EE's: Question about aliasing

[jmcoreymv]

I've always been told that aliasing can only occur when sampling a signal from the analog to the digital domain (or when downsampling in the digital domain as I've found out...). However, I'm running into another scenario which seems as if aliasing is occurring. I am sampling a 4 MHz analog signal with a 12 MHz sampling rate. Then in the digital domain I am multiplying that 4 MHz sampled signal with a 4 MHz NCO. This mixes the 4 MHz down to baseband and in theory mix another signal up to 8 MHz. However an 8 MHz signal cannot be represented with the 12 MHz sampling rate, and the signal is 'folding' back onto 4 MHz.

Can anyone provide any insight as to whether this is supposed to happen?
I'm running a simulation in matlab and trying to determine if my simulation setup is wrong or my theory is wrong.

 

[MrDudeMan]

wouldnt you want to be sampling at 16MHz?

 

[jmcoreymv]

I'm sampling a 4 MHz signal and then mixing it down to baseband. I dont want the upconverted signal anyway so I'd just lowpass filter it if I had a higher sampling rate (16 MHz+).

 

[bobsmith1492]

I might be able to help next semester... (taking DSP then). We've only touched on some of that in signals class this semester, and most of that was talk after class. Sorry. I'm sure anything I could add you already know.

 

[TuxDave]

It sounds right to me. For a 12MHz sampling rate, you can only get up to 6MHz signal. If you're sampling an 8MHz signal it in essence appears as a negative 4MHz signal. If you want to bring your 4 MHz signal to baseband and you can only sample at 12MHz you may need to down convert it twice using 2MHz oscillators and filter out the unwanted high frequency signals after each down conversion.