Average vs. Peak Symbol Power

[cochenob]

I have an application where I require that the RF power I define on the front panel be recognized as the peak symbol power of a QAM signal set.  I have come to realize that by default, the RF power you enter on the front panel is the average power. 

After some investigation, I've come to realize that you can define this via a property node of the RF Configure VI.  Unfortunately, my attempts to change this property throws an error at the Initiate VI.  I am using the Generate QAM example program shipped with the device. 

Is this the correct way in going about doing this? 

---
Brandon

[Mauricio Capistran-Garza]

Hi Brandon,

Could you please post the error number that you are getting? One thing that you have to be careful about when setting the Power Level Type to "Peak Power" is to make sure that the instantaneous magnitude of your signal (which is calculated as the square root of I^2+Q^2) never exceeds 1.  When the device is configured to "Average Power" then the driver automatically scales the data as necessary to meet that constraint, however, when configured to "Peak Power" you must do that yourself.

- Mauricio

[cochenob]

Mauricio-

You are correct.  This is the error I am receiving.  However, I am a bit confused as to why the magnitude is ever greater than 1.  When the modulation parameters are generated, this always seems to be the case automatically.  Furthermore, the generate parameters VI accomplishes this without any knowledge of what the RF power is.  I have verified this, for example in the QAM transmitter example, by probing the QAM parameters and verifying that the symbol magnitudes are <1.

However, I did discover something interesting.  I am not currently using any baseband filtering.  However, if I choose the Root Raised Cosine filter,  things work fine.

This partially makes sense in that filtering the original IQ pairs changes their value....however....I've verified that the symbols that are originally generated have a magnitude <1...so if I don't use any baseband filter, the should be the same.  Why do they turn out >1?

---

Brandon

[Andy_Hinde]

Hi Brandon,

Can you post your VI so we can look at it?

Regards,

Andy Hinde

National instruments

[cochenob]

I am just using the QAM trasmitter example that was shipped with the toolkit.  I haven't done any extra development, with the acception of adding another field to the property node to define "average" or "peak" power.  If you feel that there might be an issue (old version, etc.) I will attach it to another message (it's on a different machine right now).

[Andy_Hinde]

Hi Brandon,

Can you confirm you're using 'MT RFSG Generate QAM.vi' located at:

C:\Program Files\National Instruments\LabVIEW 8.2\examples\Modulation\RFSG examples

Andy

[cochenob]

Sorry...yes...that's the one!

[Andy_Hinde]

Hi Brandon,
Are you using a PXI-5670 or PXI-5671 (i.e. is your arb a PXI-5421 or a PXI-5441)?

Thanks,
Andy

[Andy_Hinde]

Hi Brandon,
In peak power mode the NI-RFSG driver software and PXI arb hardware expect IQ sample values with magnitudes no greater than 1.0. In average power mode, you can use values greater than 1.0 as the driver scales them internally.

The difficult part here is the Resample and Write utility which resamples the data and writes it to the arb in chunks. The resample process changes the magnitudes of the IQ samples slightly (especially when no pulse shaping is used) so even if you normalize the IQ magnitudes prior to giving the data to Resample and Write (R&W), the R&W routine slightly changes them in some cases enough to push the magnitude above 1.0 for some samples.

The attached VI works around this by resampling the waveform prior to the R&W VI. Now that the waveform is resampled, it can be scaled with some included utilities prior to the R&W VI. Since the data is already resampled, the R&W VI will only be writing the data to the arb and won't manipulate the magnitudes at all.

Hope this helps!

Regards,
Andy Hinde
National Instruments

[Andy_Hinde]

Hi Brandon,
In peak power mode the NI-RFSG driver software and PXI arb hardware expect IQ sample values with magnitudes no greater than 1.0. In average power mode, you can use values greater than 1.0 as the driver scales them internally.

The difficult part here is the Resample and Write utility which resamples the data and writes it to the arb in chunks. The resample process changes the magnitudes of the IQ samples slightly (especially when no pulse shaping is used) so even if you normalize the IQ magnitudes prior to giving the data to Resample and Write (R&W), the R&W routine slightly changes them in some cases enough to push the magnitude above 1.0 for some samples.

The attached VI works around this by resampling the waveform prior to the R&W VI. Now that the waveform is resampled, it can be scaled with some included utilities prior to the R&W VI. Since the data is already resampled, the R&W VI will only be writing the data to the arb and won't manipulate the magnitudes at all.

Hope this helps!

Regards,
Andy Hinde
National Instruments

Message Edited by Andy Hinde on 07-17-2007 02:51 PM