Re: [Discuss-gnuradio] regd: IIP3 and OIP3 measurements for USRPN210

[David Bengtson]

Hi-

Comments Interspirsed

On Wed, Apr 16, 2014 at 4:22 AM, Gayathri Ramasubramanian <address@hidden> wrote:

Hi

My project is dealing with Tx/Rx Characterization. I wanted to know if anyone has tried finding the IIP3 and OIP3 for the USRP N210 with WBX and SBX daughter-boards.

I did an experiment on Tx side ( USRP behaving as a TX, with the GRC on host generating the two tones & output from USRP goes to an attenuator of 10 db before going to a spectrum analyzer) and got the following results. (as in the excel sheet attached.)

I am planning to do this for the RX side using Two signal generators instead of the GRC flow graph. However I had a few questions:

1. I found the IIP3 for SBX to be given as 0 dBm & for WBX as 50 ~ 100 dBm on receive, What is it for Tx?

IIP3 is really only relevant to receivers. OIP3 is the relevant metric for a transmitter. I'm not too familiar with the WBX, but a receive IIP3 of 50 to 100 dBm is way to high. Conversely, your attached spreadsheet seems to imply that the actual IIP3 is on the order of -50 dBm, which is too low. I suspect you've got some unit conversion things going on. Why are your measurements in mV? dBm would be a more useful measurement.

You described your transmit test configuration. How are you testing receive?

2. Usually IIP3 and OIP3 are said to have more importance on the RX side, to measure non-linearity. Do IIP3 and OIP3 hold same significance on TX side? if not, what could be judged as a good measure of non linearity for TX side.

Are you comfortable with the concept of IIP3/OIP3? IIP3 isn't really relevant to transmitters, but OIP3 is a relevant measure. There are a whole handful of relevant transmitter metrics that have been defined. What you use depends on the system that you're using.

 

3. In this test, When I increase the amplitude of the signals, o/p varies. But for SBX the Inter-mods appear at 250mV while for WBX it is ard 550mv, and here the plot is very unsteady, and the NF is seen as curving.
however if I do the test with attenuation as 30 dB before the spectrum analyzer, The plot has a steady NF. Also after a certain point, I see a lot of spurious signals. Is there a specific amplitude range I should constraint myself to, to get the IIP3 and OIP3 points

I tend to set up measurements of IP3 so that the interfering signals (the IP3 tones at +/- deltaF) are 6 to 10 dB above the measurement noise floor. That ensures that the IP5 products aren't also being generated. If you get close to the point that the receiver or transmtter is going into compression, then your IP3 measurements are going to be masked by other nonlinearities.

Dave
 

Thanks

Gayathri

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio