Frequency converter: IGBT questions

[JStuffer]

switching frequency of igbt

Hello, i'm designing a frequency converter and have some questions on it's H-bridge part. It's going to be 1-phase 230V converter, PWMed sine wave with output frequency between 0 to say 500Hz, modulation frequency about 100kHz. The high-rail potential turns out to be 325V. The power rating is not yet determined.

Now, i'm planning to make the high-side semiconductors switch only at the low output frequency and the low-side ones do additionaly the pulse-width modulation. I'm going to use IRG4PC40W which is fast (max. 150kHz hard switching) IGBT for the low-side and slower & weaker IRG4BC20U (max. 40kHz (!)) for the high side.

I wonder if you can help me answer some questions:
1) When the low-side IGBT switches at such a high frequency and the corresponding high-side one is held conducting, will it's slower switching capability impact the fast currents or will it just pass-through as it's meant to?
2) Is the desired modulation frequency of 100kHz a reasonable value?
3) When the sine wave signal for the PWM reaches it's peak and the IGBTs are possibly held conducting continuously for some while won't they go aflame at moderate loads? I'm not sure how an IGBT operates - the datasheets provide only a pretty wide turn-off SOA.
4) Could you foreshadow the final product's power rating or it's limiting factors? Like something i should bear in mind, bottlenecks and stuff? The IGBT's are rated at 600V, 13A for high-side, 40A for low-side (current rating halved at high temps).
5) Or should i stop tempting fate and buy (aaargh $$$) additional two of the faster and tougher IGBTs?

I'm attaching an image of the h-bridge if my descriptions were any cloudy or something . Protection diodes, filters and stuff will go to their appropriate places.

Thanks for your time.
John

Edit: fixed a typo and a device's name

 

[FvM]

igbt converter

The diodes in an IGBT H-bridge aren't just protections diodes. They must be supposed to conduct full switching current in PWM operation.

As another point, your intended PWM scheme doesn't allow 4-quadrant operation, required e.g. to handle the apparent current of an inductive load.

 

[JStuffer]

igbt frequency converter

Well, FvM has a good point here.
The reverse-parallel diodes are to be selected later.
But the major flaw here is in the fact that the high-side remains on even at times when low-side switches off. In case of inductive loads there is current path from high-rail through the load into ground, which is chopped by the pulse width modulation - but only on the low side. And as the load is inductive, it shoots some emf currents at the turn-off times. This is a problem because the high side is on and creates loop with the high-on-transistor and the high-off-one's reverse diode. This current is limited by voltage drops in the semicoductors, some parasitic resistances and by nothing else.
The solution to this is (i hope i'm not wrong here) to modulate both the low and high side and hope that these will be somewhat synchronized . The switching losses will increase, as will the circuit complexity.
I suppose this answers my previous questions number 1 and 5, it no longer seems acceptable to use the IRG4BC20Us (i made a mistake in the previous post, it's '20U, not '40U).

I still need an advise at questions 2, 3 and 4, however. If you have an opinion on this matter, hesitate not and post what's on your mind!
Thanks.

John