input of mixer with BANDWIDTH .What about the settings of L.O ?

[Fovakis]

Hi there. I have a DSP that has an output of 16QAM fcenter=10khz and BW=17khz and i want to upconvert it to an IF=30Mhz. So i use for L.O a signal generator of 30010 Khz now. (i used a vco but it had a lot of phase noice so i replace it with a signal generator. But what about the bandwidth of the L.0?

Will i have a problem at the output of the mixer ? I am beginner at RF systems so please answear as simple as you can.

Thanks a lot George

 

[biff44]

yes, you will potentially have a problem. The output of the mixer will have the following signals:
1) your desired 30 MHz IF signal
2) an undesired 30.01 MHz LO leak thru, that will be around the same amplitude as the flat top of your 30 MHz IF
3) an undesired 30.02 MHz IF signal

It is a QAM signal, so you can not screw up its phase or amplitude very much or you will induce bit errors. But if you intend to transmit this signal wirelessly....you will have to reduce the amplitude of signals 2 and 3 (possibly by as much as another 40 dB or so). You will find that you can not buy a 30 MHz bandwidth filter that will pass 17 KHz of bandwidth while also attenuating 30.01 MHz LO leakage at all.

 

[Fovakis]

Thanks for your answer biff44. At the time i am calculating this in simulation and i will see it in spectrum analyser in some days!What you propose for solution? maybe to change the fcenter of DSP output? I am not sure if i can change it ... Also if the L.O has the same BW with the QAM16 Bw=17khz .Is it helpful ?

 

[biff44]

yes, the problem is alleviated with higher DSP output frequency, if for no other reason than that the 30 MHz bandpass filter becomes more realizable, and flatter in the passband.

It is tempting to use a low DSP output frequency, since the chip is cheaper, the clocks all run slower etc. But...you really have to do a tradeoff--run as high a clock rate as you can without significantly increasing DSP cost or loosing ENOBs.

You might want to post on the DSP forum for a better answer.

ALSO, at 30 MHz, if is very feasible to use a "single sideband modulator" instead of a simple mixer as he upconverter. That will reduce the unwanted upper sideband another 25 dB.

 

[Fovakis]

Hi biff44 again. So, after the mixer i put one filter to cut the unwanted zone. So instead the mixer and the filter you propose a
single modulator. As i understand the SSM wil need also a filter to cut the unwanted L.O freq=30.1MHz. So why a SSM and not a MIXER?

Also the SSM that i see here **broken link removed**
they have inputs I and Q. But from the output of my DSP i have a single output that transfer the I(t)+Q(t) signal. Can i use only the I or Q port of modulator ?

George