A transient recorder in general is a device which captures a signal, although in the context of this study, the added aspect of signal digitization is included. This procedure is of general importance but it is particularly so when the signal is related to an infrequent event. Consider for a moment some natural phenomena that may last a millisecond, but only occur once every few hours or days. Woe betide the observer armed only with an oscilloscope who happens to nod off or even sneeze at the critical moment!
This problem is solved with the arrangement shown above. The START signal indicates the beginning of the measurement period. For the rare events mentioned above it could be generated by triggering on the signal itself, just as in an oscilloscope. This signal sets the control flip flop, setting the contents of the counter MAR, acting as a memory address register to zero, and allowing clock pulses to initiate ADC conversions with the SMPL command. With each sample the MAR up counter is incremented. Thus the memory address is a surrogate for time so that the nth memory location corresponds to
At the end of conversion (EOC) time after the nth clock pulse the data register will contain the digitized value of the signal at tn and with the RAM write enabled this quantity is transferred into memory. The process of sampling and writing the sampled value into memory continues until the MAR overflows and the carry resets the control flip flop. At this juncture the memory contains an array the dimension of which is determined by the capacity of the MAR. The components of the array satisfy
where v0 is the ADC quantization voltage. Implicit in this design is that the ADC conversion time is less than the clock frequency. The conversion time is the essential limiting factor with regard to the sampling time interval.
Note that in the system shown the clock continues incrementing the MAR after the completion of the measurement cycle but the data is not altered since no new sample measurements are made. This feature is optional but common. The MAR and MDR are usually connected to a display unit in the form of a pair of DACs driving the x- and y-inputs of an oscilloscope. In this way the memory contents are continually presented so that a continual display of the captured signal is achieved. The display circuitry is not indicated.
The display feature is indicative of an important feature of the whole process. Once the signal is captured it may be manipulated in all sorts of ways, which is why digital signal recording is so powerful. In the display mode described above the non-repetitive or infrequent signal is now continually reconstructed producing a repetitive signal appearing at a controlled rate.
Other possibilities arise. If the signal is created once, after a suitable pause, a flexible signal delay is achieved. If the time span of the captured signal is inconveniently long, provision can be made to switch to an increased clock frequency at the carry indicating the end of the measurement phase. This results in a time compressed repetitive version of the signal. This version might for example be fed to some simulated system representing a mock up of a design intended to handle the actual signal, for which testing might be virtually impossible.
An extremely important aspect of this form of transient recording is that the captured signal in digital form is ready for detailed computational analysis, a procedure referred to as digital signal processing. This activity is central to modern information science, which forms the basis of modern metrology in science, engineering and medicine. To give one example, considerable effort is being spent on digital processing of electro-cardiographic signals with the objective of extracting more detailed diagnostic information.
7.1 Single Sampling
The limitation on sampling frequency imposed by the conversion time renders the system inapplicable for the recording of transients the duration of which is of the order of this time. One approach is the use of oscilloscopes with long persistence phosphors, but unless a large array of photo-detectors is employed to form an image the information is still both volatile and analogue. A modified procedure may be used when the transients are repetitive, ie pulses of identical shape occur over and over again. This procedure makes use of a digital delay. Consider a DAC connected to one input of a comparator. The other input is connected to an integrator with resistance R and capacitance C. The comparator will fire when the integrator equals the DAC output. At t = 0 the DAC reference voltage is applied to the integrator, producing a ramp voltage vI(t). If the contents of the DAC correspond to the word X, coding for the integer n = I10(X), then the relations
follow and the comparator transition occurs at
Thus the delay may be controlled digitally through manipulation of the contents of the DAC.
A conceptual arrangement is shown above. The input signal is sent both to the ADC and to a Schmitt Trigger S.T. The S.T. output acts as the clock giving one clock pulse for each pulse input. The MAR is thus incremented by each input pulse. The MAR content is connected both to the RAM as normal and to the digital delay. Hence the sampling time is advanced on each signal in step with the memory address. The ADC is coupled to the MDR as for the transient recorder and the rest of the circuit functions as before. With this arrangement only one point on a given signal is measured, but a sampling of the signal may be generated by measuring many duplicate signals. The spacing between time points is no longer limited by the conversion time. It is necessary that the spacing between signals be greater than the conversion time, but not between time points.