Sampling Rate in AD Converters: Samples vs. Signals

[Femme_physics]

The word "sample"...

In AD converters it said that:

The sampling rate - Controls how many samples are taken per second

Shouldn't the word be signals instead of samples? Are they interchangeable? The word sample is a bit odd-sounding to me in the case of electronics.

 

[vk6kro]

Sample means a portion of a signal which would be OK for an A to D converter since the A-D conversion captures a voltage for just a small instant in time.

The whole input would be considered a signal, but the A-D converter converts this into many digital samples.

 

[Dadface]

I don't know much about this but you can take an analogue signal and it digitise it by taking samples(bits) of that signal at regular intervals of time.The sampling rate is the number of samples taken per second.

 

[Femme_physics]

Sample means a portion of a signal which would be OK for an A to D converter since the A-D conversion captures a voltage for just a small instant in time.

The whole input would be considered a signal, but the A-D converter converts this into many digital samples.

Ahh..many thanks for the clarification!

 

[yungman]

Usually ADC capture the value at the input at either on the rising or falling edge of the clock. Say the ADC is capturing the input on the rising edge ( positive edge trigger), every time the clock rising edge, the ADC capture the data and output as binary bits. Sampling rate is a specification of the rate ( frequency) of the clock.

For example if the ADC spec max sampling rate is 1MHz and no minimum. That means you can clock the ADC up to 1MHz, which means you that take up to 1 mega samples per second.

Signal is usually refer to the input of the ADC to be capture. Sample rate cannot be the same as signal rate as you ask. ADC usually has a specified input bandwidth that is not necessary the same as sample rate( clock rate). I worked with a lot of ADCs and some of them have signal bandwidth much higher than sampling rate. We actually test that. Hope this clarify a little for you.

 

[Kholdstare]

A Sample is instantaneous value of a quantity, whereas a signal is the evolution of a quantity in time domain.

 

[sophiecentaur]

All varying voltage values are "signals". A string of samples, derived from a continuous signal, is still a 'signal'. The fact that this signal is a string of samples, makes it special, though because it is the result of a Modulation process and we just describe it as 'samples'. All streams of digits in a circuit are also analogue 'signals' of course.
But 'sampling' is a different issue from Analogue to digital conversion, which also involves Quantising the information contained in the original signal.
In the 'Old Days', when we didn't have the luxury of digital processing, they used 'Sampling Oscilloscopes', which could produce an analogue display of a very high frequency signal (up in the GHz region) by sampling it at a low frequency with a stream of very short pulses. You needed to have a repeating HF signal and to sample at the appropriate rate but the 'scope gave you picture of the HF signal as a set of analogue samples. This was extreme sub-sampling and you had to be extremely careful in interpreting the display, of course but my point is that they were Analogue Samples.

If you want to convert an analogue signal to a digital one then the digital values need to be associated with particular time values on the signal. Hence, you 'sample' the continuous analogue signal and then convert that sample value. If you don't do that then your ADC process may get confused. There are all sorts of ADC methods but they all take time to deliver a value, in between sample times you could imagine all sorts of weird combinations coming out of 8 parallel bits. You need a 'clock' to sort all this timing out. Even if you don't actually use analogue pre-sampling, it is still important that the timing of the digital values is consistent with the timing of the values of the original analogue signal.

BTW, it is not necessary to sample at a regular rate as long as the reconstruction equipment (DAC) knows the timings of the individual samples (and as long as the sample spacing is narrow enough to satisfy the Nyquist criterion for that portion of the signal). A lot of bit reduction methods involve this strategy, aamof.