Answering some questions ...




In this page you will find some answers to some questions. My intention is to answer all the questions that arrive to the email of this site, which is, by the way,

 csc.rf.electronics@gmail.com

Since the foreseen number of questions that will arrive to the email will not be, for sure, inmense, I think I will be able to answer some of then. More precisely, those ones that I know what to answer for.

I don't know if the number of these questions will exceed the million, (may be not, who knows ...) ,but I will try my best to answer all of them.

Even I will try to solve those doubts that I do not have answer for. Make no mistake, there will be some ... (not many, I hope ...). It is for sure that I will learn more from those ones, than the ones that I know the answer.
 
Let's say that I will try to answer as many as I can.

And to start the answering section let's present the first one that I ever had to answer regarding QUCS Studio.   

It is one regarding a frequency dependant attenuator based in a frequency response.
The question was phrased in the following way: (something like that)
"I would like to model a device which I know the frequency response from. Only the magnitude of this response. Nothing about the phase".

  1. - Question about how to model a frequency dependant attenuator, once you know its frequency response. The answer that I proposed was the following:

How to design a frequency dependant attenuator in QUCS Studio

      2 .- Question about how to model a commplex rectangular train of pulses:

Test file of a Complex Rectangular Voltage source

    3.- Question about convergence problems during harmonic balance simulation. The following document summarizes some good practices and several guidelines oriented to enhance the convergence of the harmonic balance analysis in QUCS Studio.

Convergence problems in the Harmonic Balance using QUCS Studio
 
The files used in the sumulations included in the previous file are the following:

 Convergence problems QUCS Studio files

22 comments:

  1. Hi! I found this project interesting, but I have a problem with the simulation in Win7.
    Elementary schema with 1 sine source and load resistor in trancient analysys produce no graph. Examples from distributive do not work either. In WinXP - all these examples work fine. What a bag, and how can I fix it?

    ReplyDelete
    Replies
    1. Hello:
      First of all, my apologies for answering so late. But I have been working in examples that recreate the scenario you are telling me, and I have found no way of reproducing it. It is true that I run the program in different operating systems, but, till the day, no way of getting more information with your first description of the problem.
      Can you provide me a little bit more of info, regarding this issue?
      Again, my apologies for answering so late.

      Delete
  2. I will answer myself. Simulation does not work due to the presence of Cyrillic letters on the way to the project address.
    How to change the path to the project address in Qucsstudio?
    In Qucs, this can be done through Application Settings.
    In Qucsstudio, I did not find such an option.

    ReplyDelete
    Replies
    1. Wow !! You have been really fast. Sadly, faster than me. And the Cyrillic issue is not an easy one, because is quite local. Thank you very much for your question and "self" answer.
      It is true what you say. In QUCS is quite easy, inside the "Application Settings", and then in the "Location" tag, the directory where the projects are stored can be changed. But this not work in QUCS Studio, because de "Location" tag is not available in the same place yet. May be in the following versions ...

      Sorry. Next time, I will try to be faster.

      Delete
  3. I found that the project path can be changed on the command line.
    Is there a key for storing these paths for future use? For example:
    D: \ qucsstudio \ qucs.exe - e: \ qucsstudio ... additional key

    ReplyDelete
    Replies
    1. Thank you very much for sharing this. I was not aware of this possibility. I think It is a good time to consult the software designer and to get the best answer. I will try to get an answer as soon as possible.
      Thank you again.

      Delete
    2. Here we have an answer regarding by Michael Margraff regarding the issue of storing a new project path in QUCS Studio permanently.
      The best way is to create a link on the desktop, i.e. drag'n'drop the
      .\QucsStudio\bin\qucs.exe file to the desktop by holding the Alt key pressed.
      In the properties of this link file, the command line can be stored, and
      double-click on it will start QucsStudio with the wanted project folder.

      Delete
  4. Replies
    1. Wow ... This is embarrassing. I did not try my self the recipe. Let me dig into it a little bit more. I hope I get an answer quickly .. a working one this time ...

      Delete
  5. Sorry to bother you again.
    How can I fix the error in Rectangle Voltage Source? The Tl parameter acts as a period, and the Th parameter does not respond. Source produces only meander with duty cycle 50%. :(
    In QUCS - all OK.

    ReplyDelete
  6. Hello. You don't bother me at all. Yes, the ideal rectangle voltage source seems to be flawed in QUCS Studio. In my experience, the problem of this source is that the duration of the pulse in high level is irrelevant. You can edit this parameter and put whatever you want, and the outcome of the signal is only dependant of the duration of the pulse in low level. And always with a duty cicle of 50%.
    Yes it is a bug in the definition of the voltage source and it has been already reported to the designer.

    ReplyDelete
    Replies
    1. Hello again:

      Have your tried to use the file based voltage source? You can define one entire period of the signal and then, indicate that the signal must be repeated in time. With only a few lines you can have your rectangular signal.
      In this very page, above of these questions, look at the point 2, where I am going to put a link to an example of this idea I have been explaining.

      Delete
  7. Hello! Thanks for your answer.
    I know about file based voltage source but I just used QUCS to solve my problem.

    ReplyDelete
  8. Thanks to you for your question. Next time I hope you will find the solution before.

    ReplyDelete
  9. This was a question addressed to the email csc.rf.electronics@gmail.com

    Hi,

    When plotting sinusoidal power results of a mixer model in dBm, I noticed the format from an example on the web is dBm(IF.Vb,1e6). What does "1e6" represent?

    Is there a document with a list of commands and their formats?

    Thanks for your help,

    ReplyDelete
    Replies
    1. Hello:
      Thank you very much for your question. And thank you for sending it to the email. It is the fastest way for contacting me.
      It would be very helpful that you told me what is exactly the example file that contains this expression. But, normally this value references a frequency. In your case, 1 MHz.
      You can check this if you open the "symmetricMixer.sch" file from the example project "Simulation_HarmonicBalance". It is also available as part of the QUCS Studio helping projects.
      In this file, it is analyzed a single-symmetrical diode mixer. In the results window, you will find two diagrams. One of them plots the Input reflection coefficient. In this one, the IF signal is used to calculate the Input reflection coefficient and as a mathematical term of this Input reflection coefficient, it is used the expression (RF.Vb, 1.001e9). If you open the source device that feeds the circuit, you will see that the frequency set as a paremeter is 1001 MHz, which is, exactly, 1.001e9 Hz.
      I think that your case is the same.
      Regarding the list of commands you will find all of them in the help file. In the main window of QUCS Studio, go to "Help" and click over " Help index" or press F1. A help window will pop up and all the different commands and expressions allowed in QUCS Studio are included. Of course, the files contained in the help projects available in the QUCS Studio website are very interesting because they will teach you how to use these commands in a very advanced manner some of them.
      In my case, when I press the F1 buttom for opening the help window, it does not work. One error window pops up, saying that it is not possible to find a file. I think that this is some sort of bug in the program, and I have to report it to the author of the software. Meanwhile, I use the PDF file that I have attached to this email. It is the printed version of the list of commands that I got from a previous version of QUCS Studio, to cope the problem until a new edition of the software that solves the problem.
      Best regards.

      Delete
  10. This question came to the email of the website csc.rf.electronics@gmail.com

    Hi,

    The first attached file shows a simulation with two double-balanced mixers connected in series. (https://drive.google.com/open?id=0BzxO_nDyVboGZUpmelpQLVFKUFhPYlpWN3E3N3FkXzM0TEVZ)

    The simulation does not resolve, and seems to get stuck in a loop with a warning message - see the second attachment.
    (https://drive.google.com/open?id=0BzxO_nDyVboGVW02S19KYmlNOEZCejlPdDhSYnlUeU1fRmZV)

    The simulation ran just fine with one mixer, and results were as expected, but after adding the second mixer, the problem developed.

    Any help would be appreciated,

    ReplyDelete
    Replies
    1. Jose Manuel Campelo Ortiz

      11 sept. 2019 8:56 (hace 10 días)

      para Steve

      Hi again, Steve:


      Convergence problems in harmonic balance analysis are a classic. I think every single human being involved in this stuff have struggled with them in many occasions. You will find plenty of technical papers talking about it. But most of them describe the issue in terms of the mathematical algorithm that is used to build the harmonic balance analysis in the CAD simulators. Not many of them help indeed to understand what can be done once the simulation has failed to converge.


      As the technical literature says, the concrete actions to be taken to improve the problem of the convergence would be the following:


      Analyzing the amplitude of the signals that are introduced as the excitations to the circuit under analysis. Large amplitudes of these signals will go against the convergence of the harmonic balance. Normally, this is something that is fixed by the requirements, and, therefore, this line of action is normally useless.
      Taking into account that, even when the excitation signal is composed only by one frequency tone, due to the non linearities of the elements included in the circuit under analysis, the number of significant frequency components (harmonics) that must be analyzed in the harmonic balance can be numerous. If the number of harmonics defined is not enough, the currents and voltages calculated by the algorithm will not satisfy the Kirchoff’s laws and will provoke the failure in the convergence.
      The models of the components in the circuit under analysis can be the source of the problem. Errors in the definition of the one non – linear element models can provoke very large errors in the electrical signals obtained, mainly in currents, around the circuit. Those errors will lead to non - convergence results during the evaluation of the Kirchoff’s laws in the solution process.

      It is also probable that the component models contains mathematical discontinuities But this would be the most difficult parameter to analyze, because, the models of the components, normally, are obtained from the manufacturer download website, and there is no room to improve anything.

      Delete
    2. So, what can be done in order to improve the convergence in QUCSStudio?


      Since your question of the convergence is not the first one that I have received, I have decided to write a note as a new entry of the blog. I will publish it in a day or two. I encourage you to read it. It will include examples of non linear circuits analyzed with harmonic balance.
      In any case, as main guidelines you can test the following:
      Regarding the amplitude. Very few things to do in that front. Reduce the amplitude of the input signals as much as your requirements allow. And that is all.
      Regarding the number of harmonics considered in the analysis. This is the best way of resolving convergence issues. Normally, increasing the number of harmonics enhace the convergence. And the increase of this number can be really huge.
      Regarding the diodes electrical model. It is something close. The model is what it is, and there almost nothing to do about it.

      But, finally, do not think that you will be always able to analyze every single circuit using harmonic balance. That will not happen. In many cases, the convergence problems will be unsolvable, because the non - linearities present in the circuit are so numerous that the analysis will be almost impossible resolve.
      In the case of the mixer that you sent me, I have not been able to obtain just a single solution, (even it it is wrong due to lack of accuracy or whatever) from the harmonic balance.
      I have used this example in the note that I have mentioned to you, as an example of one circuit that need another strategy to be simulated. In this case, I have analyzed the response of the mixer using the transient analysis. So far, it is the only way that I have obtained something useful. That is the approximation that I suggest to you at this point. I am trying to get something better, but it seems that this is one of those times when trying another way is actually the only way.


      You will also find this analysis in the blog entry that will be available in a couple of days.

      Regards.

      Delete
  11. BFR520.sch missing from files. I tried using the BFU590.sch from another project but could not get anything to converge.

    ReplyDelete
  12. Where exactly is this file BFR520.sch missing?
    Is it the tutorial about the design of a VCO, or the one dedicated to the L Band Amplifier?
    Let me know it and I will fix it as soon as possible.

    ReplyDelete
  13. I am afraid, this is not your blog ... Call me insane ... but I am pretty sure that you need other type of website ...

    ReplyDelete