Selecting Resonance Tank Capacitor

[mrinalmani]

I designed a high frequency transformer (100kHz) for an inverter (12V/230V, 800VA).
The leakage inductance turned out to be approximately 1uH. Now, although small (or is it larger than typical?), the reactance when referred to the primary turns out to be 600 ohms @100KHz, This is 10 times larger than the load resistance at full load!!
So I decided to add a capacitor at the secondary side with an impedance of -600 ohm. This comes out to be approximately 2.2nF.
Since the capacitor must carry the load current in series (≈3.5A) the voltage across the capacitor turns out to be a whooping 2.5KV. Also since it is handling 10 times the rated VA, even a slight loss due to dissipation factor will be magnified 10 times.

1. Please suggest an appropriate capacitor type.
2. What type of capacitors are generally used in resonance tank?
3. Is this approach of adding a series capacitor a standard practice, or do there exist other better ways to reduce the effect of leakage inductance?

 

[chuckey]

These voltages and currents correspond to those of a medium wave broadcast transmitter where silver mica capacitors would be used. Very large and expensive! :-(
Frank

 

[FvM]

Firstly I can't follow your calculation. 1 µH translates to 370 µH with transformer 230:12 ratio, or about 230 ohms. But it's too much for a 100 kHz transformer anyway.

Besides the problem of finding a suitable capacitor, you have this issues:
- dielectric strength of winding isolation
- high Q of resonant circuit, respective bad transient behaviour
- due to asymmetrical capacitor circuit, the transformer flux will be probably increased, resulting in higher losses or even core saturation

Realistically, the transformer Xk (inductive voltage drop at rated current) shouldn't be more than some 10 % of nominal winding voltage.

 

[mrinalmani]

230V RMS is actually 325V peak. Now the calculations will lead to a figure around 500 to 600 ohm for a 1:30 transformer.
I know that internal drop should be no more than 10 to 20%. But then this would translate to 10nH leakage. Now this seems like an impossible figure. How are HF transformers optimized. Do they really have such fancy value of leakages, or are there other compensation techniques. I mean I have never heard of a 1KVA inverter that does not work on resonance, has anyone?

And as far as about silver-ceramic capacitors, do HF inverters available in the market actually use such types?
Another question would be, 'why dont we see high power HF inverters in the market?'. Is the design not feasible practically.

 

[FvM]

Commercial 100 kHz inverters hardly use silver mica capacitors.

The resonator design becomes feasible if you manage to bring down Xk (the usual english formular sign is Xeq, I think) at least to 50 %. Then polypropylene HV capacitors can be used.

10 nH leakage inductance is quite small as a number. Other designers have apparently difficulties to imagine wiring with below mOhm resistance as you see in recent related threads. The problem is to think about implementation options.

For the first step, it's probably reasonable to go for less demanding designs, e.g. 30 to 50 instead of 100 kHz.