12V to 230V 50Hz Converter - Output Filter

[rentner]

Hello!

I am working on a car power Converter to drive anything - even transformers. I will compare a triangle wave to a sine wave and drive a ferrite core with the resulting 100kHz PWM Transformer. First question of all: Will it even change the voltage? Or will it always be 325V with just modified duty cycle, or will it change the output Voltage at all? And than... I want to add a Filter on the output - the problem is, that the filter effect depends on the resistance which is different at different power output levels.

How can I do that? And the more important thing: How do I even make the output voltage change over the PWM cycle? If this only works with a filter... then how do I make the filter work at very low power levels like 5W? For a 3 stage Filter I would need about 30mH, which actually is quite expensive, since I want to support 5A of current.


But realy... How Do I make the output voltage change by just having the PWM? That is my most concern.

Please help me to get this done. I cannot find ANY professional Car power converter. Just some cheap circuits, that drive a regular mains transformer with 50Hz square wave which obviusly won't ever work out as planned and is highly uneffective. That's, why I chose PWM. I got the circuit done with only discrete parts and one comparator.

What exactly determins the output voltage through a regular ferrite core switching transformer?

[Odysseus]

To start, let's consider an ideal transformer.  The output voltage of the transformer will be equal to the input voltage multiplied by it's turns ratio.  In a well designed circuit, this assertion will hold true even for a real world, non-ideal transformer.

PWM works by reducing the average input voltage in order to produce a lower average output voltage.  To turn the square wave output into the average DC value, a filter is used.  This should be done with an LC filter, and it's characteristics will not change with load.

[rentner]

The characteristics won't change over load? Than that filter is that, what would convert my PWM to the desired Voltage, right? Well, well, how exactly do I make a good filter especially for only THAT purpouse? There are hundrets or thousands of different filters out there and I cannot seem to find out, which one is THE one.


I was actually going for a 3 Way filter L-C-L. So... how?

[Marco]

How much power do you need any way that normal off the shelf inverters won't work for you?

PS. I think inverters generally use DC-DC to go to 325, then use a full bridge with a special PWM modulation to approximate the 50 Hz wave as well as possible, with a LC filter to make a decent sine out of it.

[rentner]

Marco: My solution is exactly the same - only, the PWM is before boosting the voltage. But a DC-DC Converter is exactly the same - my solution is just different at the fact, that I use a transformer, to isolate it. In general both ways are the same.



I don't want to build this because of a high power requirement, just because of learning something. And what I have to learn is, how to make the filter. It should be about 500W or more. I only need 2 Mosfets for just 5 bucks, to drive that highly efficient. I got Mosfets, which exceed the power handling capabilities of the TO220 package. They can switch over 200A continously and support peaks of over 1kA! They are especially designed for low voltage, high power switching applications, work hell of fast and have ultra low RdsOn of below 2mOhm. The switching devices so are perfect for 10-16V applications (they have Vds of 60V).



Just HOW to realise the Filter on the output? Please help me with that. What do commercial Converters, to filter the output?

[Marco]

You are talking about a ferrite transformer though ...

Can't you just use a 50 Hz transformer? Hell it might do enough low pass filtering by itself.

PS. I think you pretty much have to use a 50 Hz transformer if you modulate the low voltage and then step up the voltage. The input is still going to be 0 or positive for 10 ms and 0 or negative for 10 ms, you can't push that through a 100 kHz ferrite transformer.

[tom66]

You're going to smoke/incinerate your output capacitor of your SMPS trying to modulate the output voltage at 100Hz (50Hz x 2; ripple current is a thing); not to mention, you'll need a minimum load as it won't go to zero under light loads. Or am I missing the question entirely?

[Odysseus]

Let's design a simple 2nd order LCR low pass filter.  For a 500W output power at 230Vrms, then we have an approximate expected load resistance of (230V)^2/500W = 100 ohms, so we'll use twice this for the filter's characteristic impedance, Z, so that we end up with a critically damped filter.  Also, we need to pick a reasonable corner frequency, Fc, let's say 10kHz for 40dB of attenuation (100x) of the switching frequency, 100kHz.

Now we use these two equations to solve for L and C:

• Z=sqrt(L/C)
• Fc=1/(2pi(LC)^2)

Pop it into wolfram alpha (or actually do the math, ha) and we find that L=3.2mH and C=80nF.

Sounds reasonable to me.

Alternatively, you could put the filter before the transformer and re-design it for a reflected impedance of 100ohm*12V*325V=3.7ohm , but then you need a massive hunk of iron to step it up at 50Hz instead of 100KHz.

[Marco]

AFAICS he already needs the big piece of iron ... you can't use AM modulation, a low pass filter is not an envelope detector ... the standard modulation for inverters is zero or positive for 10 ms and zero or negative for 10 ms. The transformer won't reset, if the signal already has a 50 Hz component it needs a transformer capable of dealing with it.

The only way to do it without the lump of iron is to go to 325v first and then do your modulation with your full bridge at that voltage AFAICS.

Or am I just completely lost?

[Odysseus]

Hmm, yes, you're definitely correct about that, Marco. That must be why I needed a transformer with such large inductances in my simulation in order to keep the magnetizing currents reasonable.   

So, there you have it, rentener. Looks like you need a two stage solution to make this work.  The filter part still stands, though.

[Marco]

Also looking around, I don't think critical damping is necessarily a valid design aim ... the LC filter design is a multi-objective optimization problem (further complicated by the potential to use feedback). It will become underdamped at high impedance loads any way, it can't be allowed to be a problem ... it should work all the way from a Watt to 500 Watt.

[rentner]

Ok, that will be much more complicated. PWM the high voltage to 50Hz sine... Then I propably need a high side driver for mosfets.

And I don't think, that you can hook up 1W load only. I think you are safest with a 5-10W dummyload.

And: only 80nF and 3.2mH will work fine? Well, I will soon simulate it in LTSpice first. Never deal with HV with just putting stuff together.

[rentner]

Nice to know! Well, the noise wil rise, since your ear is more sensitive to that frequency.

For audio applications and some other noisy operations it is not good to use that frequency.

[Odysseus]

Ok, that will be much more complicated. PWM the high voltage to 50Hz sine... Then I propably need a high side driver for mosfets.

And I don't think, that you can hook up 1W load only. I think you are safest with a 5-10W dummyload.

And: only 80nF and 3.2mH will work fine? Well, I will soon simulate it in LTSpice first. Never deal with HV with just putting stuff together.

You can get away with just two low-side mosfets if your transformer has a center tap.  And I actually made an LTspice simulation of the circuit before my first post just to check my math.