Chip Parade (Part 3)
Get switched on
More application notes for common chips and devices — Robert Penfold outlines electronic switching.
In many areas of electronics there has been a steady trend towards the replacement of mechanical switches by electronic switching. With so much electronic music equipment requiring complex switching it is perhaps not surprising that this trend has been as strong here as elsewhere. There are two main reasons for using electronic switches instead of mechanical types. One is simply that mechanical components are more prone to failure and have only a limited lifespan, whereas electronic switches have no moving parts, do not wear out and will often outlive the user's interest in the equipment to which they are fitted!
The second reason for their use, and one that is becoming increasingly important in electronic music, is the fact that electronic switching enables equipment to be controlled by a microprocessor or other form of digital control system. In conjunction with other digitally controlled circuits (VCAs, VCFs, etc) this provides almost limitless possibilities.
There is a popular misconception that electronic switching is less 'noisy' than mechanical switching in the sense that it is less prone to produce switching 'clicks'. This is not really true, and if anything mechanical switches are probably slightly less likely to produce switching 'clicks'. Mechanical switches are more 'noisy' in the sense that they do not switch cleanly from one state to the other, and suffer from so called 'contact bounce'. This is relevant only when controlling digital circuits, but when switching analogue signals this factor is of little or no importance.
There are quite a number of integrated circuit electronic switches, but probably the two best known devices are the CMOS 4016BE and 4066BE types. These have identical pinouts, as shown in Fig 1, and are also similar electrically. The 4066BE has a slightly higher specification with a lower 'on' resistance and wider bandwidth. In audio applications the wider bandwidth of the 4066BE is irrelevant since the 4016BE can handle frequencies way beyond the upper limit of the audio band. Also, it is unlikely to matter much whether the switch has an 'on' resistance of (say) 100 ohms or 300 ohms, and the marginally cheaper 4016BE is therefore the type that is normally used in electronic music applications.
Although in Fig 1 some pins are marked inputs and others are labelled outputs, they are genuinely bidirectional devices. In other words, they will work just as well with the inputs and outputs transposed!
With these devices there is a separate control input for each switch, and the switch is set to the 'on' state by a high control voltage (ie, a voltage equal or nearly equal to the positive supply potential). It is important to realise that these switches cannot be used just like ordinary mechanical switches. The first point to bear in mind is that when switched on they provide a significant resistance of typically between about 80 and 500 ohms, with the precise figure depending on the type used and the supply voltage (higher supply voltages give lower 'on' resistances). The switches can only handle low level signals, and could not, for example, be used to control the output of a power amplifier.
A more important limitation is that the input voltage must always be between the two supply rails, and ideally the voltage swing should be kept quite low so that minimal distortion is generated. The distortion occurs because the 'on' resistance is not a true resistance, and does vary slightly with changes in input voltage.
Figure 2 shows a typical circuit using a 4016BE, and this circuit simply feeds one of the inputs through to the output while blocking the other. IC2 is used in what is almost a standard operational amplifier two-input mixer circuit, but a CMOS analogue switch has been added in each input's signal path. The switch for input A is controlled direct from the control voltage, but the one for input B is controlled via a CMOS inverter. Thus only one switch at a time can be turned on. When the control voltage is high channel A is selected, and when it it low channel B is selected (the CMOS inverter can be one section of a 4069BE).
A slight problem with this basic arrangement is that the switches do not fully block the signal when in the 'off' state, due to stray capacitance causing coupling through the device. This is only likely to be significant at high audio frequencies, but the problem can be eliminated by splitting each input resistor into two sections and coupling each junction to earth via another CMOS switch. The circuit is arranged so that each additional switch is turned on when its channel is disabled. It then reduces the signal level reaching the main switch to a much lower level so that the small amount of breakthrough becomes insignificant.
An advantage of using the switches in conjunction with an operational amplifier is that most of the input signal is developed across the input resistors, rather than the much lower 'on' resistance of the switches. This gives a low level of distortion.
There is another CMOS analogue switch which uses the pinout configuration of Fig 1; the 4416BE device. This differs from the 4016BE in that switches B and C are 'on' when their control voltage is low, rather than when it is high. The 4416BE could therefore be used in a circuit of the type shown in Fig 2 with the inverter stage omitted. However, the 4416BE is relatively difficult to obtain and nearly ten times the price of a 4016BE!
CMOS switches are not only usable with AC signals, but can be used with DC signals. Devices such as the 4016BE are sometimes used in applications such as envelope generators.
The circuit of Fig 3 uses another CMOS switch, the 4051BE, in an 8-channel selector circuit. The 4051BE is effectively eight CMOS switches having their outputs taken to a common terminal (pin 3). By turning on the appropriate switches the desired input can be selected. However, the control inputs are not directly accessible, and the desired switch is activated by feeding the correct address to the address inputs. For instance, taking address input two high would select channel 2, or taking address lines one and four high would select channel 5. This kind of circuit is obviously ideal where the equipment is to be controlled by a microprocessor.
With the 4051 it would be awkward to use additional switches to suppress breakthrough of the disabled channels, and the alternative approach of using a lower input impedance (about 4k7) is used to render stray capacitances insignificant. A slight drawback of this is an increase in distortion, but the circuit still performs well in this respect provided the input level is no more than about one volt RMS or so.
The 4051BE has bilateral switches and it can be used in applications where a signal source must feed one of up to eight inputs. It is also suitable for applications where DC signals must be multiplexed or demultiplexed, such as where a single analogue to digital converter is to be used with up to eight input channels. In electronic music there is increasing use of analogue sound generators and signal processors with digital control circuits, and this approach gives great versatility at moderate cost; see for example the Prophet or Chroma synthesizers. The 4051BE is ideal for use in equipment of this type. In DC applications the VEE terminal at pin seven can be taken to a negative supply rail so that the input/output voltage can go right down to, or even below, the negative supply rail.
There are several similar devices to the 4051BE incidentally, such as the 4053BE triple two channel type, and the 4067BE which has no fewer than 16 input/output channels.
CMOS analogue switches are popular due to their low cost, but for the ultimate in performance there are superior devices, intended for use in digitally controlled hi-fi equipment. They are also just as suitable for electronic music applications. Two examples of such devices are the LM1037 and LM1038.
Fig. 4 shows a circuit which employs an LM1038. In fact this is a stereo four channel device and only one channel is shown (the pin numbers in brackets are for the other stereo channel). Pin 12 provides a bias voltage which is decoupled by C6 and fed to the inputs via the resistors. C5 is a supply decoupling capacitor which should be mounted as close to IC1 as possible. The other capacitors are just input and output coupling components.
There are four control inputs, one of which is just an inhibit terminal which cuts off all the inputs when taken high. There are two address inputs which select the desired input in standard binary fashion. The fourth terminal controls the latches on the address inputs, and this operates the latches on a negative to positive transition. This makes it very easy to control the circuit from a microprocessor. If the latches are not required, pin 16 is simply tied to the positive supply rail. If more than four channels are required two or more devices can be connected in parallel, but pin seven of each device must be wired together.
The performance of this device is very impressive with just 0.04% THD with a one volt RMS input, an output noise voltage of 5uV, and 90dB suppression of disabled inputs. This should be sufficient to satisfy the most critical of users!
The LM1038 has a slightly more straightforward method of control with the appropriate one of the four control inputs being taken high to select the desired input (pin 3 for channel 0, pin 16 for channel 1, pin 18 for channel 2, and pin 1 for channel 3). With both devices the control inputs are 'high' when they are at any potential between two and 50 volts; the LM1038 is thus better suited to non-microprocessor digital control.
Feature by Robert Penfold
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!