I have a signal with components at 1200 and 1800 Hz. When I mix this signal with a 1500 Hz signal to create components at -300 and +300 Hz (then LPF) a DC offset presents itself when the input signal is going form 1200 to 1800 Hz at a period of 1200 Hz (msk modulated signal). This strange affect only occurs with specific time vectors. What could cause this? I understand how mixing two signals at different frequencies create signals at f1+f2 and f1-f2 however where is this power at DC coming from?
Below is an image to show my problem:
The first signal is the MSK signal, the second is the 1500 Hz signal I'm mixing with it, the third is the result and the fourth is the filtered version of the result to recover only the low frequency parts.
The problem I believe lies in that at certain time offsets there exists an alias of 1500 at 300 (and -300) that is destructively mixing with my msk signal and producing a dc value. The result in the center there when you see the dc value should be a series of half sine wave pulses at 300 Hz.
Is there a way I can avoid this problem and recover my +/- 300 Hz signal from the msk signal?