Why can I not achieve datasheet additive(residual) phase noise performance at frequencies <350MHz with the E5052 phase noise analyzer?
In fairness to the E5052, we have seen the same issue discussed below with other vendor's phase noise analyzer's. However, since the E5052 has been the gold standard in phase noise measurements for many years, the questions we receive on this topic often revolve around the E5052 (simply due to its popularity).
Additive vs Absolute Phase Noise (E5505 vs E5052)
The LTC695x additive phase noise plots were created using the E5505 phase noise measurement system. This instrument was designed specifically for additive phase noise measurements. By definition, additive phase noise only measures the phase noise of the device under test (DUT). Additive phase noise does not include the phase noise of the source that is driving the DUT. Conversely, The E5052 measures absolute phase noise. Absolute phase noise measurements include the total phase noise of the DUT plus the source that is driving the DUT. One reason the E5052 may not reproduce the E5505 datasheet plots is the difference in measurement techniques (additive vs absolute). However, in theory, if the source's absolute phase noise is 6dB lower then the DUTs additive phase noise, then the source's phase noise contributions to the absolute phase noise measurement should be minimal (<1dB error). Under this condition the E5052 results should be within 1dB of the E5505 additive phase noise datasheet plots. Unfortunately, theory does not hold when measuring the LTC695x devices with E5052 for frequencies <350M. Below describes a solution to overcome this issue.
Amplitude to Phase noise Conversion <350MHz (Measurement Error)
The E5052 is an excellent instrument and has been the standard in phase noise measurements for many years, but at frequencies less than 350MHz the E5052 suffers from some amplitude noise to phase noise conversion. This amplitude noise to phase noise conversion is seen when measuring the LTC695x products at low frequencies. To correct the amplitude to phase noise conversion error we recommend using a phase discriminator circuit in the measurement path between the LTC695x and the E5052. The discriminator circuit shown below has worked well for us from 20MHz to 350MHz.
The basic concept of this circuit is to increase the signals slew rate so that the E5052 has less chance for amplitude noise to affect the phase noise measurement. The circuit above has 2 stages that apply gain followed by diodes. The diodes maintain the increased slew generated from each gain stage and the diodes clip the amplitude at the output of each amplifier to avoid sending the 2nd stage amp into compression and exceeding the E5052s max input amplitude specifications.
Below are some measurements with and without the above phase discriminator.
For more on additive phase noise