6000W By IRS2092

[Andrew Eckhardt]

Originally Posted by Eva

No bias at all ensures no cross-conduction. Note that these transistors are driven with a square wave, a logic signal, not with an audio signal, and that bipolar transistors tend to exhibit longer turn-off than turn-on delays. Note also that conventional bias schemes only work well at audio frequencies, not for the ~100ns current transients that these buffers have to deal with.

Yeah most of the time will be spent trying to charge the 15,000pF, after crossover, even with those 15A drivers.

Does anyone know what the switching frequency is? This thing looks like a woofer amp.

[Mark Allen]

And here is a photo of the results of the first testing done on the 6kw power amp at IR's development plant....

[mgm2000ro]

pcb ?

[Eva]

5KW at 2 ohms (20 msec) are achievable with a full bridge of IRFB4227 pairs, IR2113 with buffers, a tight layout and a good active current limiting system. I have done that.

Testing at these power levels is difficult, supply voltage sags and available power is quickly reduced to 4KW or less after 1 sec.

Note that this is neither a project for unexperienced people nor something that you can just build from a schematic.

[sphinx]

Originally Posted by Eva

A full bridge is required to obtain 6KW on a not too low impedance with 200W MOSFET.

Eva, what are the limiting factor(s) when targeting max output with the half bridge implementation? I'm not only referring to this design. Let's assume there's an appropriate buffer between the IRS2092 and MOSFET gates, the MOSFET are matched, inductor is optimal, appropriate power supply, and load = 4 ohms. I'm guessing switching losses will be very high but maybe it can work by reducing the switching frequency?

[Eva]

Sorry, I meant "200V MOSFET"

200V blocking capability limits maximum practical half bridge rails to +/-90V or so, which translate into (90*.707)^2/2=2kw @ 2 ohm

Full bridge doubles output swing resulting in (180*.707)^2/2=8kw @ 2 ohm

Full bridge also allows to use a supply voltage closer to 200V because there is no pumping.

[darkfenriz]

Feedback network seems wrong (no lowpassing of the pre-filter signal)

[wg_ski]

Originally Posted by Eva

Sorry, I meant "200V MOSFET"

200V blocking capability limits maximum practical half bridge rails to +/-90V or so, which translate into (90*.707)^2/2=2kw @ 2 ohm

Ok, then... how about IRFB4229's instead? I did manage to score 85 of them from Farnell (all they had at the time). Enough for a couple of full bridges....

[Eva]

+/-120V rails only give 3.6kw on 2 ohms, not much improvement considering that Rds-on is doubled and reverse recovery charge of body diode is doubled too (from 200V IRFB4227 to 250V IRFB4229).

Note that these figures are very optimistic. There will be a lot of sagging unless the power supply is regulated (which increases complexity, size and part count), resulting in substantially less power. Not to mention that regulating a single rail supply for full bridge is much simpler...

[wg_ski]

But you'll sure as heck get the 6000W at 2 ohms with a full bridge from +/-120. Even if it sags to +/-85 under load. Most real amplifiers do this.

As far as the higher Rds and Qrr of the higher voltage part, it seems you have two choices. Either put up with higher switching and/or conduction losses or put up with it being fragile. Pick your poison. The same goes for any amplifier, really - high fT Jap outputs don't have the Is/b of the old hometaxials.