only I mixer instead of IQ mixer?

Started by mary1234567 February 6, 2006
I have tested use only cos(2*pi*fmix*n/fs) to demodulate a BFSK input
signal. And it worked. Question is how much worse will it be? Will it be
3dB worser?
Can I compensate this by oversampling instead? For example will a double
oversampling give 3dB more SNR.

Any answer is welcome!
/Mary


mary1234567 wrote:

> I have tested use only cos(2*pi*fmix*n/fs) to demodulate a BFSK input > signal. And it worked. Question is how much worse will it be? Will it be > 3dB worser?
You can have some trouble with locking carrier and symbol sync. How much worse it is going to be depends on the application.
> Can I compensate this by oversampling instead? For example will a double > oversampling give 3dB more SNR.
You can compensate this by reading more books. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
mary1234567 wrote:

> I have tested use only cos(2*pi*fmix*n/fs) to demodulate a BFSK input > signal. And it worked. Question is how much worse will it be? Will it be > 3dB worser? > Can I compensate this by oversampling instead? For example will a double > oversampling give 3dB more SNR. > > Any answer is welcome! > /Mary >
Without knowing what fmix, n and fs are (I figured out the '2' and 'pi') it's hard to say anything definitive. Do you really mean 'binary frequency shift keying', or did you mean 'binary phase shift keying', which would be BPSK? If you are mixing the signal down to some frequency other than DC then you will get complete information, but the amount of noise you get will depend on the electronics in front of your mixer. The reason to use I/Q mixing is to go straight from RF to DC, without needing much electronics in front of the ADC pair. With a good radio you can get by with a single mixer (see http://www.wescottdesign.com/articles/MSK/mskTop.html) but these days you spend less money with an I/Q downconverter and a faster DSP chip. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com