Magnetising current in transfomer

[bowman1710]

Hi Guys,

I am currently looking at a transformer design and I am a bit unsure on what is optimum for the design , the waveform on the transformer will be a short pulse of square waves ranging from 2KHz to 20Khz, maximum pulse length 50ms every 15% rep rate.Design specs are

Vin-200V pk
Vout 2kV pk
Iout-0.42A rms
Iin-4.243A rms
np/ns:1/10

Now I have already have looked at the minimum turns for this for it not to go into saturation but the magnetizing current from what I have worked out is very large! so my questions are:

1) I am unsure what the magnetizing current would be given this pulse waveform, whether the pulse waveform and rep rate effect this?
2) How do you determine when lowering the magnetizing current becomes non beneficial and you start losing more through copper losses etc?

 

[srizbf]

the selection of the core decides the maximum fluxdensity .

a suitable core is arrived at first from manuf datasheets.

 

[FlapJack]

"Vout 2kV pk
Iout-0.42A rms"

First i will state the obvious. That transformer will be a real man killer.

I believe that the core will have to be large because of the voltage, current and power issues. The larger the core the lower you typically have to run the Gauss level, because large cores hold the heat in. Your only saving grace is the 15% duty cycle, but i do not have enough experience to know how much that will help you.

Modern ferrites are in the range of 2800 to 3200 Gauss. If no one else has a better answer, start with 1600 Gauss, then measure the core temperature after 2 to 3 hours of operation to see were you are.

 

[FvM]

15% duty cycle means that you can design the transformer for 40% of the intended output current, 335 VA instead of 840 VA.

I presume that cut-tape is preferred over ferrite for a compact transformer with 2 to 20 kHz frequency range, ultimately a modern amorphous metal core.

 

[bowman1710]

That transformer will be a real man killer

Yeah I know with the core I am looking at either a ferrite or an advanced powder really im having to analyse the what is optimum at the moment but i think it will likely be the advanced powder core. Any idea's on the magnetizing current issue I have brought up with how the magnetizing inductance effects the overall design?

 

[FvM]

I believe that cut-tape or even standard 0.35mm laminated sheet is more appropriate than powder core. Just my 50 cents.

Magnetizing current must be handled by the driver amplifier. Not more, not less.

 

[Warpspeed]

I agree, cut tape would be well worth thinking about.

Two advantages, you can run grain oriented silicon steel up to 12,000 gauss easily without worrying about saturation at lower frequencies.
And the permeability is pretty high, lots of inductance and low magnetising current.

What you need to watch out for is eddy current loss, but because of the reduced duty cycle, core heating should be much less of a problem.

The other evil to avoid is a too low self resonant frequency.
2Kv means quite a few turns, and significant distributed capacitance.
It absolutely must have a much higher self resonant frequency than 20 Khz, and that may be quite difficult to achieve with what you are trying to do.

I think the solution may be a fairly large tape wound toroid, so large that fewer turns will be required, and sufficient secondary turns can fit onto a single well insulated layer. That will produce the highest possible self resonant frequency, which I suspect may the greatest hurdle to overcome.

And I believe eddy current loss will turn out to be greater than magnetising current loss, but the final effects of both will be fairly similar.

 

[FvM]

I agree that winding capacitance and respective self resonance of the 2 kV winding is the real problem.

 

[FlapJack]

Z winding on a sectioned bobbin greatly reduces capacitance, i think by a factor of 4 to 6 over regular layers. I do not know very much about cut tape but it sounds expensive to custom wind each core. Lamination's are always expensive and can they do 20 khz.

- - - Updated - - -

BTW, your original question about Magnetising current in a transformer is a little strange. I think that is why you are not getting any answers about it directly. As a general rule you pick a transformer core with enough room for your winding's and then run the highest flux you can without exceeding a 60 deg C rise (for ferrite). This process ensures the smallest and hopefully cheapest core.

Rethinking ferrite as a choice, 2khz is very low frequency. Now i think i see why you are talking about tape wound cores and lamination's.

 

[Warpspeed]

Ferrite is very good stuff at higher frequencies, but the biggest limitation of ferrite is the very low flux saturation level.
At 2Khz and 2Kv it will require a HUGE number of turns which creates a whole series of other major problems. Its exactly why there are no ferrite audio transformers made. A ferrite core large enough would be prohibitively expensive.

To a certain extent you can trade off the number of turns for core cross sectional area, fatter core = fewer turns.
This should really help with the self resonant frequency problem.

Winding toroids by hand is a real pain, but once you have your prototype working, automated machine winding is now commonplace and quite economic for commercial quantities. The tape wound silicon steel cores are now produced in numbers that make them economic even for mains frequency power.

Making the very first one yourself, will be an ordeal.
Getting quotes to have a hundred or more toroids wound commercially is not a big deal, and should be fully competitive with other alternatives construction methods.

 

[Warpspeed]

One further thought...

The only application I can think of for something similar might be the design of an audio transformer for driving electrostatic loudspeakers.

You could try a literature search (and sniff out the internet) to see how others have approached these same or similar problems.

 

[FvM]

The only application I can think of for something similar might be the design of an audio transformer for driving electrostatic loudspeakers.

The parameters in post #1 don`t suggest an audio application.

I think, whatever the application is, the discussion has shown that the transformer design problems are on a different field than expected by bowman1710.

 

[Warpspeed]

The parameters in post #1 don`t suggest an audio application.
.

True, but 2Khz to 20Khz does put it into the audio spectrum.

 

[Dan Mills]

Sonar?

This is exactly the sort of thing you see used to drive big piezo ceramic stacks (But there the leakage is usually a designed parameter to tune out the stacks fixed capacitance).

And yea, tape wound toroids are fairly commonplace, most transformer companies will wind you a custom one off for a price that is not too horrible for a prototype, do not try to hand wind this.....

Regards, Dan.

 

[D.A.(Tony)Stewart]

Hi Guys,

I am currently looking at a transformer design and I am a bit unsure on what is optimum for the design , the waveform on the transformer will be a short pulse of square waves ranging from 2KHz to 20Khz, maximum pulse length 50ms every 15% rep rate.Design specs are

Vin-200V pk
Vout 2kV pk
Iout-0.42A rms
Iin-4.243A rms
np/ns:1/10

Now I have already have looked at the minimum turns for this for it not to go into saturation but the magnetizing current from what I have worked out is very large! so my questions are:

1) I am unsure what the magnetizing current would be given this pulse waveform, whether the pulse waveform and rep rate effect this?
2) How do you determine when lowering the magnetizing current becomes non beneficial and you start losing more through copper losses etc?

1kVA at 20kHz is no problem.
NASA has been using 100kW 20kHz power supplies that are lower cost and weight for a long time.


here is one method for 10kVA

• Vin= 587V (quasi-square wave).
• Vout=50kV
• Power rating 25kVA.
• Switching frequency 20kHz.

In order to accommodate two bobbins (one for the low voltage and one for the high voltage winding) a rectangular core was used with a 50 mm 50 mm cross-section. This was made from 100 mm 25 mm 25 mm sections, since they were the largest available; these were ground to give good fits between the sections. To ensure low lossees the flux density was limited to 0.25 T.for ferrite. ( for Silicon Steel laminate, limit to 1 or 1.3T max for this not 8T ( expensive))

Two primary windings in series, each containing 13 turns of AWG7 wire giving a primary copper loss of 4.6 W.
• Two secondary windings in parallel, each containing 700 windings of AWG21 wire.
• Total losses (copper and powdered “iron”) of 510 W for 25KW out or 2% loss.

 

[Warpspeed]

1kVA at 20kHz is no problem.

Designing for 20Khz with ferrite is pretty straightforward.
But that 20Khz transformer design is not going to work at 2 Khz.

 

[bowman1710]

Hi guys been away and only just seen all the replys i'm amazed!

To start with from FVM and warspeed

I believe that cut-tape or even standard 0.35mm laminated sheet is more appropriate than powder core

I agree, cut tape would be well worth thinking about.

I've never heard of cut-tape before?? Are you referring to something similar to this or am I way of the mark???

https://www.mag-inc.com/products/tape-wound-cores

I didn't think of the self resonant issue so thank you for bringing that up!!!!

Magnetizing current must be handled by the driver amplifier

Could you explain in a bit more detail about this, I am unsure what you mean by handling the magnetizing current, and is this the same for eddy currents?

The only application I can think of for something similar might be the design of an audio transformer for driving electrostatic loudspeakers.

Thanks i will look into that!

 

[Warpspeed]

Tape wound cores are now very common, and as time goes on are slowly displacing the old stamped soft iron E and I laminations.

The most frequently used material is 0.35mm thick grain oriented silicon steel strip, which is wound on a machine under extreme tension onto a mandrel which can be round, square, or rectangular.

The round cores are sold as toroids, and the rectangular ones are usually sawn in half to produce two U shaped core halves, which are subsequently clamped together around the completed winding.

The silicon steel 0.35mm cores are the most common material, but other materials and thickness’s are available for very special applications, but can become quite expensive in larger sizes.

Once you have wound your transformer, the primary will have some value of inductance. When you connect it across a voltage source, current will flow depending on the applied frequency and voltage.
That is the magnetizing current.
It will decrease directly as the frequency increases, and is one part of the no load primary current.
Because it is an inductance, the current will lag the voltage by ninety degrees.

The will also be eddy current loss due to circulating currents in the core.
This acts somewhat like a partial shorted turn, and causes an increase in primary current which looks something like a resistive load. Eddy current loss is resistive in nature, the primary current will be in phase with the voltage.

Eddy current loss does however increase with frequency, and the rate of increase is exponential as frequency rises.

So there are two types of no load loss.
One is out of phase and decreases linearly as frequency rises.
The other is in phase and increases exponentially as frequency rises.

There will be a magic frequency of minimum no load current which is often around 350 Hz for the common type of cut tape wound cores.
As frequency rises the no load loss rises increasingly steeply purely from eddy current loss.
That can be reduced by lowering the design flux density.