them winding their way between controls and circuit blocks, threatening a mediocre crosstalk performance. The use of double sided PCBs helps greatly with this, but very often there are still awkward points such as the need for the feed to and from the faders to cross over the mix bus area. This can easily wreck the crosstalk performance; one solution is to use what I call a ‘three-layer board’. The mix buses are on the bottom of the PCB, the top layer above it carries a section of ground plane, and the fader connections are made by wire links above that. Given a tough solder-resist, no further insulation of the links is necessary; if you have doubts then laying a rectangle of componentident screen print under the links will give another layer of insulation without adding labor cost.
Medium-sized mixers are commonly made with separate modules, connected together with ribbon cable bearing insulation-displacement connectors (IDCs). The advent of IDC ribbon cables (a long time ago now) had a major effect on the affordability of mixing consoles. These cables naturally join Pin 1 to Pin 1 on every module, and so on, leading to acertain inflexibility in design. Large mixers use a motherboard system, where each module plugs into a PCB at the bottom of the frame, which is typically divided into ‘bins’ holding eight or 12 modules. This provides (at considerably increased expense) total flexibility in the running of buses and the interconnection of modules.
To me, it is a ‘mixing bus’ or a ‘summing bus’. I realize that some of the world spells it ‘buss’ and I am probably wasting my time pointing out that the latter is wrong but I am still going to do it. The term dates from the dawn of electrical distribution, when circuits were connected together by copper bars called ‘omnibus bars’. This inevitably got shortened to ‘bus-bars’ and in mixing consoles it was further abbreviated to ‘bus’, which somehow turned into ‘buss’. The Oxford English Dictionary says buss means ‘to kiss’ (archaic), which seems somehow not quite appropriate. I have grave doubts if my protest here will make any difference to common usage, but sometimes you’ve just got to make a stand.
Most mixer channels have both microphone and line inputs. On the lower-cost consoles these are usually switched to a single amplifier with a wide gain range, the line input being attenuated to a suitable level first. This approach is covered in Chapter 13 on microphone amplifiers. High-end consoles have separate line-input amplifiers, removing some compromises on CMRR and noise performance. Dedicated line-input amplifiers are dealt with in Chapter 14 (Full Version of book available at FocalPress.com).
Mixer input channels have more or less sophisticated tone controls to modify the frequency response, either to correct imperfections or produce specific effects. This subject is fully dealt within Chapter 10 (Full version of this book available now at FocalPress.com).
The addition of effects for general use, such as reverberation, is normally handled by an effects send system. However, if a specific effect (say, flanging) is going to be used on one channel only then it is far more efficient and convenient to connect the external effects unit in series with the signal path of the channel itself. This is done by means of an insert point (usually just called an ‘insert’), which is a jack with normalling contacts arranged so that the signal flows to it and back to the channel again when nothing is plugged in. When a jack is inserted the normalling connection is broken and the signal flows through the external unit. Inserts are also often fitted to groups. Inserts come in two versions, illustrated in Figure 17.1. The single-jack version is economical in panel space but is restricted to unbalanced operation. The two-jack version is superior because it allows the use of balanced send and returns, and in addition the OUT jack socket can be used as a direct output because inserting a jack in it does not break the signal path; only inserting a jack into the IN jack socket does that.
When the mixer has a patchbay, the insert sends (and, indeed, other console outputs) are likely to find their way there through a quite considerable length of ribbon cable, which has significant capacitance between its conductors. It is easy to get into a situation where the crosstalk performance of the console is limited by capacitative crosstalk between outputs, despite their low impedance. Output amplifiers commonly have a series resistor to isolate the amplifier from the capacitance of the cabling and prevent HF instability, and the minimum safe value of this resistor defines the output impedance, which is usually in the region of 47–100 U.
Things get worse when the layers of ribbon cable are laid together in a ‘lasagne’ format; this is very often necessary because of the sheer number of signals going to and from the patchbay. In some cases layers of grounded screening foil are interleaved with the cables, but this is rather expensive and awkward to do, and does not greatly reduce crosstalk between conductors in the same piece of ribbon. The only way to do this is to reduce the output impedance.
In a particular mixer design project, the crosstalk between the insert sends from the channels, with an output impedance of 75 U, was found to be _96 dB at 10 kHz. This may not sound like a lot, but I didn’t get where I am today by designing consoles with measurable crosstalk, so something had to be done. An effective way to obtain a near-zero output impedance is shown in Figure 17.2. Here the main negative feedback for the op-amp goes through R1, from the outside end of isolating resistor R2 and so reduces the output impedance, while the stabilizing HF feedback is taken through C1 from the inside end, where it is not subject to phase-shift because of load capacitance. With this insert send stage the output impedance was reduced from 75 U to less than 1 U and the crosstalk disappeared below the noise floor. Very similar circuitry can be used with stages that have gain, see Chapter 15. Arrangements like this must always be carefully checked to make sure that HF stability with a capacitative load really is maintained; this circuit is stable when driving a 22 nF load, which represents 220 meters of 100 pF/meter cable.
This arrangement is sometimes called a ‘zero-impedance’ output; the impedance is certainly much lower than usual but it is not of course actually zero.
In a group module, an inverting insert send amplifier is often used to correct the phase inversion introduced by the summing amplifier. A zero-impedance version of this is illustrated in the section below on summing amplifiers.
Excerpt from Small Signal Audio Design by Douglas Self.
About the Author
Douglas wears his learning lightly, and this book features the engaging prose style familiar to readers of his other books The Audio Power Amplifier Design Handbook and Self on Audio. You will learn why mercury cables are not a good idea, the pitfalls of plating gold on copper, and what quotes from Star Trek have to do with PCB design.