Power Transformer behavior at higher frequencies

[alireza7]

Suppose a 220VAC to 110VAC ,50Hz, 20w transformer.

How is it's behavior in higher (1khz up to 10Mhz) frequencies? Does it attenuate signals with high frequencies or not?

Actually i want to know that if using a transformer on the mains line improves EMC of my system and protection against burst signals or not?

[eb4fbz]

Does it attenuate signals with high frequencies or not?

Yes, because impedance is very low at high frequency (differential mode). However, at very very high frequency you have to accound for interwinding capacitance for common mode rejection.

[alireza7]

Does it attenuate signals with high frequencies or not?

Yes, because impedance is very low at high frequency (differential mode). However, at very very high frequency you have to accound for interwinding capacitance for common mode rejection.

I would be appreciate if you explain more about interwinding capacitance.
How can i increase this capacitance in order to increase common mode rejection?

[T3sl4co1l]

Yes.

The small-signal model looks like this:



where the transformer part is an ideal transformer.

To understand capacitances, add capacitors between pairs of terminals too.  This gives the second order equivalent, which for a given load impedance, shows a low-frequency cutoff (due to the magnetizing inductance, Xo), a high frequency cutoff (due to leakage inductance X1+X2 and capacitances), and poor isolation at high frequencies (due to the capacitors from primary to secondary).

The capacitances and inductances can resonate, giving peaks or valleys in the frequency response around the HF cutoff.  A real transformer, with multiple layers of windings, can have many such peaks and valleys, which needs a higher-order model to represent.  Let's just quietly ignore that for now, and approximate it with this model instead.

For a typical split-bobbin mains transformer, leakage inductance dominates around a few kHz.  Actually, leakage is pretty significant at mains frequency itself, too, which combined with DC resistance of the windings (R1 + R2), allows some transformers to be "impedance protected" -- that is, you can short the secondary and it won't burst into flames.  It might still get hot though, as you will have noticed if you've seen a melted wall-wart before.  This is the feature, that comes with the bug: poor regulation, i.e., Vout varies strongly with Iout.  Often you'll measure an open-circuit voltage ~1.5 times nominal output; the output comes down to nominal at rated load.

Anyway, the capacitance on this type is also pretty low, which is nice for mains noise isolation purposes.

For a typical "shell" style transformer (EI core, windings overlapping on a single bobbin), leakage is lower, taking over in the low 10s of kHz.  Regulation is good.  Capacitance is higher.  For low-noise purposes, an electrostatic shield (a layer of foil, slitted to avoid it becoming a shorted turn) is usually inserted between primary and secondary, basically solving the capacitance problem with regards to noise.  (The capacitance is still relatively high (100s pF?), but it's to ground only.)

For a toroid style transformer, leakage is even lower and capacitance is even higher.  I don't know that I've seen one of these made with a shield; you could, but I don't know if anyone does?  Kind of awkward wrapping flat foil around a donut, after all.  Anyway, these have good efficiency, surprisingly high bandwidth (sometimes low 100s kHz), and good regulation.

If you take the question a little further and ask about SMPS transformers as well, then the bandwidth continues to rise -- it has to carry sharp square waves -- but there are again a number of variations.  Resonant SMPSs can have intentionally poor transformers, making use of leakage inductance as part of the resonant circuit.  The bandwidth in this case will be relatively low (a bit less than the switching frequency itself), at least when loaded with all the capacitors and diodes and stuff in circuit.  On the other hand, very compact, high frequency DC-DC isolators need to have very low capacitance and high bandwidth, which means making the transformers as small as possible.

Tim

[exe]

Just in case, there are devices designed for this purpose: power line filters.

2 T3sl4co1l: there are toroidal transformers with electrostatic shielding, but they are rare. The easiest way to find them is to search for "toroidal audio transformer" . Like this one: https://www.ebay.co.uk/itm/DIY-Audio-Grade-Toroidal-Transformer-300V-A-GOSS-band-and-screen-grid/382615178797?hash=item5915a44e2d:g:s04AAOSwCfRbFQTr:rk:2:pf:0 .

[alireza7]

Yes.

The small-signal model looks like this:



where the transformer part is an ideal transformer.

In addition to the facts that you said can we also say since the burst signals have large amplitude ( e.g. 2kv peak, 50ns duration ,5Khz frequency )  the core become saturated and Xo become low (Iu,the magnitization current become high) and the transformer is somehow short circuit and doesn't pass the signal?


And also i think core loss become high at high frequencies so Ro became low. Am i right?

[Jwillis]

I did a little experiment with a 120V to 30V  30VA  laminated  power transformer. I used a frequency generator on the primary and an oscilloscope on the secondary.I injected 10Vpp at varying frequencies to see what the output would look like at the secondary.
 I haven't tested this particular transformer to see the actual ratio but that didn't seem important for this experiment.  The experiment was to see at what frequencies this transformer would hold a stable peak to peak voltage.
The transformer is supposed to have a  4:1 ratio and showed  at the secondary approximately 2.4Vpp  between 13hz and 270Khz on the oscilloscope.From 270khz to 490khz it climbed 2.4Vpp to 5.6Vpp.Above 490khz the Vpp started to drop of .At 1Mhz the secondary only read  0.8Vpp.

I took other arbitrary readings like capacitance of the transformer but couldn't make sense of them at the time.But capacitance seemed to drop off to zero at around 9.3khz.I also took DMM readings on the primary and secondary .Primary held 3.9Vac through the whole  range of frequencies .The secondary showed the highest voltage at at the highest capacitance but also dropped to zero of at 9.3khz.

Anyway I thought it was an interesting little experiment and will look into it more .

[T3sl4co1l]

Yeah, not bad, also not unexpected.  It's not very meaningful without a load resistance -- again, transformers have a characteristic impedance, and best results are had when source and load resistance match their respective windings.

Capacitance measurements are only meaningful at low frequencies, where the windings aren't interfering with the measurement.  And even then, that's only between windings, or winding and core, or shield, or etc.  You can't meaningfully take the self-capacitance of a winding -- you probably need an impedance vs. frequency measurement to fit that.

Tim

[bsfeechannel]

Under full load, the most important parameter in the frequency response of a linear transformer is the leakage inductance, and the cut-off frequency is in the range of a few kHz or less.

[Wolfgang]

Suppose a 220VAC to 110VAC ,50Hz, 20w transformer.

How is it's behavior in higher (1khz up to 10Mhz) frequencies? Does it attenuate signals with high frequencies or not?

Actually i want to know that if using a transformer on the mains line improves EMC of my system and protection against burst signals or not?

Hi,

from a 220V:110V 50Hz line transformer you can expect

- useful power transfer up to ca. 400Hz
- above that iron core losses rise too much so that the transformer gets hot

above ca. 400Hz interwinding capacitances, iron losses and stray inductance spoil the broth. You will see a lot of resonances and an erratic transfer rate and input and output impedance.

If you need a wideband transformer, it is probably some type of injection transformer, like here :

electronicprojectsforfun.wordpress.com/injection-transformers/

These go from a few 10Hz to some MHz and they have a defined amplitude/phase transfer characteristic. The drawback is that only smaller voltages are allowed.

Have fun experimenting
  Wolfgang

[exe]

BTW, rumors ([1]) say a cheap current transformer can serve as a poor man's injection transformer. Never tried myself.

[1] "APPLICATION NOTE 3245" https://www.maximintegrated.com/en/app-notes/index.mvp/id/3245

[Wolfgang]

BTW, rumors ([1]) say a cheap current transformer can serve as a poor man's injection transformer. Never tried myself.

[1] "APPLICATION NOTE 3245" https://www.maximintegrated.com/en/app-notes/index.mvp/id/3245

Its not a rumour. For a limited frequency range, they work fine. See

electronicprojectsforfun.wordpress.com/injection-transformers/

Its also possible to make homebrew transformers as good as commercial ones, like here:

https://electronicprojectsforfun.wordpress.com/injection-transformers/measuring-commercial-and-homebrew-injection-transformers/

[alireza7]

Suppose a 220VAC to 110VAC ,50Hz, 20w transformer.

How is it's behavior in higher (1khz up to 10Mhz) frequencies? Does it attenuate signals with high frequencies or not?

Actually i want to know that if using a transformer on the mains line improves EMC of my system and protection against burst signals or not?

If you need a wideband transformer, it is probably some type of injection transformer, like here :

electronicprojectsforfun.wordpress.com/injection-transformers/

No i need essentially a base band transformer which eliminate surge , burst and conducted RF signals.

[Wolfgang]

If I get you right you want a common mode choke ?

[T3sl4co1l]

You can eliminate those factors with a synthesized sine wave inverter / UPS -- you cannot eliminate them with a transformer.

A transformer can attenuate them to some extent or another, so if you have some acceptable attenuation figure or noise floor or threshold, we can give a meaningful answer.

Tim

[alireza7]

I want to reject both common mode and differential mode signals

For this purpose i designed a circuit like this:



Vout goes to a switching power supply

Do you think this circuit will pass surge and burst and conducted RF tests?

If you don't know what are surge and burst and conducted RF see this website:

[T3sl4co1l]

I want to reject both common mode and differential mode signals

Again, if you do not have a meaningful spec, you cannot possibly hope for a meaningful answer.

Taking the above statement at face value, the best rejection of both common mode and differential mode signals is to short the input of whatever this connects to.  That way all signals -- transients, noise, and mains voltage -- are rejected about as well as can be (>120dB, if a little shielding is provided?).

Tim

[alireza7]

I want to reject both common mode and differential mode signals

Again, if you do not have a meaningful spec, you cannot possibly hope for a meaningful answer.

Taking the above statement at face value, the best rejection of both common mode and differential mode signals is to short the input of whatever this connects to.  That way all signals -- transients, noise, and mains voltage -- are rejected about as well as can be (>120dB, if a little shielding is provided?).

 
Tim

I read a lot about EMC(surge,burst,...) but my goal is just to designed a protection circuit in order to prevent any  MCU reset or malfunctioning in secondary circuit (which contains micro-controller and analog circuit) ,when we apply burst and surge signals . So unfortunately i don't have any sense to numbers and required specs.