The source signal is a voltage sine wave at the frequency being measured. It is typically between 1 mW and 100 mW, so, because it is usually in 50 ohms, the peak voltage could be between sqrt(10) V and 1/sqrt(10) V. The VNA steps through the frequency range you specify, using the number of points you specify (for instance 201 points), measuring at each frequency.

Thank you. I have measured from 27 GHz to 40 GHz. With 401 points. Could please explain me how to obtain the source signal curve ?

regards

There will be 401 different source signals. Each is a sine wave. Each will last long enough to get an accurate measurement with the IF filtering, averaging and smoothing that you are using. Find out what the sweep time is on the VNA. It will stay at each frequency for 1/401 of the sweep time. The frequency will change by (40-27)/(401-1) GHz after each measurement.

So the sweep time is 10.947 msec. Calculating the Frequency change (40-27)/(401-1) = 0.0325 GHz. Also, I checked the sweep type and it says linear frequency. So, Based on this how can I produce the sin wave source signal?

Could you help me, please?@Malcom White

Sebastian Tamayo you can use SCPI command to read the VNA parameters directly through its interface. [You can use the commands to set parameters also.]

??? pure monochromatic low-noise sine signal is in the basis of all VNAs. OK, first network analyzers were based on tube electronics... but still, no matter what components you base the analyzer on, it comes down to interconnect of PLL circuits, and mixers. If you want to know more about your PNA unit, here is the service manual: https://www.testworld.com/wp-content/uploads/service-guide-keysight-agilent-n5230a-4-port-pna-l-microwave-network-analyzers.pdf Section 5-6.

About SCPI: HP/Agilent analyzers allow directly reading out binary buffers for S-parameters, and even non-calibrated voltage readings for software-processing. It is a blazing fast operation, but it requires some binary buffer management in your software, which may be compicated in high-lever research software like LabView... or Excel.. .yes Excell's VBA is very useful for quick-coding graphical interface to VNA's, but it lacks many native binary operations (use native OS Win32 calls), and interrupt control. Interrupt handling is very desirable when dealing with Agilent VNA's, since state registers and ready polls overload the bus and may fail (VNAs and PNAs may simply freeze up until reset manually).. And still, there are non-native tricks to call VBA functions when the instrument issues a ready interrupt.

The same as reading, you can load whatever sweep plan you like into your vna in binary or text. Again, loading preconstructed binary plan is much-much faster than reading/writing formatted strings. And actually easier, since you don't have to deal with delimiters, line-ends, value suffixes (i.e. Mega, Giga...)... just convert and multiply with your predefined scale explicitly.

Could you help me, please?@Malcom White

The input signal is a sine wave at 27 GHz lasting 10.947 msec. This is followed by a sinewave at 27.0325 GHz for 10.947 msec, followed by all the other steps until the last 10.974 msec at 40 GHz.

What is it that you need to know? Do you know what a sine wave is? Do you need to find out how to generate a sine wave using the language of your simulation code?

Thank you.

Yes, I know the sin wave but what are the values of the amplitude? I understand that at every frequency there is a sin wave, but with what values? I would like to simulate the signal source as an input to a Horn Antenna in CST software.

In my first reply I said "The source signal is a voltage sine wave at the frequency being measured. It is typically between 1 mW and 100 mW, so, because it is usually in 50 ohms, the peak voltage could be between sqrt(10) V and 1/sqrt(10) V."

The V|NA will tell you the power output level you have chosen. Do you know how to calculate the peak voltage from the power level and the impedance? The impedance will be 50 ohms in the analyser. If you want the voltage in the horn, that is a lot harder, because it depends on the impedance which varies with frequency and distance along the horn, and the voltage varies across the width of the horn.

It might be better to define the power rather than the voltage. The power hardly changes from the inside of the VNA to the mouth of the horn.

I expect you will not be able to simulate the VNA signal on CST because 10.947 msec is about 100 million cycles and I don't think CST will manage that many in reasonable time, or before timing out. In the time domain it will do about 10 steps per cycle, so that is a billion steps. In the time domain or frequency domain it is not even sensible to include the cable from the VNA to the Horn because that adds extra calculations to each timestep and means a longer time before the result is stable. In the frequency domain, solving for the cable as well will increase the required memory a lot.

You may be able to just run CST until it gives you a converged result. That will be the same result that the VNA would get in 10.947 msec. In the time domain one run will give you S11 at all the frequencies you need in one go.

In addition to all the previous answers, it's important to remember that a VNA is a "four-receiver" instrument, where a single source is monitored using four receivers to construct the two-port s-parameters matrix. The significance of this is that most VNAs will have variations in their signal generator, which you can actually observe if you disable the calibration, and the "accurate" response is only obtainable through the two-port four-term matrix after applying the error correction based on the calibration. Using a waveport (properly setup) in CST is similar to performing the two-port measurements with the VNA at the calibration plane. Adding the coaxial connector (you can find 3D models online) will mimic the setup better as the calibration plane varies based on the kit you're using.

As for CST's TD analysis, the main difference between that and the VNA measurement is the simulation duration, in CST the TD solver terminates after reaching sufficient stability in the s-parameters. If your VNA includes a time-domain option, you may want to investigate the effect of time-gating your s-parameters, which will resemble the CST simulation, if you set the time window to the same simulation duration to CST's.

exactly. And technically there is no requirement to keeping the amplitude of the source strictly constant. It may vary with frequency, and may fall down when going to the higher band. But it should lie inside the normal range defined by the manufacturer, which is checked and fixed during annual maintenance by a certified metrology lab. The same goes to the noise floor, which is more important. Some instruments, particularly made in Russia, allow user-level direct tuning of the output power.

The calibration and what is basically homodyne mixing (perhaps in several stages, and perhaps including digital sampling) in the VNA mean that the calibrated signal displayed by a VNA is close to the ideal signal from a code such as CST, if it launched and measured at the calibration point.