Is there known any case in which non-instant output settling of a digital-to-analog converter is preferable? Ideally, the output of DAC is a step function, with immediate transitions. Perhaps if the output was settling in a proper manner (for a given case) it could better resemble a signal (decrease quantization noise?)? Is this necessarily the same thing as ideal DAC + filter?
qualify the DAC input and filter any output noise, you will be fine depending on PCB layout and signal speed. Ground planes are paramount in design. i.e. A bread board prototype may operate totally different from an appropriate PCB layout in terms of noise and signal integrity.
Dear Dr. Adam Rudziński, DAC settling time is the elapsed time from input code application until the output arrives at and remains within a specified error band around the final value. It is usually specified for a full-scale 10V transition. The delay time is very small and is almost entirely due to propagation delay through the DAC and output amplifier.During this interval there is no output movement. During During this interval there is no output movement. During slew time the output amplifier moves at its highest possible speed towards the final value. Ring time defines the region where the amplifier recovers from slewing and ceases movement within some defined error band. slewing amplifiers generally have extended ring times, complicating amplifier choice and frequency compensation.
Source:
http://cds.linear.com/docs/en/application-note/an74f.pdf
Dear Sirs, my question must have been imprecise, because it did not concern these subjects. Here, I am not interested in desing guides for using DACs or reasons for settling. The problem I am thinking about is whether non-instant settling of output, which is considered to be a kind of "distortion", could be, in opposition, used for better "conditioning" of a signal. For example, perhaps a more smooth output waveform could better match the true signal than approximation by step function (provided by ideal DAC) and decrease somehow noise related to the conversion process (not related to implementation on PCB)? I tried to find some papers, but I have found just one, and not quite relevant.
Adam, I think the simple answer is no, when using the sampling theorem. Reason being, in order to reconstruct the original signal "perfectly," the samples have to be infinitely narrow (and must occur at least at twice the frequency of the highest frequency component of the original signal).
A quick and dirty explanation. When sampling a signal, in the time domain you are convolving the original f(t) with a string of narrow pulses. In the frequency domain, you will see a train of spectra which are images of the Fourier transform of the original function, and an envelope over these that is the Fourier transform of the sampling pulse.
As we know, the Fourier transform of an impulse is simply 1, from -inf to +inf. So if the sampling pulses are impulses, infinitely narrow, you will see a train of unattentuated "perfect" spectra of the original. A low pass filter in the DAC will restore the original.
If the pulses are not infinitely narrow, their Fourier transform will instead be a sinc(x) function (sin(x)/x). It looks like a rounded bandpass filter applied to the original signal's spectrum, and then a decaying series of harmonics. The wider the sampling pulses are, the narrower is the main lobe of this sinc function (potentially attenuating the desired signal spectrum at the extremes), plus all this noise created by the harmonics. So the DAC may end up with a less that flat rendition of the original signal's spectrum, not to mention noise created by the harmonics, encroaching on the desired spectrum. The wider the sampling pulses are, the worse is the effect.
Here's a good paper on the subject.
http://lavryengineering.com/pdfs/lavry-sampling-theory.pdf
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing