#99 ideal transformer - again

0.0.17
closed-invalid
Richard C
None
5
2014-07-28
2013-06-07
Thomas Bechteler
No

Dear Qucs-Team!

It seems that there is a lack of profound knowledge in electronics.
I still believe that there is a bug in the (ideal) transformer.
At the end of this message I cited the last statement about the ideal transformer.

This statement is wrong!

The ideal transformer is defined by the governing equations

turn ratio N:=N2/N1

u2 = N * u1
i1 = -N * i2

These simple equations can be looked up in
1) Leon O. Chua: Linear and Nonlinear Circuits. McGraw Hill, 1987.
2) Jiri Vlach: Computer Methods for Circuit Analysis and Design. VNR, 1994.
and in many other books.

Due to the definition above, the ideal transformer does NOT
act as a short circuit, not even for dc!
A REAL transformer acts as a short circuit for dc only!
For ac signals it acts not as a short circuit, as practice shows us.
Even on the cited webpage (http://qucs.sourceforge.net/tech/node48.html)
it is stated "And in matrix representation this is for DC and for AC simulation".

The current which flows into, e.g., port 1 is only defined by the resistance
connected to port 2 (and, for the real transformer, some current due to the
stray field of the transformer).

The ideal transformer is not only a perfect current transformer,
but ALSO a perfect voltage transformer!

The equations which can be found on the cited webpage
(http://qucs.sourceforge.net/tech/node48.html)
can mislead the reader. They try to make a link to the
scattering parameters. In my humble opinion, this was didactically
not accomplished.

As a result, there is still a bug in the ideal transformer!!!

Cordially

Thomas Bechteler

=====start citation============================================================

Having investigated this, I think there is not a bug after all, and instead there is a misunderstanding of what the ideal transformer component actually does.

The ideal transformer acts as a perfect current transformer, the current in the primary is mutiplied on the secondary. If you just put a voltage source across the terminals, it acts like a short circuit, so there is infinite current. At the same time in this case, there is no resistance, and therefore no voltage drop across the tranformer inputs, hence the output voltage is also zero.

See here for more information:

http://qucs.sourceforge.net/tech/node48.html

The ideal transformer component does not reflect a real physical transformer device, a real device must have winding resistance and inductance, and only works with AC voltages. An ideal transformer can transform DC current, which is physically impossible in the real world (maybe some exotic superconducting device I'm not aware of).

What changed in Qucs 0.0.15 -> 0.0.16 as far as I can tell was that the behaviour was actually corrected for DC simulation.

=====end citation============================================================

Related

Bugs: #99

Discussion

  • Richard C
    Richard C
    2013-06-08

    Thomas, ok, lets make this simpler, could you outline how the mna
    equations shown in the link I gave to the qucs docs are incorrect, as it
    is these equations exactly which are implemented in qucs.

    The ideal transformer does act as a perfect voltage transformer but what
    is the voltage u1 across the winding when the winding has no resistance
    or inductance?

    Richard

     
    • Dear Richard,

      all right, I studied a little bit on the MNA,
      since I was not very firm in this topic.

      I checked the equations on the webpage
      http://qucs.sourceforge.net/tech/node48.html
      and they are correct. I didn't mean that
      they are incorrect. There is no clear explanation
      how to come from Eq. (9.26) to the S-parameters.
      (Maybe there was no such intention?)
      The link between Fig. 9.2 and the S-parameters
      is very dangerous. In Fig. 9.2 the numbers are
      terminal numbers and not port numbers! Actually,
      the shown ideal transformer is a two-port where
      terminals 1 and 4 make up port 1, and terminals 2 and 3
      make up port 2. But that is another story ...

      Back to Eq.(9.26): While the matrix itself is
      correct, the rightest side seems not correct to me.
      The currents are all set to zero which is not
      the case when the ideal transformer is connected
      to a circuit including sources. In Section 5.3.13
      (Ideal Transformer), on the webpage
      http://www.analog-electronics.eu/analog-electronics/modified-nodal-analysis/modified-nodal-analysis.xhtml
      the current vector is not a zero vector as it
      should be for the general case. Maybe this is
      a flaw implemented in QUCS.

      Remark 1:
      In the literature, the currents I_t and T \times I_t
      are shown all into the transformer. In Fig. 9.2 current
      I_t is shown flowing out of terminal 1. This is,
      of course, possible since any designation can be
      chosen arbitrarly. Only the signs in the matrix
      change accordingly.

      Remark 2:
      I want to mention again that in QUCS the ideal transformer
      does not work anymore in the TRANSIENT analysis.
      But is does work in the AC analysis.
      By the way, the mutual inductor in QUCS works as it
      should be in both, the TRANSIENT analysis and the
      AC analysis. This is now the work around for me.

      To your question:
      Actually, when speaking about an ideal transformer,
      there is no winding, coil, inductor or resistor to be
      considered. The ideal transformer is only defined by
      mathematical equations. So it works perfectly well for
      dc or ac, in the time domain or in the frequency domain.
      Unfortunately or not, the schematic symbol for the
      ideal transformer is depicted as two inductors.
      This may mislead some people that there are indeed
      inductors. However, there are no inductors!

      Please, find attached to this message a pdf-file
      which describes two examples for the MNA with an
      ideal transformer. I hope that I did not introduce
      any mistake and that I illuminated the topic from a
      different point of view.

      Cheers

      Thomas

       
  • Richard C
    Richard C
    2013-06-10

    So, I went back and tried the schematic in the original bug report (very simple, but I've also attached it). This seems to work fine in the development version of Qucs. You're right about it not being a short-circuit, because the impedance is propogated from the other side.

    Thomas, could you try the development snapshot to confirm this works for you?

     
    Attachments
  • Richard C
    Richard C
    2013-06-10

    This works fine on the dev version compiled on Mint Linux 14 I should add, to be specific.

     
    • qucs 0.0.17 compilation fails because of

      componentdialog.moc.cpp:15:34: fatal error: private/qucomextra_p.h: No such file or directory

      I guess it is a problem with qt, as usual ... I tried to fix it,
      but couldn't succeed up to now. Maybe OpenSuSE (12.1) always causes
      qt-problems ... or I wait until an rpm for SuSE is available ... lets see ...

      Thomas

       
  • Frans
    Frans
    2013-06-10

    <html>
    <head>
    <meta content="text/html; charset=ISO-8859-1" http-equiv="Content-Type">
    </head>
    <body bgcolor="#FFFFFF" text="#000000">
    Have you installed libqt4-devel?



    On 06/10/2013 09:41 PM, Thomas Bechteler wrote:



    qucs 0.0.17 compilation fails because of


    componentdialog.moc.cpp:15:34: fatal error:
    private/qucomextra_p.h: No such file or directory


    I guess it is a problem with qt, as usual ... I tried to fix
    it,

    but couldn't succeed up to now. Maybe OpenSuSE (12.1) always
    causes

    qt-problems ... or I wait until an rpm for SuSE is available
    ... lets see ...


    Thomas




    [bugs:#99]
    ideal transformer - again


    Status: open

    Created: Fri Jun 07, 2013 09:57 PM UTC by
    Thomas Bechteler

    Last Updated: Mon Jun 10, 2013 06:02 PM UTC

    Owner: Richard C


    Dear Qucs-Team!


    It seems that there is a lack of profound knowledge in
    electronics.

    I still believe that there is a bug in the (ideal)
    transformer.

    At the end of this message I cited the last statement about
    the ideal transformer.


    This statement is wrong!


    The ideal transformer is defined by the governing equations


    turn ratio N:=N2/N1


    u2 = N * u1

    i1 = -N * i2


    These simple equations can be looked up in

    1) Leon O. Chua: Linear and Nonlinear Circuits. McGraw Hill,
    1987.

    2) Jiri Vlach: Computer Methods for Circuit Analysis and
    Design. VNR, 1994.

    and in many other books.


    Due to the definition above, the ideal transformer does NOT

    act as a short circuit, not even for dc!

    A REAL transformer acts as a short circuit for dc only!

    For ac signals it acts not as a short circuit, as practice
    shows us.

    Even on the cited webpage
    (http://qucs.sourceforge.net/tech/node48.html)

    it is stated "And in matrix representation this is for DC and
    for AC simulation".


    The current which flows into, e.g., port 1 is only defined by
    the resistance

    connected to port 2 (and, for the real transformer, some
    current due to the

    stray field of the transformer).


    The ideal transformer is not only a perfect current
    transformer,

    but ALSO a perfect voltage transformer!


    The equations which can be found on the cited webpage

    (http://qucs.sourceforge.net/tech/node48.html)

    can mislead the reader. They try to make a link to the

    scattering parameters. In my humble opinion, this was
    didactically

    not accomplished.


    As a result, there is still a bug in the ideal transformer!!!


    Cordially


    Thomas Bechteler


    =====start
    citation============================================================


    Having investigated this, I think there is not a bug after
    all, and instead there is a misunderstanding of what the ideal
    transformer component actually does.


    The ideal transformer acts as a perfect current transformer,
    the current in the primary is mutiplied on the secondary. If
    you just put a voltage source across the terminals, it acts
    like a short circuit, so there is infinite current. At the
    same time in this case, there is no resistance, and therefore
    no voltage drop across the tranformer inputs, hence the output
    voltage is also zero.


    See here for more information:


    http://qucs.sourceforge.net/tech/node48.html


    The ideal transformer component does not reflect a real
    physical transformer device, a real device must have winding
    resistance and inductance, and only works with AC voltages. An
    ideal transformer can transform DC current, which is
    physically impossible in the real world (maybe some exotic
    superconducting device I'm not aware of).


    What changed in Qucs 0.0.15 -> 0.0.16 as far as I can tell
    was that the behaviour was actually corrected for DC
    simulation.


    =====end
    citation============================================================




    Sent from sourceforge.net because you indicated interest in https://sourceforge.net/p/qucs/bugs/99/


    To unsubscribe from further messages, please visit https://sourceforge.net/auth/subscriptions/




    </body>
    </html>

     

    Related

    Bugs: #99

    • Yes, libqt4-devel was installed.
      Other qt4-related development packages
      which I installed didn't change the situation ...

      By the way, the directory /private only exist
      in the qt3-path

      Thomas

       
  • Stefan Jahn
    Stefan Jahn
    2013-06-13

    I think my email didn't reach anyone...


    Hello,

    thank you for the explanation :)

    The change from 0.0.15 to 0.0.16 was introduced since some
    users expected the ideal transformer to do something different
    in the time domain.

    Actually, both trafo and strafo where just designed for AC domain.
    They transfer ideally voltages and impedances.

    In the DC steady state domain, since it is ideal element, there is a
    short at input (only L) -> thus also V=0 on output.

    Now, when using the thing in time domain, L for ideal trafo
    is not defined: L --> 0 we still have a short as in DC. For
    L --> infinite we get a transfer as in AC domain, BUT: also
    a transfer for DC signals... (which is not physical)

    Conclusion: depending on assumptions it will do different things.
    My suggestion: Use mutual inductance models in transient simulations
    instead of the ideal trafo models.

    BR, Stefan.

    On Fri, June 7, 2013 5:57 pm, Richard C wrote:

    Having investigated this, I think there is not a bug after all, and
    instead there is a misunderstanding of what the ideal transformer
    component actually does.

    The ideal transformer acts as a perfect current transformer, the current
    in the primary is mutiplied on the secondary. If you just put a voltage
    source across the terminals, it acts like a short circuit, so there is
    infinite current. At the same time in this case, there is no resistance,
    and therefore no voltage drop across the tranformer inputs, hence the
    output voltage is also zero.

    See here for more information:

    http://qucs.sourceforge.net/tech/node48.html

    The ideal transformer component does not reflect a real physical
    transformer device, a real device must have winding resistance and
    inductance, and only works with AC voltages. An ideal transformer can
    transform DC current, which is physically impossible in the real world
    (maybe some exotic superconducting device I'm not aware of).

    What changed in Qucs 0.0.15 -> 0.0.16 as far as I can tell was that the
    behaviour was actually corrected for DC simulation.


    [bugs:#83] ideal transformer

    Status: open
    Created: Mon Feb 20, 2012 01:30 PM UTC by Thomas Bechteler
    Last Updated: Fri Jun 07, 2013 01:29 PM UTC
    Owner: Richard C

    Hi QUCS developer!

    The ideal transformers (both under "lumped components") in version
    0.0.15 work.
    However, they do not work in version 0.0.16!

    Thomas

     

    Related

    Bugs: #83

  • Richard C
    Richard C
    2013-06-13

    Stefan, thanks for clearing this up. I think my understanding was not
    quite wrong then after all, as really the behaviour in transient sims is
    undefined.

    Perhaps we should we consider issuing a warning in transient sims in
    this case to this effect?

    My conclusion though is that, whatever happens, there is no bug, or at
    least nothing that holds up a release.

    Richard

    On 13/06/2013 07:31, Stefan Jahn wrote:

    I think my email didn't reach anyone...


    Hello,

    thank you for the explanation :)

    The change from 0.0.15 to 0.0.16 was introduced since some
    users expected the ideal transformer to do something different
    in the time domain.

    Actually, both trafo /and/ strafo where just designed for AC domain.
    They transfer ideally voltages and impedances.

    In the DC steady state domain, since it is ideal element, there is a
    short at input (only L) -> thus also V=0 on output.

    Now, when using the thing in time domain, L for ideal trafo
    is not defined: L --> 0 we still have a short as in DC. For
    L --> infinite we get a transfer as in AC domain, BUT: also
    a transfer for DC signals... (which is not physical)

    Conclusion: depending on assumptions it will do different things.
    My suggestion: Use mutual inductance models in transient simulations
    instead of the ideal trafo models.

    BR, Stefan.

     
  • Frans
    Frans
    2013-06-14

    • Description has changed:

    Diff:

    --- old
    +++ new
    @@ -1,4 +1,3 @@
    -
     Dear Qucs-Team!
    
     It seems that there is a lack of profound knowledge in electronics.
    
    • status: open --> closed-invalid
     
    • Dear Qucs-Team,

      maybe not calling it "ideal transformer".
      Some different name indicating that only
      ac simulation works ...

      But still a pity that there is no ideal transformer!!!

      cheers

      Thomas

       
  • Frans
    Frans
    2013-06-14

    I am closing this bug and add a feature request in stead, for adding a warning for transient and dc simulations.