TRF371125 Baseband EVM vs different frequency

Leo:

Enclosed is a measurement report on the TRF371125 EVM (Error Vector Magnitude).  This data is measured on the -25 device around 2400 MHz; however, the performance will be similar with the -09 device operating around 900 MHz.

8463.TRF3711_Evaluation_CustPresV3.ppt

Here are some things to consider in your approach:

• The gain of the device changes with respect to frequency: higher gain at low frequency and lower gain at high frequency.  Higher gain generally translates to degraded linearity and improved NF.  Lower gain is the opposite.  EVM% is definitely impacted by IP3 performance and NF performance so there is a "balance" to get optimum performance.  It may make sense to shift to the -09 device for the 750 MHz and 800 MHz bands and keep the -25 device for the 1.8 GHz and 2.6 GHz bands.
• Choice of RF balun does impact performance.  I have found that the wideband balun/transformers do not actually work very well because they introduce a lot of gain and phase mismatch.  Typically, there is one small range of frequencies where it will look godd and the rest of the performance is poor.  I recommend using the more specialized, narrow band baluns as outlined in the datasheet.  I have measured these devices and found good gain and phase balance.  The mismatch will impact IIP2 performance of the device most severly and will have some impact on the IIP3 performance.  Excessive loss will directly impact NF which degrades EVM.
• Phase noise is a significant contributor to the EVM% performance.  The TRF3765 phase noise performance will definitely impact EVM% performance.  I would expect this would have more of an impact at higher frequency where phase noise performance is degreaded (interestingly you do not quite see this).  Try injecting a known good LO source from test equipment to gauge the impact of phase noise on your set-up.
• Verify that you are loading the TRF3711 in high impedance.  If you are injecting into the IQ input ports of the MXA then make sure the input impdedance of the instruement is set to 1 kohm or high impedance.  Measuring EVM through the on-board transformers (for 50 ohm load) is not recommended as the baluns will introduce variations over frequency and do not pass DC.
• Verify that the gain setting is optimized for each band.  The gain changes with respect to frequency so you will need a lower setting for the lower frequency bands.  As mentioned previously, the 750, 800 MHz band likely need the -09 device for best performance.  Also verify that the automated DC offset correction is initiated for each different frequency band.

--Russell