Rank: Advanced Member Groups: Registered
Joined: 21/11/2017(UTC) Posts: 114 Location: Olt county Was thanked: 1 time(s) in 1 post(s)
|
Greetings Thanks CBG for interest and involvement. As I have already said, a domain-defined function is not useful, since these domains can not be specified through the script that puts this function into a specific simulation program, such as Multisim - for example. For this reason, a single function must be defined to be continuous and derivative on the intervals -4 <Ug <0 and 0 <Ua <375. Otherwise, all the best. Edited by user 03 June 2018 09:45:08(UTC)
| Reason: Not specified
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: ola_nicolas Thanks CBG for interest and involvement. As I have already said, a domain-defined function is not useful, since these domains can not be specified through the script that puts this function into a specific simulation program, such as Multisim - for example. For this reason, a single function must be defined to be continuous and derivative on the intervals -4 a domain-defined function is not useful, since these domains can not be specified through the script that puts this function into a specific simulation program, such as Multisim. If Multisim can't take a mathematical expression from Smath or else ... what can you do? Just recycle Multisim because as you say it is not a math tool, just a gadget. Maybe it can take the plot [te 9 plots in sequence] of your Tesla. 1. Those plots, you have them from Carlos and I offer from more compacted polynomial fit. 2. Those plots, you have them as from the "x*ln(1+exp(Ug/x)) , which basically the function you attempted in your Mathcad that failed miserably. 3. Both proposals [Carlos, Jean] are smooth, monotonous, continuously derivable/integrable. Don't neglect understanding the note: "The polynomial fit tends to fit as well the defective digitized data. Digitized data are inaccurate by themselves and more inaccurate from the french curve spline creator from not true analog experiment in the original Tesla lab work." In short: Smath recovers your TESLA [1959]. Multisim is left to be coded to read either maths or Smath graphs Genfit Minimize TESLA RECONCILIATION.sm (390kb) downloaded 16 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud In short: Smath recovers your TESLA [1959].
... please: refresh the previously attached document φ(x):=[1,x,x^2]
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud ... please: refresh the previously attached document ... refresh,There was an error in the collapsed "Recover the fits" Genfit Minimize TESLA RECONCILIATION.sm (391kb) downloaded 22 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Shall I drop dead @ $ 6110 Canadian Multisim Full ?
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 14/10/2015(UTC) Posts: 308
Was thanked: 77 time(s) in 58 post(s)
|
Thank you very much Jean for your accurate response. HOLA Ola_Nicolas!!! What I did, was simply find the equation that represents each curve of the given graph. The functions are continuous, but when representing a polynomial of 4 degree, they are unstable outside the specified ranges. On the other hand, doing a little research, I found that the curves of the TESLA graph were constructed with the equations of Scott Reynolds and Marshall Leachfon, or with the improved equations of Normal L. Koren. The plate current equation for triodes is IP = (EG + EP /� )^3/2 / kG1 for (EG + EP /� > = 0 = 0 otherwise. The plate current equation for pentodes is IP = (2 / kG1p) (EG + EG2 /� )^3/2 arctan (EP / kV for EG + EG2 /� > = 0 = 0 otherwise. � is the amplification factor Unfortunately, these equations also contain a decision. Here I leave the information link, which I hope will be of interest and help to you. http://www.normankoren.c...article.html#PlatecurvesHere is a small attempt to graph the curves with the formulas indicated above. Tesla_CBG_R1.sm (138kb) downloaded 18 time(s).Best Regards Carlos
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: CBG Thank you very much Jean for your accurate response. Hello Carlos, I read quick the paper. As it looks, the equation(s) apply to the circuits with all resistances ... etc. Some tips to optimize the project. Nicolas should be happy ... are gone fishing ... how big the fish ? A most interesting/educative project ... Thanks. Jean Tesla_CBG_R1 [optimized].sm (263kb) downloaded 23 time(s).
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 21/11/2017(UTC) Posts: 114 Location: Olt county Was thanked: 1 time(s) in 1 post(s)
|
Hello everybody.
I've seen your previous posts. I will keep them as such and I will respond as quickly as possible. But current life and activity is on the forefront. Electronics and math are just a hobby for me. With all due respect.
Nicolas
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 21/11/2017(UTC) Posts: 114 Location: Olt county Was thanked: 1 time(s) in 1 post(s)
|
Hello to everyone. I studied Norman Koren's articles - everything I found on the internet. I have also studied Eugen Karpov's articles, which seem to be the last to bring improvements to the regressions concerned. I tried myself, to make some modest contributions through an article published here: http://www.tehnium-azi.r...catodici-si-de-grila-r43If someone is interested, the article also contains a PDF file with the English translation. It does not bother me that you have modeled each curve on the basis of polynomial equations, but that these equations are dependent on a single variable. Assuming that there are regressions in the curves that include the second variable, these regressions could not be used without an interpolation protocol between them, so that they could be covered for the grid voltage a continuous range of real numbers. The model presented by Jean Giraud (though I have not studied it yet) seems closer to what it should be. The problem is the emergence of an additional variable (that β) that has only some discrete values. So it (β) can not be attributed to a concrete physical size and as such the relationship can not be implemented on electronic simulators. That's how things stand until I study the issue of collaboration between SMath Studio and my simulator, my greatest interest is focused on a complicated method of determining coefficients for the Koren equation that I presented in the first post . Otherwise, all the best. Nicolas. Edited by user 09 June 2018 16:03:03(UTC)
| Reason: Not specified
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: ola_nicolas That's how things stand until I study the issue of collaboration between SMath Studio and my simulator, my greatest interest is focused on a complicated method of determining coefficients for the Koren equation that I presented in the first post . Otherwise, all the best. Salut Nicolas, I was not gone fishing ! I read most of the link about Norman Koren work. Norman formulation(s) meets his experimental data, fine, gorgeous work. IMHO, Norman and others are not correct from the onset ... Explain: The physics of the system is the XFR of the system [Transfer Function] Each system has its own XFR ... one is exemplified in the attached, from the data generated by Carlos from Norman simplified formulation. To resume your concern: 1. get reliable data from your simulator 2. attach those data, we will try solving the XFR. Jean Project Tesla_CBG_R1 [optimized] Copy.sm (213kb) downloaded 16 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud my greatest interest is focused on a complicated method of determining coefficients Answer: Similar was done Mathcad 11 via a specific method unpublished.It is not possible in Smath, only manual . I have added another model at the end. Can't be automated either. Jean Project Tesla_CBG_R1 [optimized] Copy.sm (235kb) downloaded 20 time(s).
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 21/11/2017(UTC) Posts: 114 Location: Olt county Was thanked: 1 time(s) in 1 post(s)
|
Hello Jean Giraud
Explain the abbreviations please.
Nicolas
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: ola_nicolas Hello Jean Giraud
Explain the abbreviations please.
Nicolas IMHO => "in my humble opinion"
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud IMHO = ... this model should solve all curves. It solves/iterates nicely for this particular data set. f(x,β) recovers from the reduced solving procedure. Genfit Minimze Tesla.sm (46kb) downloaded 26 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
From what I understood, the data sets are from unknown lab test. Digitized from some simulation software or else valid method. The project is collected and attached in "Samples" If you have more questions, please do not hesitate. High quality document for publishing. Cheers Nicolas, be good ... Jean Genfit Minimize Tesla (2) Copy.sm (214kb) downloaded 17 time(s).
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 21/11/2017(UTC) Posts: 114 Location: Olt county Was thanked: 1 time(s) in 1 post(s)
|
Bonjour Jean Giraud Originally Posted by: Jean Giraud ... Cheers Nicolas, be good ... Jean I'm just trying to do it. Otherwise today there are all kinds of complex electronic devices for automatic drawing of the curves characteristic of the triodes and the pentodes. The technology has advanced to the point where, besides viewing the characteristic curves, coefficients from Norman Koren's phenomenological equation are provided. This is not what I want, but only to be able to determine these mathematical calculators at any time. The reasons are different. For example, the equation presented in the attachments in the 1st post contains an innovation introduced by myself. Thus, in the original equation, there is a factor, sqrt(kVB+Ep ^2). I considered that both the order of the radical and the power of the anode voltage could be an irrational number, very close to 2, but not exactly 2. That is why I introduced a new coefficient y so that the previous relationship becomes (kVB+Ep^y)^1/y. I also work with models of the Koren equation, in which the natural logarithm base is a base, very close to e, but not exactly e. This innovation has given me fewer reasons for satisfaction, which is why I did not insist. So, let go of the further development of Kore's phenomenological equation, but be able to determine it at any time. Here's what I'm want. ... Voila Nicolas Edited by user 13 June 2018 17:49:51(UTC)
| Reason: Not specified
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: ola_nicolas I'm just trying to do it. Otherwise today there are all kinds of complex electronic devices for automatic drawing of the curves characteristic of the triodes and the pentodes. Ok ... Savant gadgets are just gadgets, proof: the one you have used has muffed your lab test. L. Koren equation is phenomenological as it says. Unfortunately, experimental data are not exactly reproducible. Nothing in the circuitry is reproducible not even the triode/pentode ... think of similar physics wrt T/C T/C = Thermocouples. In short, the Creator was short of fully understanding what He was creating. Your complex attempted formulation does not speak to me. The attached proposal fits your data sets based on 2 parameters and 2 fixed constants. The constants are the initials X, Y of each experiment. It needs only two parameters for each experiment. The model function is not suitable for auto-iterate ... peanut work to calculate trial/error. If possible, it would be interesting your collect ~ 20 points, raw ... as measured with good equipment. No processing, no spline curve junction. Salut Nicolas, will read you soon ... Jean Genfit Minimize Tesla(5) Copy.sm (208kb) downloaded 16 time(s).
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: Jean Giraud The model function is not suitable for auto-iterate ... peanut work to calculate trial/error. ... easy to trick the argument (x+ε) Now, with a single set of initial β it iterates and fits all data sets ! Genfit Minimize Tesla(5) Auto_Iterate.sm (212kb) downloaded 24 time(s).
|
|
|
|
Rank: Advanced Member Groups: Registered
Joined: 14/10/2015(UTC) Posts: 308
Was thanked: 77 time(s) in 58 post(s)
|
|
|
|
|
Rank: Guest
Groups: Registered
Joined: 04/07/2015(UTC) Posts: 6,866 Was thanked: 981 time(s) in 809 post(s)
|
Originally Posted by: CBG It contains a file in excel for the calculation of the triodes graphs Thanks Carlos, interesting visit. This project got lot of attention ! Jean 0Appendix_00000 TESLA Models.sm (17kb) downloaded 25 time(s).
|
|
|
|
Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.