Question about sampling an input sine wave to system using A/D

[andrew257]

hi all

i am sampling an input sine wave to my system using an A/D, i am then measuring using a scope the output from my D/A.

The problem i have is with a 1khz sine wave going in from my function generator the output looks distorted and does not look like the input signal.

Now i was aware i must sample at least twice the highest frequency component which i am doing, in this case i am sampling at 44.1khz. More than enough for a 1khz input.

when i increase the input frequency the waveform becomes even more distorted

Am i wrong to think they wavefroms should look identical in shape size etc? What frequency does a sine wave have to be sampled at to get an identical match at the output?

i am assuming the Nyquist theorem is only to retrieve the frequency component of the waveform and not the actual shape?

 

[electricpete]

Re: sampling problem

You need a low-pass filter operation to recreate the original signal.

 

[bulx]

Re: sampling problem

Nyquist tells you that there is way to recreate the original signal if you have samples of the original signal at twice the BW of the highest component. The way is to have a brickwall (ideal filter) which will pass through all the signal components.

In short, exactly like what the earlier response says: filter it
-b

 

[burnsmedia]

Re: sampling problem

For example, if you use this filter

you must sample with your ADC at
Fs=Fc+Fmax.

See Discrete time signal processing. Oppenheim.