NI-USRP Fetch RX Data Measurement Units (V?)

[beck.278]

Hey guys,

I am using the VI "NI-USRP Fetch RX Data" to obtains IQ samples from my radio.  I have it set to the Complex Double instance (CDB).  Does anyone know what the measurement units are for the samples at the output?  ie, is it in Volts, mV, dBm etc?

Furthermore, would it be any different if I chose waveform data type? 

What I want to do is compute the average power in a set of samples, in dBm.  Obviously in order to do that I need to know the units that I'm starting with.  Thanks to anyone that can help!

-Brian

[Sarah_Y]

Hey Brian,

This function outputs volts (peak-peak).  If you change the data type, it will not change the units of the data.  It will change how the data is bundled together which can make a difference on what type of functions will accept the data. For example, the FFT power spectrum.vi requires the waveform data type.  

The niUSRP EX Rx Continuous Async.vi has a nice front panel that displays this.  You can see in the IQ graph, the amplitude displayed in volts.  The graph below shows the baseband power spectrum and the amplitude in dB.  This bottom graph was generated by sending the IQ data fetched by the USRP through the FFT power spectrum.vi, where the IQ graph shows the raw data.

Hope this helps.  Post back if you have any further questions!

[Stefan_Bieder]

Dear all,

I still have some problems of understanding. What do I have to do to calculate the instantaneous power of a sampled I/Q point using the niUSRP Fetch Rx Data.vi?

Analog to the niRFSA Fetch IQ.vi I tried to determine the instantaneous power in Watts by calculating the complex magnitude (I2 + Q2)/2R with R = 50 [Ω]. By summing up the instantaneous power divided by the number of samples it should return the total received power. To calculate the total received power in [dBm] I use the following expression
Total_Received_Power_dBm =  10 * log(Total_Received_Power_Watts / 1e-3)

To test my VI, I use an externa RF generator to input a pure CW signal with -30 dBm signal power.

Gain [dB] = 0 --> Total received power = -14,3 dBm

Gain [dB] = 10 --> Total received power = -5,7 dBm

Gain [dB] = 20 --> Total received power = 3,8 dBm

Can you please comment on this?

Best regards,

Stefan

[cochenob]

Stefan-

I think at one point I convinced myself that the units of I and Q from the radio are in Volts.  As Sarah mentioned, this doesn't change with data type (waveform, cluster, etc).

If you want to measure the power, you have two options...

1. Calculate the magnitude from I and Q as M = sqrt(I^2 + Q^2).  Note that unless you input a single tone, this will likely give you the entire power within the BW that you determine via the sample rate.  If you're just inputting a single tone, and there's no other spectral content in your band, then this will be the power of that single tone.  If you've got other junk in your band, this method might be less ideal.

2. If your input is a single tone...then you could also take an FFT, and then just look at the peak.  The "Spectrum VI" (or whatever it's called) that is included with most of the shipped examples gives you the power spectrum in terms of Vrms^2 or dB (NOTE: dB...NOT dBm!).  You can get to dBm from Vrms^2 via an easy calculation...but I got tired with that and just made my own VI that calculates the power spectrum and returns things in dBm (referenced to 50 ohms) or Vrms^2.  I attached that code here (Simple Two Sided Spectrum.vi).  It's basically a stripped down version of the Power Spectrum VI that's in the examples, but without a lot of fluff and in units that I usually want.

It also looks like you're seeing discrepencies between the power of the tone you put in and that which you measure.  While I am not entirely certain of the calculation that you're doing now...note that the USRP is not a calibrated device.  In other words, all of the gains and losses in the front end are not accounted for.  In my experience, there's about 23-ish dB of gain between the power of a tone that I put into the radio, and that which I calculate.  You seem to be seeing 15-16 dB, but I'm not entirely certain of your method as you describe it.

In any case...I also attached a simple program that I've used to help calibrate the magnitude response of my radios.  It's fairly similar to one of the shipped Rx examples, but uses my Power Spectrum vi that I mention above, reporting the power spectrum in dBm.  You can calibrate your device by measuring the difference between what you put in, and what this program measures.  It will change a bit depending on your BW (sample rate) and center frequency.

Good luck!

---

Brandon

[psisterhen]

The values returned from niUSRP Fetch Rx Data are not really in units of Volts.  It's better to think of them as units of the ADC codes.  Even that is tough because they are processed in the DSP chain and so don't correspond to the reference levels of the ADCs.

The important thing is to not consider them as a voltage.  That is why you will see the values increase if you increase the Gain (which increases the gain from the input terminal to the ADCs), but the signal at the input terminals doesn't change.

To relate these values to absolute units of V or dBm, you need to do a calibration.  For example, input a known -10 dBm signal on the connector, set the Gain to 15, and calculate a RMS value from your measured waveform.  Then you'll be able to relate the Fetch values to an actual signal power at the input terminal (for that particular Gain setting, and at that particular Frequency... it varies for both!).