Hallelujah for Two Switching Thresholds

When higher data rates are transmitted by cable over longer distances, noise is generated that overlays the wanted signal and makes it difficult to clearly determine the digital signal. This can result in digital logic elements no longer being able to process the wanted signals. The noise can be reduced by shielding the line, but not completely suppressed. The use of input filters, e.g. RC networks, additionally reduces noise – but also the maximum data rate.

A better way to regenerate degenerated signals is to use Schmitt triggers. They are used, in particular, in devices that establish a connection between an analog and a digital environment – in other words, wherever noise suppression is absolutely essential.

How does a Schmitt trigger work?

A Schmitt trigger is a comparator circuit with hysteresis (Figure 1). It analyzes the analog input signal and converts it into a switching pulse when one of two defined threshold voltages is undershot or exceeded. The difference between the two switching thresholds, the switch-on voltage (V_on) and the switch-off voltage (V_off), is referred to as hysteresis.

Figure 2 shows how noise is removed from a digital signal using a Schmitt trigger: if the input voltage exceeds the value V_on, the Schmitt trigger switches on and outputs a signal. This remains stable despite the noise until the input voltage undershoots the value V_off. The further apart the two threshold voltages, the greater the noise immunity of the system.

Click image to enlarge

Figure 2: Noise removal of a digital signal using a Schmitt trigger

Digital inputs do not like analog signals

Schmitt triggers are therefore used for digital signal transmission, as digital circuits require discrete voltage signals at the input. If a continuous signal with a voltage level outside the defined levels is fed into a logic gate, for example, unique signals are not issued at the output. A consequence of this can be oscillations with indeterminate signals, which lead to instability and increased power consumption. 

Square wave signal as the basis of digital signal transmission

The best-known special case of digital systems is the binary signal. In this case, only the states “0” and “1,” for example 0 V and 5 V, exist. A signal of this kind, which switches back and forth between two values and has a rectangular progression over time in a diagram, is called a square wave signal. Square wave signals consist of a fundamental sine wave and odd harmonics. To transmit them in as loss-free a way as possible, you need to know more than just the basic frequencies, which can be read from the amplitude-time diagram. Rather, the actual signal must be analyzed in terms of the frequencies it contains. The relevant amplitude-frequency diagram shows the other frequencies present in the signal, along with their components.

Click image to enlarge

Figure 3: A square wave consists of a fundamental sine wave and odd harmonics (Fourier analysis)

Many factors distort the wanted signal

When transmitting digital signals over large distances at high data rates, various sources of interference impact the wanted signal:

The low-pass character of the line distorts the signal linearly, i.e. low frequencies are transmitted while high frequencies are filtered out. This limitation makes it difficult or even impossible to detect the signal unambiguously at the output. The square wave signal is scattered.

Electromagnetic interference comes from fluorescent lamps, power grid switchgear, or inductors (coils or motor drives). They have randomly distributed frequency spectra that interfere with or overlap the line frequencies and thus distort the signal. To ensure the quality of the wanted signal is still sufficient, the signal-to-noise ratio (SNR) must be high enough. However, a simple amplifier does not distinguish between wanted and unwanted signals, but simply increases both equally.

Other lines or adjacent data channels generate crosstalk due to capacitive or inductive coupling. This is indicated by rapidly rising or falling signals. If a signal of this kind passes through a transmission line, it induces crosstalk noise in an adjacent line, which propagates as a pulsed noise.

The bandwidth of the transmission path also impacts the wanted signal, as the frequency space of real signal transmission systems is limited. The smaller the bandwidth, the more the square wave signal is scattered, since the higher the sine frequency, the lower the amplitude component. And of these frequencies, only those that are within the bandwidth of the transmission path pass through the line. As the gain increases, the usable bandwidth decreases, as the gain bandwidth product (GBP), which is constant, applies to each amplifier. The high frequency components of the input signal enter the output range with a smaller amplitude.

The composition of the signal from several sinusoidal oscillations with varying frequencies becomes increasingly noticeable the more components are in the transmission path. The reason for this is the time differences of the various frequencies through these components that prevent them from reaching the receiver at the same time.

Impedance mismatches cause a portion of the transmitted signal to be reflected at both the transmitting and the receiving ends of a line. These reflections cause additional errors. For example, in CMOS logic ICs, they lead to a greater signal delay and ringing, as well as overshoot and undershoot.

With mechanical switches and relays, chattering (contact jump) often occurs: instead of a clean transition from zero to full current, a rapidly pulsed electric current is generated. This causes problems, especially with logic ICs, since they respond fast enough to misinterpret the on-off pulses as data signals. Schmitt triggers filter out these “bounces” thanks to the two switching thresholds. This process is also referred to as “debouncing.”

To avoid signal output instabilities, logic components must be controlled by pulses with steep edges. However, external signals often have a finite rise or fall time (slew rate). For example, clock signals from crystal oscillators used to drive PCB devices may have a low slew rate. By reducing the track resistance and/or capacitance on the board or increasing the drive capability of the input signal it is possible to avoid distortion of the output waveform. These methods are, however, usually time consuming and costly. A simple way to eliminate these problems is to use a Schmitt trigger. It converts a slow or noisy signal into a clean signal with sharp edges before passing it to the logic gate.

Click image to enlarge

Figure 4: A Schmitt trigger converts slow edges into sharp rising edges

Wide range of applications

Assembling a Schmitt trigger from discrete parts is a demanding and sometimes time-consuming process; calculation of the individual components is very complex. A stand-alone product such as the 74LVC14A from Diodes is much more convenient. It provides six independent Schmitt trigger inverter buffers and is designed for operation with a wide power supply voltage range of 1.65 V to 5.5 V. The inputs are tolerant to 3.3 V or 5.5 V, allowing the device to be used in a mixed voltage environment. The CMOS technology guarantees low power consumption. The 74LVC14A is fully specified for partial power-down applications. This circuitry disables the output, thus preventing damage to the current backflow when the device is powered down. The gates perform the positive Boolean function, i.e. they work as inverters, thus negating the output signal.

Rutronik