IQ Sampling and Oversampling Differences? *Confused*

Question

Two different thought process that produce two different alias band calculations .... What concepts are these?

If I have two 1 GHz ADC's and they sample with 90 degree offset (A comment said this is Over Sampling)

• Each ADC has an Fs/2 of 0.5 GHz
• The combined system sample streams have an Fs/2 of 1 GHz because the second ADC captures additional sample points between the samples of the first - doubling the highest frequency possible by one alone.
• The system can be thought of as having Fs of 2 GHz, whereas a single equipment Fs is 1 GHz.
• After an FFT, the entire spectrum repeats every 2 GHz, the 'system' Fs not the equipment Fs of 1 GHz.
• After an FFT, The positive frequencies are 0 to 1 GHz and the negative -1 to 0 GHz.

I am still a bit confused by the thought that maybe everything above is false because maybe IQ sampling means ...

• The FFT is being fed a complex number via IQ sample streams.
• The FFT fed a complex number does not have negative frequencies - it can span the entire single equipment Fs of 1 GHz.
• So Fs/2 is still 0.5 and Fs is still 1, each one is not doubled like the first bullet points suggest because the complex notion of FFT allows frequencies up to the single equipment Fs of 1 GHz when complex sampled.
• So also, Fs is 1 GHz and not 2 GHz (3rd bullet point from first set of points) and so the spectrum repeats every 1 GHz and not 2 GHz.
• Also in this case, there is no 'spectral inversion' from negative frequencies.

It makes a big difference to where I calculated the alias band and thus my analog filter whether the first set of statements is true or second set is true

Please can you offer advice. Still require assistance if you can 02/03/2020

Answer 1

I think you're confusing IQ sampling with oversampling. Your first scenario is oversampling, or taking twice the number of samples on a real signal. Other than stating the sampling takes place 90 degrees apart (it would actually be 180 degrees), your analysis is sound.

In the second section, you describe IQ sampling. This is usually done by mixing a signal with a nonzero center frequency to baseband using two mixers, which multiply the same signal by a sine and a cosine, respectively. The two mixing signals, separated by 90 degrees, are generally output from the same oscillator. The outputs of the two mixers are separately sampled to get the I and Q components of the sampled stream. This is typically represented as a stream at the original sample frequency but with two values (I and Q) for each sample.

IQ sampling essentially allows one to differentiate between positive and negative frequencies in the baseband signal, so it does in fact double the useful bandwidth of the signal just as oversampling does, but in a different way. Instead of doubling the Nyquist rate, it revitalizes the band from -Fn to 0.

Answer 2

Oversampling refers to the case when anytime the sampling frequency is greater than the Nyquist frequency. IQ sampling or Quadrature sampling occurs either when you perform complex down-conversion or simply use a Quadrature Hybrid coupler if no down conversion is required. You will need two ADCs in this case.

The benefit of IQ sampling is that it cancels the Image frequencies therefore if your signal bandwidth is say 200KHz you can sample at 200KHz using two ADCs instead of 400KHz single ADC.

Oversampling will occur if you sample more than 200KHz IQ or more than 400KHz using single ADC.