Current source inverter

Sourcing a amp with a current source makes no sense. I think it's placed there to model some feture rather than being the actual design. The MOSFET's are configured in a H-bridge. You can't avoid the need for deadtime with this configuration.
Placing an inductor on the supply can prevent large shoot through currents, but will intruduce a lot lot of ringing and most likely worsening things rather than improving anything.

You can avoid large deadtimes by selecting proper transistors with low gate charges. FDD3682 or IRF6665 can be recomended.
A deadtime about 10ns will produce arround 0.1% THD. Sinse 10ns deadtime is hard to achieve, theres no way arround a bit of open loop distortion. A self osccilation design controls the outputstge cycle by cycle and removes some of the errors. Feedback should also be applied.
 

Kenshin

Member
2005-01-09 9:51 am
China
1.It appears on a SMPS forum, it isn't a class D audio amplifier at all. But it may be turned into a current mode controlled, voltage output class D by feedback.

2. Since it's a current src instead of voltage src, the inductor may be a big one to produce a constant current in a switch cycle. And the dead time may be "negative".

3. New transistors & integrated gate drivers are expensive and difficult to order. BTW: Could discreet gate drivers be as fast as integrated one?

4. Does the number 0.1% means open loop distortion? How much could a self-osc one reduce distortion? Maybe the loop main pole frequency / switiching frequency?

I have built a low power BJT one with 100ns rise/fall time, but don't know how much THD.

sovadk said:
Sourcing a amp with a current source makes no sense. I think it's placed there to model some feture rather than being the actual design. The MOSFET's are configured in a H-bridge. You can't avoid the need for deadtime with this configuration.
Placing an inductor on the supply can prevent large shoot through currents, but will intruduce a lot lot of ringing and most likely worsening things rather than improving anything.

You can avoid large deadtimes by selecting proper transistors with low gate charges. FDD3682 or IRF6665 can be recomended.
A deadtime about 10ns will produce arround 0.1% THD. Sinse 10ns deadtime is hard to achieve, theres no way arround a bit of open loop distortion. A self osccilation design controls the outputstge cycle by cycle and removes some of the errors. Feedback should also be applied.
 

Kenshin

Member
2005-01-09 9:51 am
China
parameters of FDD3682 TrenchFET :

Resistive Switching Characteristics (VGS = 10V)
VDD = 50V, ID = 32A VGS = 10V, RGS = 16§Ù

tON Turn-On Time 83 ns
td(ON) Turn-On Delay Time 9 ns
tr Rise Time 46ns
td(OFF) Turn-Off Delay Time 24ns
tf Fall Time 26ns
tOFF Turn-Off Time 75 ns

how to get a dead time of 10ns and keep it reliable?