Capturing and downconverting a 868MHz signal

[cynop]

Hello,

I am trying to use a USRP-2920 to demodulate a BPSK signal. I know that the signal has a data rate of 100 bps and that the carrier is around 868 MHz (not sure about the exact frequency but it's 868MHz +- 20KHz). In essense that means that during modulation a cosine of frequency 100Hz has been multiplied with a cosine of frequency 868MHz, with the amplitude being regulated by the NRZ bitstream (I hope that's right). However I have a couple of questions:

1) First of all, I would assume that to accurately capture and demodulate the signal I would need the IQ rate to be at least 2x868MHz. However in the example "USRP Packet Receiver" this is not the case. An IQ rate of 400K is more than enough to demodulate the signal. Obviously I have misunderstood something, but I'm not exactly sure what.

2)I am trying to downconvert the signal using the "MT Downconvert Passband VI" in an effort to record the 100Hz wave. Is there any way to downconvert the signal? Considering that the maximum IQ rate is 50M samples/second, there is no way of satisfying the Nyquist criteria right?

I am sure I am missing something obvious here, but I can't figure out what. Any help will be greatly appreciated

[ITUbahadir]

Hi,

I realized same project like you. I try to help you. Firstly,

1) IN USRP IQ Rate does not depend on carrier frequency. IQ rate means sample number of hardware. You can adjust the IQ rate with symbol rate and number of samples per symbol. your symbol rate is equal to bit rate so, data rate in order to that your demodulation is BPSK. If you product samples per symbol and sampling rate, you can reach IQ rate. Also, amount of IQ rate causes overload of buffer. For example, for 9600 baudrate, you can use 192k IQ rate and 20 samples per symbol in 430 Mhz.

2) In implementation of demodulation, you cannot use downconvert block. It just is used in simulation. You can use the MT Resample (Complex Cluster) for downconvert in order to decrease sampling rate. 

[cynop]

Hi ITUbahadir,

Thanks for taking the time to reply and sorry for taking so long to respond. I am not sure i understood everything you said.

1) The sampling rate, but satisfy the Nyquist Theorem right? Therefore for a carrier of 868MHz, it must be sampled at least doubled that. And if assume there are 8 samples per symbol that would mean that the IQ rate is 868x2x8x10^6 samples/sec, which obviously is wrong, as the example works fine with a much smaller IQ rate

2) Downconvert and Downsampling is not the same thing though?

[emnajaoua]

Actually I think that the IQ rate here correspond to the IF frequency that means the sampling IQ frequency should be superior to twice the IF frequency.So you should look at the IF frequency not the carrier frequency (868 Mhz).

[muffin.vi]

Hi cynop,

I don't think I fully understand your application, but I can help out a bit about your questions on demodulation. The USRP 2920 has a direct conversion architecture. That means that it uses a mixer with a LO at roughly the same frequency as the carrier signal. The result is that the carrier frequency gets mixed out, and you're only left with the signal modulated on the carrier. This is called the "baseband" signal. So, when you select your IQ rate, think of that being the IQ rate for the signal, not for the signal+carrier. So, the thing that you really need to know is the bandwidth of your BPSK signal. That will tell us what the IQ rate needs to be set to. 

To make things a bit more complicated, IQ rate can be confusing, too. Since the I and Q are sampled separately, you're really getting double the number of samples for a given rate. So, a Nyquist rate for a 400 kHz bandwidth signal, is actually just 400 kS/s IQ rate. Because of filter roll-off, however, it's a good idea to sample about 20% faster. 

[emnajaoua]

I am sorry you are right muffin.vi .the IQ sampling rate correspond to the bandwidth of the signal.

You can take a look at this document page 18.

title of the document : National Instruments RF Platform
Cristóbal Rus – Ingeniero Especialista en Academia
Ángel Sendín – Ingeniero de Sistemas