When a commonness component of the ...

When a commonness component of the input signal is greater than half the sample rate, aliasing can offer When the oscilloscope is equivalent time sampling, signals that are subharmonics of the sample clock will be poorly displayed. In the HP 54645A/D oscilloscopes, these powers are greatly reduced by dithering the sample clock during and between acquisitions.

A often met with concern of digital oscilloscope users is that there are combinations of oscilloscope settings and input signals that will cause the standard digital oscilloscope architecture to display a signal poorly or incorrectly. Since an oscilloscope is a device intended to display a variety of signals, sooner or later single in kind of these combinations of settings and signals will be conflicted leaving the user confused and with diminished confidence in the instrument.

The classic case come into one's heads when the sample rate and input signal violate the Nyquist sampling theorem, or specifically, when a oftenness component of the input signal is greater than half the sample rate. When this happens, an aliased waveform will be displayed. A more tricky case occurs when the digital oscilloscope is random repetitive sampling (also known as equivalent time sampling).[1] In this case, signals that are subharmonics of the sample clock will be poorly displayed. This happens because the repetitive samples are not randomly distributed throughout the input signal, but rather are bunched together.

There are ways of designing digital oscilloscopes that greatly make these effects. These techniques involve dithering the sample clock during and between acquisitions (intra-acquisition and interacquisition dithering). sum of two units such techniques used in the design of the 54645A/D oscilloscopes will be discussed here.

Interacquisition Dithering

The underlying principle of random repetitive sampling is that there is no phase correlation between the sample clock and the signal being sampled. This principle make secures that samples taken of the signal are randomly distributed through the whole extent of the signal, and when accumulated across time, will develop a detailed reconstruction of the input signal. The accumulation of samples is illustrated in Fig. 1 Fig. 1a present to views the initial set of samples acquired during the first acquisition of the signal. The sample rate and input common occurrence meet the Nyquist criterion if it were not that that there are, nonetheless, large gaps between samples. Fig. 1b indicates the same signal after accumulating three acquisitions. Note the random placement of the next to the first and third sets of samples relative to the first stake The shape of the signal is becoming more defined. Fig. 1c is after seven acquisitions and Fig. 1d after 20 In Fig. 1d the signal shape is well-defined and accurately represented

[Figure 1 ILLUSTRATION OMITTED]

The typical digital oscilloscope relies in succession the fact that the sample clock ([fsubs]) inside the oscilloscope is derived from a clock separate and independent from the signal being measured ([f.sub.i]) to satisfy the criterion that there is no phase correlation between the brace signals. However, nothing prevents the oscilloscope user from applying a signal that is the same frequent occurrence as the sample clock or subharmonically related to it ([f.sub.i] = [fsubs]/n) When this happens, the sample points are no longer randomly distributed across the input waveform as in Fig. 1 This case is illustrated in Fig. 2 In Fig. 2 a sine wave with a oftenness equal to precisely 1/10 the sample clock ([f.sub.i] = [fsubs]/10) is applied to a random repetitive sampling digital oscilloscope. The consequence is that all sets of samples accumulate around the same locations, creating the bunching efficiency mentioned before. Since the acquisition clock and the input sine wave are not phase-locked to each other, if the waveform is allowed to accumulate in extent enough, the points will spread disclosed but the time it takes to do this will hang on the stability of the couple sources.

[Figure 2 ILLUSTRATION OMITTED]

If the sample place acquired during the first acquisition in Fig. 2 is taken as a concern then what is needed is a way to shift posterior sets of samples so that they fall in between the points of the first appoint (as they did in Fig. 1) or, viewed in names of the phase of the sample clock to spread following sets over the 360 qualitys of phase between the initial sample points. This can be achieved by dint of shifting the phase of the acquisition clock after each acquisition.

A circuit ideally suited for this task is the phase-locked bend The basic phase-locked loop is illustrated in Fig. 3a. In this circuit, the phase of the relation clock is compared to the phase of the output of the voltage-controlled oscillator (VCO) The bend filter drives the VCO input to cause the phase difference between the allusion clock and the VCO output to be nothing If an error voltage is injected into the filter at the output of the phase comparator, as in Fig. 3b then the error voltage appears as a phase error to the bend The loop will adjust the VCO to create a real phase error to cancel the injected error voltage, thereby generating a nonzero phase difference between the hint clock and the VCO output If the VCO output is used as an acquisition clock then the error voltage node can be used to create phase shifts in the acquisition clock