Electronics > Projects, Designs, and Technical Stuff
Transformer Design for Power Amplifier
<< < (2/2)
xavier60:
If the tone burst starts at the zero crossing point, the flux will be doubled for the first half cycle. This needs to be taken into account when deciding the volts per turns to avoid saturation.

Extra: Or make the first half cycle half amplitude.
Archangel1235:

--- Quote from: xavier60 on November 30, 2018, 10:24:49 am ---If the tone burst starts at the zero crossing point, the flux will be doubled for the first half cycle. This needs to be taken into account when deciding the volts per turns to avoid saturation.

Extra: Or make the first half cycle half amplitude.

--- End quote ---

Thanks a lot for the reply

Hi can you please explain why this happens..?

Thanks ..

--- Quote from: xavier60 on November 30, 2018, 09:06:23 am ---You should tell us what you have tied so far. Although I have successfully made a few SMPS transformers, I don't fully understand what I'm doing.
The primary turns are calculated to avoid core saturation at the combination of the lowest expected frequency and highest primary voltage.
The core size doesn't directly determine power handling, mainly copper loss does.
Larger cores allow for less turns per volt and thicker wire which increases power handling. Because your busts have such a low duty cycle, power handling becomes less of a problem.
For a recent project I used an ETD49 with 3C90 cores at 40Khz. Although I utilized the winding window rather inefficiently, it easily handles 600 watts continuously.
With your project, minimizing winding capacitance would be important. https://coefs.uncc.edu/mnoras/files/2013/03/Transformer-and-Inductor-Design-Handbook_Chapter_17.pdf

--- End quote ---

This my current transformer

core           -  ETD 34 with N97
primary      -  Awg23(3x litz from) 4 turns
secondary  -  Awg 30 84 turns

With this below 40 KHz everything is fine only above that the gain drops...  And no precautions were taken to reduce capacitance.,.. so that might be a problem too..


--- Quote from: fourtytwo42 on November 30, 2018, 09:07:35 am ---I am not surprised at your problems the site is laughable, nowhere does it show or include in calculations the effects of operating frequency!!  More specifically skin effect in winding's that reduces the effective area of round wires as frequency increases nor core losses that again increase as frequency increases.

I would suggest you find yourself a copy of ExcellentIT7100 (Russian authored freeware) that at least takes these into account.

Expect to be using expensive Litz wire and copper foil as a minimum in this design. A 1Kw core able to operate over 30-150Khz is going to be expensive too. But at least you can get some solid idea's using the software I have suggested, it is not perfect BTW but gives a very good starting point for estimating the design parameters.


--- End quote ---

Thanks a lot for the input  I think the primary windings in my setup is to blame AWG gauge 23 is not suitable for anything above 53KHz according to this site https://www.powerstream.com/Wire_Size.htm This is more or less what Im observing above 40KHz the transformer gain drops with load..

I will reduce the AWG in primary windings and report back the results..

Thanks a lot for helping me :)
xavier60:
It is something you really need to think hard about.
If the sine wave has already been applied to the Primary inductance for many cycles, the magnetizing current lags the voltage waveform by 90°.
At each voltage zero crossing point the current has reached peak from the previous quarter cycle of applied voltage. Moving on from the voltage zero crossing point to the voltage peak, the current decreases to zero because the applied voltage polarity is now reversed to that of the previous quarter cycle. After the voltage peak the current begins to increase again in the opposite direction. And so on.
When a voltage half cycle is first applied starting at the zero crossing point, the current increases in the one direction for the whole half cycle.
The term "volt seconds" is relevant.
dmills:
Is this a sonar application? If so, how capacitive is the load? Piezo ceramic transducers generally have quite large fixed C.

A 1:400 impedance ratio transformer is a big ask, I would probably have designed for a much higher rail voltage or used a tuned transformer (requires a gapped core) and L network or such (30k-150k is a little wide for that however, usually driving a piezo ceramic in water you can get about an octave useful transmit bandwidth).

Try splitting the transformer, make say a 4:1 (voltage ratio) which can be easily done as a TLT, then about a 5:1 or so high voltage jobbie, much less of an interwinding capacitance problem that way.

Regards, Dan.
T3sl4co1l:
What is the load?  Is it actually resistive, or is it reactive?  In that case, is the 1A at 1kV typical of the reactance?  (I.e. ~1nF capacitance.)

Conventional transformer windups are a practical simplification (and to some extent, perversion) of a more general concept, transmission lines.

All transformers have a characteristic impedance and length (or equivalently, some cutoff frequency, or Cp and LL in the 2nd order LF equivalent model).  These parameters aren't necessarily straightforward to derive (e.g., for multilayer or scatter-wound windings), but they are always present.

Transformers designed based on a transmission line concept, have the highest bandwidth that is possible, and the parameters are easy to derive from the construction.  The downside is, the construction may be more complicated.

So, all that said, what of it?

Transformers have impedance, so we expect that that impedance must be somewhere near the circuit impedance.  Namely, you are expecting around a 1kohm secondary, so the characteristic impedance of that winding (with respect to the primary winding, or to itself, as the case may be) should be similar.

Since the turns ratio is not very small (i.e., more than, say, 3:1), the impedance contribution from the primary, as seen by the secondary, will be small.  The primary is ~ohms, and the secondary is ~kohm.  So, it's close enough to consider the secondary by itself, as if it were a single transmission line surrounded by ground/shield.

The impedance doesn't need to match exactly, and the higher the cutoff frequency is above the operating frequency, the less critical it is.  So we desire to have as high a cutoff as possible given the power handling and turns ratio, and a characteristic impedance that's close enough to behave.

So, transmission lines.  The characteristic impedance depends upon the geometry of the cross section.  If the signal wire is large, and close to ground, the impedance is low; if thin and distant, it is high.  The geometry acts as a ratio relative to the impedance of free space, Zo = 377Ω, and the ratio is usually smaller (50-150Ω TLs are much more typical, with TLs up to 600Ω being known but uncommon, and in particular, not very practical for winding into a transformer!).

So, needing 1kohm, we already know we will be pressed for impedance.

Probably the best you will be able to do, is a cylindrical winding (around a U or P style core), with the primary being about one turn (probably made with foil), and the secondary being a single layer solenoid (helix) wound on a form, spaced away from the primary.  The intent is to allow more coupling between neighboring secondary turns, than between a given secondary turn and the primary, with the effect of raising the secondary's impedance.  (This raises LL a bit faster than it reduces Cp -- thereby reducing Fc overall, so don't go too crazy with it.  A modest distance will do well enough.)

The core cross section should be as large as possible, to allow for a single turn primary, with as much flux as is required at the lowest frequency and highest voltage.  This allows a minimal length of secondary wire, which is the limiting factor in terms of electrical length, and therefore Fc.

A conventional build, with layers of foil or wire, insulated with tape, may not be too bad, if the total wire length is still very short relative to what's required (i.e., an Fc of >1MHz allows 150kHz operation at a ~6:1 impedance mismatch, i.e., a ~150Ω winding is acceptable -- this would require a fairly heavy layer of tape but not so much that a separate winding former would be desirable).

Tim
Navigation
Message Index
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod