Some miscellaneous pulse-handling circuits
Ringing, in the world of pulses, is the usually unwanted oscillation when a level changes suddenly. The frequency of these oscillations is usually much higher than the repetition rate of the pulses. They are usually fairly heavily damped, and the cure for them is usually the introduction of some resistance to absorb the energy of the ringing.
In the world of telephony, ringing is the signal that indicates that the receiver should be picked up. It is a high-voltage alternating current signal at a frequency below the voice passband of the system. We aren't concerned with this meaning of ringing in this page.
A ringer is a circuit for deliberately producing ringing. The circuit at the right can be adjusted to produce a series of pulses repeated at the frequency of the input square wave. It is a cathode follower with a resonant circuit in the cathode lead. The bypassed 1.3k resistor provides grid bias, so that the plate current is about 0.8 mA. The same circuit works with a transistor, except that the biasing must be supplied by a split power supply. Note that the resonant circuit rings when the plate current is suddenly cut off, and the damping is controlled by the Q of the coil. When the tube is again turned on when the input voltage rises, the resonant circuit sees the low impedance looking into the cathode, and the oscillations are strongly damped.
With a square wave input at 6 kHz, the pulses contained 5 cycles of ringing, at a frequency of about 44 kHz. The input and output of the circuit are sketched at the left.
Another ringer circuit is shown at the right. It is based on a bootstrap amplifier, which looks like a cathode follower but definitely is not. The transformer is any small audio interstage transformer, such as a Mouser TM019 with turns ratio 3.75:1. Now the resistance looking into the cathode is high at all times, so the resonant circuit is not damped and rings on both edges of the input. The output, in fact, resembles the output of a spark-gap transmitter, which was also a ringer excited by sharp impulses. The ouput is a series of damped oscillations, very pretty ones, with a decay time of about 120 μs. This is a good way to study a resonant circuit, in fact.
If you are using a transistor, the bootstrap circuit may not be convenient. In this case, just put the resonant circuit in the collector lead of an ordinary common-emitter amplifier, and the effect will be the same, since the resonant circuit will be looking at the high collector resistance.
According to the IEEE Dictionary, clamping is maintaining some point on a waveform at a fixed voltage, and this is the definition I shall use here. The word apparently meant something different in analog computers.
The sketch at the left illustrates the meaning of clamping. At the top is a signal with zero dc component, which usually results from passage through a capacitor. The capacitor charges to the average value of the signal, producing this result. At the bottom is a similar signal with its lowest value clamped at 0 V, so that it now has a finite dc value. The dc level may said to be restored. Note that we are dealing with a repetitive waveform and not an isolated pulse. There is no need that the signal be accurately periodic, simply that it should have a slowly-varying dc component. The famous case is that of an NSTC video signal. It is amplified by capacitor coupled amplifiers, but then its dc value must be restored if it is to control the brightness of the picture properly.
An easy and automatic way of clamping a signal is the use of a clamping capacitor. If a diode is used as shown in the left-hand circuit, the capacitor will charge until the minimum value of the signal no longer can forward-bias the diode. This level can be set by varying the voltage Vc, but usually the anode of the diode is grounded, so that the signal is clamped to a little below 0 V. If the diode is reversed, then the maximum of the signal will be clamped to 0 V. The circuit shown with a unity-gain buffer, because the circuit is very sensitive to loading. The circuit can work into any high input impedance, of course, equally well.
The right-hand circuit clamps the minimum of the signal accurately to 0 V. This circuit is suggested in a textbook without the 1k resistor and diode D2. It will not work without D2, as you can easily determine. The op-amp saturates negatively anyway during the cycle. Also, the output is only "half-connected" to the amplifier output, so it is very sensitive to loading, and should be buffered. This circuit does not respond well to variations in the clippling level. In fact, it is not a very satisfactory circuit, and the simple capacitor clamp with buffer will usually be better. The authors who suggested it also call it a "clamped capacitor" which is erroneous--the capacitor is not clamped, but the waveform.
Transient phenomena must always be considered in circuits like these, as well as the source impedance. The capacitor clamp has the advantage that it is automatic. When the waveform changes, the clamping level will change to suit, but this may require time. An isolated positive pulse will get through the clamper nicely, while an isolated negative pulse will be swallowed (it will actually be integrated, to very little if the pulse is short). In some cases, pulse levels will have to be changed by circuits that add or subtract a fixed voltage, and so are not automatic.
Composed by J. B. Calvert
Created 29 November 2001