Where To Find This Example
Select Help > Open Examples... from the menus and type either the example name listed above or one of the keywords below.
Or in Version 13 or higher you can open the project directly from this page using this button. Make sure to select the Enable Guided Help before clicking this button.
RF Blocks with Nonlinear Noise
This example shows the use of the general RF block elements in conjunction with the nonlinear noise capabilities in AWR. The examples progress from a simple noisy amplifier through an example with both amplifiers and mixers. You can simply open the indicated schematics - the appropriate graphs will be embedded in the schematic window. For more detail, any graph can be opened by itself from the project browser.
The simplest case is a single noisy amplifier. Please refer to the schematic called "1_Noisy_Amp" and the graph called "1 Noisy Amp". Here, we have taken the amplifier and dropped it into a 50 Ω system. The frequency is fixed at 4GHz and the power is swept from -20dBm to +10dBm. You can see the gain and power behave as expected. Note the NF in the linear region. NFmin is set at 1.6dB but we have not matched for optimal NF, so the simulated NF is higher than NFmin.
The nonlinear noise analysis element (NLNOISE) is set to analyze noise at 0.1GHz. Note that this frequency is offset from the tone frequency.
Next, we introduce an ideal matching element by using the LTUNER element. LTUNER is primarily used for Load Pull measurements but here we use it to transform the amp's Gamma(opt) to 50 Ω. Remove the wire shorting the tuner and toggle the tuner on. Running the analysis now shows that the NF has been reduced to NFmin, which is what we expect. Note that this technique is sensitive to S12. If S12 is non-zero, then the amp is not unilateral and NFmin will not be achieved with the ideal noise match.
Linear Noise Contours
The previous example can be extended to produce noise contours as shown in the schematic called "2_Noisy_Amp_Linear" and the graph called "2 Noise Contours". The Load Pull Wizard is used to perform an impedance plane sweep based on linear NF which is measured in the "2 Noisy Amp_Linear LP" graph. Then, the contours are plotted as shown in "2 Noise Contours". Note that the contours are centered about Gamma(opt), which has a magnitude of 0.38 and an angle of 54°. This all goes to illustrate the fact that the nonlinear models, such as NL_AMP2, can properly compute linearized noise results.
This case is two amplifiers cascaded. Please refer to the circuit called "3_Cascaded_Amps" and the graph called "3 Cascaded Amps".
The amplifiers are represented by the data file "MGA_86576 5V", which represents an internally matched MMIC amplifier biased at 5V. The nonlinear characteristics of the first amplifier have been arbitrarily set and the nonlinear characteristics of the second amplifier have been set 10dB higher than the first one.
The nonlinear noise analysis element (NLNOISE) has the start/stop frequencies for noise analysis set to 3.9GHz and 4.1GHz respectively. Since the fundamental is 4GHz for this circuit, we are including all noise components.
The nominal analysis produces small-signal results with an RF drive of -50dBm. Each amplifier has |S21| = 14.49 = 23.1dB and the total system gain is calculated as 46.2dB. Each amplifier has NFmin = 1.6dB and a Gamma(opt) taken from the data file. One of the significant capabilities in the harmonic balance simulator is the proper handling of mismatch in nonlinear noise calculations. Overall NF is 1.96dB, which follows from the cascaded NF equation and mismatch.
(NOTE: the pure cascaded NF equation:
F = F1 + ((F2 - 1) / G1)
would predict system NF of 1.626dB, so it is apparent that mismatch is being accounted for in the calculation.)
To complete this example, tune the input power of this circuit and watch gain and NF change. As the first amplifier begins to compress, its gain drops, which affects the cascaded noise results. At -25dBm input, the gain has dropped to 42.5dB and the NF has risen to 4.54dB.
Power Analysis of Cascaded Amps
This is a simple cascade of two identical amplifiers and they are driven by a power sweep from -40dBm to +10dBm. Please refer to the schematic named "4_Power_Analysis_Cascaded_Amp" and graph named "4 Power Analysis Cascaded Amp". The red trace shows the fundamental output power at port 2, which is a standard measurement. The blue trace, however, shows the fundamental output power at the node between the amplifiers. When setting up this measurement, the "Measurement Component" box has a button next to it with a rectangle (...). Clicking that button produces a dialog that allows you to select any node in the circuit. The notation is “AM1@2”, which indicates amplifier AM1 at node 2. The same curve could have been produced by selecting “AM2@1”, since that represents the same node in the circuit.
The results show that the first amp goes into compression as indicated by its nonlinear parameters. Since the second amp is being driven by the first amp, its compression point relative to the input signal (which is the X axis) is much lower (by the amount of gain that amp 1 has, 23dB).
Cascaded Amplifiers with Filter
The next case is the same basic circuit with the addition of a filter on the input, schematic "5_Cascaded_Amps with_Filter" and the graph is "5 Cascaded Amps with Filter”. This will allow us to look at the response over frequency, which has been set to a sweep from 1 - 6GHz in this case. The filter is a 3 pole Chebyshev band-pass with a pass-band of 3 - 4GHz. All other parameters remain the same, except the NLNOISE frequency parameters, which are now set to cover DC - 6GHz to reflect the RF sweep.
As expected, the filter characteristics have a direct impact on both gain and NF. The in-band gain is still in the 45dB range but the ripple is exaggerated by the additional mismatch introduced by the filter (compare the ripple of the filter alone with the ripple in the system gain). The losses associated with the filter roll-off translate directly to the NF out of band, since there is a one-to-one relationship between front-end loss and NF. In-band noise is still nominally in the 2dB range.
Once again, changing the input power for this circuit to -25dBm illustrates the effects of the cascaded noise equation. The gain drops and the NF increases as the first amplifier begins to compress.
Cascaded Amps with Mixer
The final case consists of the two amplifiers and a mixer, the schematic "6_Cascaded_Amps_with_Mixer" and the graphs are "6 Power Spectrum Cascaded Amp with Mixer", "6 Waveforms Cascaded Amps with Mixer", and "6 NF and Gain Cascaded Amps with Mixer".
The amplifiers are the same as the first two cases. The mixer has typical performance characteristics for a double balanced mixer.
The NLNOISE element in this case needs some adjustments, The MWO nonlinear noise calculations will properly account for noise that is frequency translated from RF sidebands to the IF band. In order to do this, the NLNOISE element needs to be told about one (and only one) of the sidebands. In this case, we have elected to specify the sideband from 3.9GHz to 4.5GHz. This decision was made after consulting the "6 Power Spectrum Cascaded Amp with Mixer" graph and choosing the upper sideband on the fundamental cluster of signals. By specifying this one sideband, the noise calculation will properly handle all sidebands and their translation to the IF band. LO signal is chosen as Large Signal (ie specifying "1" for "LSTone" parameter as LO is set as Tone 1) for noise analysis.
The results of this analysis show the system gain is 40.25dB. This is consistent with the amplifier gain seen earlier along with the mixer conversion loss. The NF is 1.97dB, which is only slightly higher than the amplifiers alone. This result is expected because the cascaded noise equation does not weight downstream noise contributions heavily. As before, we can tune the input power and see the expected changes in gain and NF as the first amplifier compresses.
As a final exercise, we can look at the "6 Waveforms Cascaded Amps with Mixer" graph and note the obvious presence of the LO on the IF output. If we activate the output filter and remove the wire that shorts it out, we can make a significant impact on the waveform's LO content.