Switched power rails?? Doesnt seem logical in classD, there is next to no advantage in it due to the rail to rail switching nature of the classD final output stage. In theory the quiescent disipation is very low anyway, unlike other classes where the savings in quiescent disipation justifies having switched power rails.
Leigh
Leigh
Hi PAFI,
do you think, it s not good enough for audio applications becouse of the relative slow correction of the global feedback?
(becouse the same in a class H where the basic topology is class AB, is just quicker, but elswhere very similar)
thanks,
Tamas
Those motor drives are already going to be operating at a low frequency compared to a class D audio amp. Any global feedback employed would be painfully slow, even for a subwoofer amp.
If you wanted to eke out more efficiency in a sub amp (or get away with slow MOSFETs) lower the PWM frequency first before doing anything more elaborate. A sub amp may be quite happy with 10-20 kHz and switching losses should fall dramatically and there would be no need for any rail stepping. Unless of course, you were using those big industrial FET packs, and putting out 50kW (100 hp). You're not thinking about running 20 lab horns off a single amplifier, are you?
If you wanted to eke out more efficiency in a sub amp (or get away with slow MOSFETs) lower the PWM frequency first before doing anything more elaborate. A sub amp may be quite happy with 10-20 kHz and switching losses should fall dramatically and there would be no need for any rail stepping. Unless of course, you were using those big industrial FET packs, and putting out 50kW (100 hp). You're not thinking about running 20 lab horns off a single amplifier, are you?
Hi,
In Fact, I dont want to have 50kW
But maybe something around 4kW @4 ohms
The feedback is not really a problem, becouse you can use local (quick) feedback for the compensating of the rail switching.
but some benefits: lower voltage mosfets could be used (lower Rds on, lower gate charge, etc) and this could make efficiency better and overall part costs lower...
of course there could be used many other techniques used at the same time...
any toughts?
Tamas
In Fact, I dont want to have 50kW
But maybe something around 4kW @4 ohms
The feedback is not really a problem, becouse you can use local (quick) feedback for the compensating of the rail switching.
but some benefits: lower voltage mosfets could be used (lower Rds on, lower gate charge, etc) and this could make efficiency better and overall part costs lower...
of course there could be used many other techniques used at the same time...
any toughts?
Tamas
What I referred to is not really similar to Class H.
http://www.ece.utk.edu/~tolbert/publications/apec99.pdf fig.3
The problems are: it is difficult to drive the gates, it needs many transistor, and the diodes makes loss and distortion higher. The controll method is more problematic too.
If you really want to switch the supply rails of a conventional ClassD amp, then there are other problems. The switchings will go out. Class AB is not only much faster, but its supply rejection is basically much better, even without feedback. ClassD is basically a modulator, while an emitter-follower is a voltage generator, it's operation is essentially independent from supply voltage (as long as supply is significantly higher then output).
And finally: it doesn't worth the effort.
http://www.ece.utk.edu/~tolbert/publications/apec99.pdf fig.3
The problems are: it is difficult to drive the gates, it needs many transistor, and the diodes makes loss and distortion higher. The controll method is more problematic too.
If you really want to switch the supply rails of a conventional ClassD amp, then there are other problems. The switchings will go out. Class AB is not only much faster, but its supply rejection is basically much better, even without feedback. ClassD is basically a modulator, while an emitter-follower is a voltage generator, it's operation is essentially independent from supply voltage (as long as supply is significantly higher then output).
And finally: it doesn't worth the effort.
Hi,
In Fact, I dont want to have 50kW
But maybe something around 4kW @4 ohms
Tamas
4kW at 4 ohms shouldn't be any real problem using conventional class D techniques. You just won't be able to buy a kit or a one-chip-wonder solution. Full bridge, 220-250V rail, 8 state-of-the-art TO-220 PDP-panel HEXFETs ($3 each) and appropriate driver circuts, 100 kHz. No need to resort to anything crazy - just a few months of homework and research.
I am developing and planning to build something similar. Just one DC power supply without reference to ground, which becomes either negative or positive synchronously with the main half-bridge switching states. In terms of charges, Rdson and body diode, halving the Vdsmax requirements for mosfets is truly revolutionary to both distortion and efficiency.
Multi-rail class D
We use variable rail switching in our class D mobile amplifiers.
Saves a lot of heat and also "helps" the MOSFETs
I have also designed a full tracking switchmode supply so one can then hang on a class D or a class B amplifier for ultimate efficiency
Steve Mantz
Zed Audio Copr.
Los Angeles CA
We use variable rail switching in our class D mobile amplifiers.
Saves a lot of heat and also "helps" the MOSFETs
I have also designed a full tracking switchmode supply so one can then hang on a class D or a class B amplifier for ultimate efficiency
Steve Mantz
Zed Audio Copr.
Los Angeles CA
Just cascode some slow high voltage high current mosfet with fast low voltage high current mosfet! Dont forget recovery diode! That work up to 6Kw with +/-280V rail. In each half leg of your bridge, you use low voltage as standard mosfet and with the 12-15V supply of the mosfet driver, you simply feed the gate of the high voltage mosfet. In that way, the switching device never ''see'' more than 10-13V. I doo it with SMD high current mosfet, without heat sink! Nearly all lose pass trough the cacoded high voltage fet and recovery diode.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.