• State Resolved
  • Locked Locked
  • Replies 1 reply
  • Subscribers 9 subscribers
  • Views 602 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TRF3703 noise floor measurement

Robert Murano
Intellectual 1275 points
Other Parts Discussed in Thread: DAC5687, DAC5688, DAC5682Z

I am having trouble measuring the noise floor on the TRF3703-17. I have measured at several frequencies with similar results, the results I give here at 2GHz are typical. Whereas the data sheet indicates -163dBm/Hz, I am getting a broad peak at 5MHz offset from the carrier of -155dBm/Hz, the rest from 1MHz to 20MHz offset being around -157dBm/Hz.

  • 0
    TI__Intellectual 1275 points

    As you are finding out, the noise measurement is one of the most difficult parameters to measure accurately.  The noise performance reported in the datasheet is a measure of the noise generated by the device itself.  Contributions from the LO source, signal source, and measurement equipment have been eliminated. 

    The enclosed presentation illustrates the technique used to measure the TX noise using a spectrum analyzer like the Agilent PSA.  Slide 3 shows a block diagram of the set-up.  For the datasheet measurements we use the same set-up except that we use the Agilent E5052 Phase Noise Analyzer as the spectrum analyzer.  The E5052 has a little bit better dynamic range than the pure spectrum analyzers.

    I LO power and common mode voltage will not be a big knob for the noise.  Even with a very low noise source, we still filter it.  This will have a little bit of an impact to the overall performance.  Alternatively to filtering, you can independently measure the phase noise of the source and mathematically remove its contribution. 

    It seems that you were not injecting a signal into the device.  The output signal was only the carrier feedthrough.  When using the E5052 we had to have a signal for the analyzer to lock on to.  We used sharp narrow band filters to pass the signal but to reject the noise from the generator.  If you are not using a generator then this may not be an issue as long as the noise from the power supply which is generating the common mode voltage is clean.  We use a specific power supply for the noise measurements that has low noise output.

    The dynamic range of the spectrum analyzer is the other key.  It seems like you are doing all the right things in setting up the spectrum analyzer with the correct dynamic range and with the pre-amp.  There will still be noise contributions from the Spectrum analyzer that can be measured and mathematically eliminated.  The enclosed presentation may illustrate some of the things that I have discussed. 

    But the real proof is how it behaves with modulated signals.  You can check out an application note found at: http://focus.ti.com/analog/docs/techdocsabstract.tsp?familyId=632&abstractName=slwa055

    This discusses the way a single carrier WCDMA measurement was taken and the superior results possible with the TRF370x product family.  The modulated measurements show a real world case where the source is from a DAC5687 device and the LO signal is unfiltered. Newer DACs such as the DAC5688 and DAC5682Z would offer even better performance.

    Robert

    TRF3703OutputNoise.pdf