Elektroda.com
Elektroda.com
X
Elektroda.com

How to design a pulse transformer

_lazor_ 10281 39
This content has been translated » The original version can be found here
  • 1. Magnetic field in matter

    Prof. Roman Kurdziel „Podstawy elektrotechniki” wydanie II całkowicie zmienione. wrote:
    The magnetic field in a vacuum depends only on the electrical circuits that generate it. In material environments, there is also the influence of molecular currents in particles of matter. An electron moving around the nucleus at the angular velocity omega_zero in an orbit with radius r represents an elementary magnetic dipole. Magnetic dipoles are usually arranged chaotically in the matter, so that the body does not show a magnetic state unless it is subjected to an external magnetic field, i.e. generated by external causes, e.g. current in any electrical circuit.
    The elementary magnetic dipole placed in the external magnetic field is subject to a mechanical moment, which sets the electron, regardless of the orbital motion, in a precessive motion, similar to the motion of a fob. The axis of the precession is the field strength vector H. As a result, an additional magnetic field is created which slightly weakens the external field. This phenomenon is called diamagnetism.
    For the reason given above, all bodies should be diamagnetism. However, in many bodies placed in an external magnetic field, the opposite phenomenon is observed, i.e. a certain strengthening of the external magnetic field. The o phenomenon can be easily explained if it is assumed that the electron, apart from orbital motion, rotates around its axis, called the electron spin. The electron spin is accompanied, regardless of the magnetic moment resulting from the orbital motion, by the spin magnetic moment p_s.
    In individual atoms, some electrons spin in one direction, others in the opposite direction, which corresponds to the opposite sense of the spin moment. If the numbers of electrons spinning back and forth are equal, the sum of the spin magnetic moments is zero and the body exhibits diamagnetic properties, resulting from the orbital motion of the electrons. On the other hand, if the number of electrons with a certain direction of spinning prevails, the sum of the spin magnetic moments is different from zero and the atom exhibits a certain resultant spin moment, which in the external magnetic field tends to assume a position consistent with the direction of the field intensity. The spin dipole field and the external magnetic field add up, i.e. the presence of this type of matter increases the magnetic field in relation to the field that a given electric circuit would generate in a vacuum. This phenomenon is called paramagnetism.
    The influence of the environment on the magnetic field is thus marked: in diamagnetic environments a decrease, and in paramagnetic environments an increase in the magnetic inductance B in relation to the inductance B_o that a given external field of intensity H would cause in a vacuum. The ratio of the resultant induction B to the intensity of the external field H is called the magnetic permeability of the environment
    $$u = \frac{B}{H}$$
    (...) At certain values of the ratio of the distance D between atoms to the diameter of the atom d, namely when 1.5

    Cool! Ranking DIY
    Can you write similar article? Send message to me and you will get SD card 64GB.
    About Author
    _lazor_
    Moderator of Designing
    Offline 
    _lazor_ wrote 2445 posts with rating 508, helped 198 times. Live in city Wrocław. Been with us since 2016 year.
  • #2
    SylwekK
    Level 31  
    I wound up my first network hit (and the next few) in primary school. My father subscribed to a young technician in which all necessary calculations were described. These transformers are still working today, and some time has already passed :)
  • #3
    And!
    Admin of Design group
    In general, the topic is beyond me, so after reading I did not speak, but since the topic of network transformers was raised,
    then I will ask for one detail, maybe I can find the answer to such a question:

    Why do some transformers on a core made of metal EI plates (e.g. in a microwave) have a "welded core" in one place so that there is a "short circuit" between the core plates?


    I found some illustrative photos:
    https://www.alibaba.com/product-detail/220v-1...ve-power-voltage-transformer_60475738378.html
    https://www.indiamart.com/proddetail/microwave-oven-transformer-20222540248.html
    https://www.kitchenwareonline.com/samsung-high-voltage-transformer-c2x14483870

    PS
    I wonder if it could be "drawn" material from pdf
    https://elnamagnetics.com/wp-content/uploads/...ube-Materials/3C90_Material_Specification.pdf
    to the post in case the pdf disappears.
  • #4
    _lazor_
    Moderator of Designing
    I'm not an expert on network transformers, but they probably weld to minimize costs. After all, everything must fit tightly so that it does not buzz too much and the use of screws and flat bars would probably be more expensive and less convenient to install than weld. These are my suppositions.

    As for pdf I will transfer the charts to the post.
  • #5
    And!
    Admin of Design group
    It can really be an optimization of production costs, but whether such welding does not increase the loss in the core and more specifically eddy current.
  • #6
    ArturAVS
    Moderator of HydePark/Cars
    And! wrote:
    does such welding increase the losses in the core and more specifically eddy currents?

    It increases, it also causes a large no-load current. Note, however, that such luck works casual.
  • #7
    User removed account
    Level 1  
  • #8
    ArturAVS
    Moderator of HydePark/Cars
    o_Tadeusz wrote:
    There is a simple way not to count complicated designs. When the mains transformer is called, count the turns when unwinding the secondary winding. By dividing the number of turns by voltage, we get the number of turns per volt. It remains to count the number of windings for the new voltage and select the wire according to the required current. The maximum power cannot be exceeded because the wire will not fit on the wireframe.

    Yeah, and who will want to count those unwound rolls?

    Simple and proven pattern:
    Quote:
    Core cross section (Qr) - for classic EI cores - cross section area of the middle column of the core. The core cross-section, operating frequency and maximum induction determine the number of winding turns according to the formula:

    n = (U * 10 ^ 8) / (4.44 * f * Bm * Qr) where:
    n - number of turns
    U - voltage [V]
    f - frequency [Hz]
    Bm - maximum induction [Gs]
    Qr - core cross section

    As you can easily calculate, for induction maximum 10,000 Gs (== 1T), frequency 50 Hz, core cross section 1 cm ^ 2 and voltage 1V we need 45 turns.
    After analyzing and transforming the pattern, we can derive a fairly practical formula for the number of turns per volt for a classic 50 Hz mains transformer:

    n = 45 / Qr [zw / V, cm ^ 2]

    As I remember, this pattern has two characters differing in number 45 vs 44, one for continuous work and the other for occasional work. A dozen or so dozens of transformers according to this and all of them worked approx.
  • #9
    _lazor_
    Moderator of Designing
    o_Tadeusz wrote:
    SylwekK wrote:

    My father subscribed to a young technician in which all necessary calculations were described.

    There is a simple way not to count complicated designs. When the mains transformer is called, count the turns when unwinding the secondary winding. By dividing the number of turns by voltage, we get the number of turns per volt. It remains to count the number of windings for the new voltage and select the wire according to the required current. The maximum power cannot be exceeded because the wire will not fit on the wireframe.


    However, times are changing and network transformers are less and less used. Instead they are switched-mode power supplies, which are highly optimized and often you can forget about rewinding. However, the cost of the set (carcass, cores, hairpins, winding wire, insulation) comes out relatively cheap these days, so it's worth just designing the optimal transformer for your own needs. And since the formulas used for network transformers are not optimal for calculating pulse transformers, I decided to present more optimal calculations.


    arturavs wrote:

    Simple and proven pattern:
    Quote:
    Core cross section (Qr) - for classic EI cores - cross section area of the middle column of the core. The core cross-section, operating frequency and maximum induction determine the number of winding turns according to the formula:

    n = (U * 10 ^ 8) / (4.44 * f * Bm * Qr) where:
    n - number of turns
    U - voltage [V]
    f - frequency [Hz]
    Bm - maximum induction [Gs]
    Qr - core cross section

    As you can easily calculate, for induction maximum 10,000 Gs (== 1T), frequency 50 Hz, core cross section 1 cm ^ 2 and voltage 1V we need 45 turns.
    After analyzing and transforming the pattern, we can derive a fairly practical formula for the number of turns per volt for a classic 50 Hz mains transformer:

    n = 45 / Qr [zw / V, cm ^ 2]

    As I remember, this pattern has two characters differing in number 45 vs 44, one for continuous work and the other for occasional work. A dozen or so dozens of transformers according to this and all of them worked approx.


    The patterns given in the article are simply more general and can be used for various waveforms and importantly fillings. In the above case, constants such as 4.44 make them optimal only for transformers powered by sinusoidal voltage.
    Instead, the constant 10 ^ 8 reminded me that I had to complete the formulas with units.
  • #10
    SylwekK
    Level 31  
    o_Tadeusz wrote:
    There is a simple way not to count complicated designs. When the mains transformer is called, count the turns when unwinding the secondary winding. By dividing the number of turns by voltage, we get the number of turns per volt. It remains to count the number of windings for the new voltage and select the wire according to the required current. The maximum power cannot be exceeded because the wire will not fit on the wireframe.


    I also used this obvious method. My father rewound the engines, I had access to virtually any size of DNE, and the winder that counted the unwound turns helped me count the turns. :) Unfortunately, it wasn't rosy with everyone. Some of them were simply melted to the ankle and even the carcass had to be made from scratch with only (or as much as ...) a core.
  • #11
    ArturAVS
    Moderator of HydePark/Cars
    _lazor_ wrote:
    The patterns given in the article are simply more general and can be used for various waveforms and importantly fillings. In the above case, constants such as 4.44 make them optimal only for transformers powered by sinusoidal voltage.

    By all means I meant mains transformers, where we have sinusoidal voltage.
  • #12
    User removed account
    Level 1  
  • #13
    Renegat_pol
    Level 20  
    [Quote = "SylwekK"]
    o_Tadeusz wrote:
    ... Some of them were simply melted to the ankle and even the carcass had to be made from scratch with only (or as much as ...) a core.


    The reality of the 1980s. There was nothing to do with the carcass. In the factory where I had apprenticeships, the bobbins were made of ... fiberboard.
  • #14
    RitterX
    Level 38  
    With microwave welding from a microwave it seems a simple matter. It is about magnetostriction and acoustic experience that would accompany this size of a fully loaded transformer. Welding the sheet metal along the edges causes some slight deterioration of the core loss, but there is no drama. Eddy current loss is a spatial phenomenon from the point of view of the magnetic field in the core. It is better to increase core losses as losses due to returned microwaves because they are more and more buzzing :) . Let's make an appointment, it will no longer be better packaged than when you leave the factory.
  • #15
    rysieklew
    Level 2  
    Hello, I used to use the article from the Young Technician for calculations, then the Excel spreadsheet. Now it is the low and medium power power supplies that are replaced by switching power supplies. Maybe someone will need this Excel sheet.
  • #16
    Mark II
    Level 21  
    Modern low power industrial on classic EI shapes are also welded. I suppose that in this case the reduction of production costs is of particular importance.
  • #17
    ^ToM^
    Level 38  
    arturavs wrote:

    n = 45 / Qr [zw / V, cm ^ 2]
    As I remember, this pattern has two characters differing in number 45 vs 44, one for continuous work and the other for occasional work. A dozen or so dozens of transformers according to this and all of them worked approx.


    No, the higher value was for small transformers and the smaller for larger ones made of better transformer sheet. The counter ranged from 43 to even 48. It all depends on which core and transformer the equation applied to. Nevertheless, the model is the most correct and practical.
  • #18
    _lazor_
    Moderator of Designing
    All in all, you don't need to include insulation between packet sheets in your formulas and add a column fill factor?
  • #19
    OldSkull
    Level 27  
    It is a pity that the assumption in the calculations is the lack of a gap. In practice, the gap is very important, because although it reduces inductance, it protects against saturation of the core. Unfortunately, sometimes detailed information is missing in the catalog notes.
    In practice, there are a lot of things that need to be measured on a finished transformer and iterate the next change of design. For example, the very presence of other windings has an effect on inductance (but this is the effect of affecting other transformer parameters). In practice, no prototype or movement. It is good that the impulse transformers are demountable (if someone does not flood them), because you can easily play with cores and moderately with windings.
    Usually it looks like this, we get to know what the transformer working conditions are and we calculate what the windings and core should be based on.

    Also remember that the transformer is an induction coil (only multiple) and the same rules apply to it as the coils.

    I have an additional question:
    Well, the epidermal effect is important for choosing the face that we wind the transformer. But there is one more effect that affects: the currents flowing through the individual threads of the weave of the face affect each other ("proximity effect"), causing uneven distribution of currents in the face. In short: the current is more likely to flow through the outer surface of the face, even reducing the depth of penetration of the current into the face. The result is that you would have to take smaller veins of the face than it results from epidermis. I met with a study on this subject, but it was too academic and came out for practical cases, quite impractical small cross-sections. Do you know anything about this? Did you meet something like that?
  • #20
    _lazor_
    Moderator of Designing
    Generally, the slit causes many unwanted effects like:
    - increasing the number of turns
    - the stream in the crevice likes to run sideways, which can cause problems with EMI
    - energy accumulation in a transformer, if it is not a topology based on it, it is undesirable
    - saturation is fought differently, e.g. the mentioned safe-start algorithm or DC component cut-off capacitors.


    I have heard about the proximity effect but I haven't really explored it so much, so unfortunately I won't write more about it at the moment.
  • #21
    ^ToM^
    Level 38  
    OldSkull wrote:

    I have an additional question:
    Well, the epidermal effect is important for choosing the face that we wind the transformer.


    After all, I don't know if you are talking about transformers with an iron core powered from a 50 Hz network or pulsed on powder cores. Because these are two different fairy tales basically. The former are usually not used with the classic DNE face. The epidermal effect should be considered, but usually not at 50 Hz.

    In addition, many network transformers do not have a traditional air gap when the core packet is alternately assembled.

    In my opinion it's getting a mess. There should be two themes, run in parallel. How to design a network transformer and another How to design a transformer for a pulse converter.
  • #22
    _lazor_
    Moderator of Designing
    The article is about the use of ferrite cores, not powder cores. Powder cores are cores with a dispersed gap and are not the same as ferrite, where the joining points are ground and adhere to minimize the gap, which is undesirable.

    In total, network transformers are closer to this topic than transformers with a slot for converters such as flyback.
  • #23
    OldSkull
    Level 27  
    ^ToM^ wrote:
    The epidermal effect should be considered, but usually not at 50 Hz.

    The article is based on the example of a pulse transformer.
    ^ToM^ wrote:
    In my opinion it's getting a mess. There should be two themes, run in parallel. How to design a network transformer and another How to design a transformer for a pulse converter.

    All in all a good idea because the approaches are different. For network, e.g., toroid winding is very important. At pulses due to the small number of turns, they even take a toroid, the problem is not great.

    _lazor_ wrote:
    Generally, the slit causes many unwanted effects like:
    - increasing the number of turns
    - the stream in the crevice likes to run sideways, which can cause problems with EMI
    - energy accumulation in a transformer, if it is not a topology based on it, it is undesirable
    - saturation is fought differently, e.g. the mentioned safe-start algorithm or DC component cut-off capacitors.

    When generating high voltage, it is generally difficult to bite the subject without a gap.
    By contrast, increasing the number of turns: it's not always bad. Sometimes, although a small gap helps a lot not to operate on the level of the number of turns, e.g. 2-3 (if the relatively low inductance value is important but high currents may be), because then the repeatability of the conductors in the transformer is of great importance.
    In other words: wherever there are relatively high currents or high voltages, the gap is very useful. E.g. when we have a large gear ratio.
  • #24
    _lazor_
    Moderator of Designing
    OldSkull wrote:
    _lazor_ wrote:
    Generally, the slit causes many unwanted effects like:
    - increasing the number of turns
    - the stream in the crevice likes to run sideways, which can cause problems with EMI
    - energy accumulation in a transformer, if it is not a topology based on it, it is undesirable
    - saturation is fought differently, e.g. the mentioned safe-start algorithm or DC component cut-off capacitors.

    When generating high voltage, it is generally difficult to bite the subject without a gap.
    By contrast, increasing the number of turns: it's not always bad. Sometimes, although a small gap helps a lot not to operate on the level of the number of turns, e.g. 2-3 (if the relatively low inductance value is important but high currents may be), because then the repeatability of the conductors in the transformer is of great importance.
    In other words: wherever there are relatively high currents or high voltages, the gap is very useful. E.g. when we have a large gear ratio.


    The article is a result of the fact that it designs a 230 / 15kV system in SR topology (series resonance with parallel load to C). There, the most fissure is something undesirable.
  • #25
    ^ToM^
    Level 38  
    _lazor_ wrote:
    The article is about use ferrite cores, not powder cores.


    The powder core is the general name for cores made by sintering or pressing particles instead of silicon iron sheets.

    Basically, powder cores can be divided into two groups:

    - powder cores made of Iron powder core,
    - powder cores made of alloys (Alloy powder core).

    Thus, the IPC core are ferrite cores.
  • #27
    ^ToM^
    Level 38  
    _lazor_ wrote:
    https://pl.wikipedia.org/wiki/Ferryt

    3C90 material is not included in the IPC. If you do not agree, please provide some evidence or do not litter the thread.


    I gave general information. The subject is, after all, general, while the core is given as an example, for which the example was considered. It is hard to believe that the topic was to apply only to the core of 3C90 material.

    https://pl.wikipedia.org/wiki/Ferryty

    Ferrites are usually made by sintering powdered metal oxides in the right proportions.

    https://de.wikipedia.org/wiki/Ferrite

    The starting materials for the production of soft magnetic ferrites are finely ground iron-oxygen compounds, such as iron (III) oxide or hematite. ..... These powder materials are mixed as evenly as possible or as evenly as possible in a water bath. Then the mixture is subjected to a chemical process, calcining at a temperature of about 1000 ° C.

    All pressed cores are made of powders. Both ferrite and all the rest. 3C90 material and successors are made of ferrite powders. I don't know what the dispute is. It's a matter of precision in naming to distinguish it from silicon metal cores and other transformer metal sheets.
  • #28
    jack63
    Level 42  
    _lazor_ wrote:
    3. Assumptions for transformers

    I suggest you change it, because it looks a bit strange in Polish.
    Like a trifle, but why spoil a job well done.
    _lazor_ wrote:
    Fig. 1 shows magnetic permeability as a complex number (u'_s and already '' _ s). The ratio of these two values (u '' _ s / u'_s) determines the material loss tangent (the ratio of the power stored in the magnetic field to the dissipated power in the form of heat).

    It should be noted that this makes sense for a sinusoidal waveform, i.e. practically only network transformer.

    I think that you should clearly separate the calculations for transformers or chokes powered by sinusoidal and impulse voltage (current).
    The phenomena that you describe as the start of the network transformer are the basis for the work of transformers in impulse systems.
    In fact, the latter work only in consecutive transient states. There is no phase shift between voltage and current. Effective values etc. must not be used.
    One more remark:
    Computational selection of magnetizing current is quite arbitrary. That is, two transformers wound on the same core with identical assumptions regarding voltage and power may have different numbers of turns! Both will work and only one can have more losses and more heat.
    Various crafters, such as manufacturers of "space" welders, use it.
  • #29
    _lazor_
    Moderator of Designing
    In general, the rectangular waveform is considered as a Fourier series and the data determined for the sine (as a sum) can be used as much as possible.

    In general, switched-mode power supplies can work in sine, for example, resonant power supplies. Anyway, the above formulas can be used for different waveforms, because we operate with the integral of voltage and not constants determining the shape of voltage on the primary winding.

    Switching power supplies can have their current and voltage phase shifts determined as much as possible. Look at such topology as phase shift full bridge.
    For example, I once built something like this on stm32f334 in half bridge topology:
    https://www.youtube.com/watch?v=VRjyxihpBYU

    The phase shift and current sine of the primary winding are the most. And using the theory (Fourier, atan2 etc.) I automagically tune it when I change the value of induction of the resonant choke (which was the coil of the heater).

    Transformer cores are elements whose accuracy is +/- 10%. It is difficult to predict with great precision how the whole core will behave.
  • #30
    RitterX
    Level 38  
    Faraday's Electromagnetic Induction Law, which is used to derive the formula for the number of transformer coils, is just as valid for the transformer core with: silicon transformer sheet, powder core, micrometal, metalgals, ... Therefore, somehow I can not see the need to divide the thread into subgroups.
    For a rectangular waveform, the factor in the formula is 4 and this is the canonical form of the derivation, and for a sine wave it should be slightly modified, it is 4.44. It also has physical justification.
    Silicon transformer sheet is quite a broad concept. Just change the amount of silicon in the range of 0.5% ... 4% to radically change the maximum values of saturation induction, usually in the range of 1T ... 1.6T and magnetic permeability. This means that without knowing the sheet metal parameters, it is impossible to optimally design tafo. Just like the powder core and any other. Although in the case of transformer sheet metal and a small, constant operating frequency, the indulgence margin is definitely greater.
    A gap in the core of a network transformer is a necessary production evil. It affects losses and significantly through the generation of not some magical EMC but the most ordinary stream of dispersion. To avoid this, not only punched sheets are used for cores but rather for cores made of coiled tape. Comparing the network toroid with such an EI one can draw two conclusions: the size and significantly lower voltage drop between no-load state and full load. Once it matters and others you can skip it.

    Current displacement in the winding does not apply only to high frequencies. That is why in high-power motors powered from the 50 Hz electricity network windings are built in no other way. They are not necessarily of solid wire and often of insulated and properly packaged tape. In this particular case, improvement is not necessarily possible because nurseries are a necessary evil :) and they cannot be enlarged indefinitely and the number of turns must match exactly as in the transformer.