R12 should at the very least be a trimmer and have a bypass with a capacitor (tens of uF un parallel with a smallr foil cap), but preferably it should be a Vbe multiplier and mounted on the heatsink. A Vgs multiplier may work better, though. In both cases bypass with a capacitor.
The reason is that you are modulating your bias voltage, and hence bias current by alowing the current charging and discharging the gates of the MOSFETs to produce a variable (and not at all linear!) drop across R12. Not only does it vary with amplitude, but also with frequency. It introduces distortion through this mechanism, where it's trivial to avoid it.
Secondly, you have no thermal compensation of bias current. Don't be fooled by literature that states MOSFETs do not need it - all but the very oldest lateral MOSFETs will exhibit thermal runaway if not thermally compensated. You might want to find the temperature parameter in your simulator and try it for yourself, you may be in for a surprise.
The reason is that you are modulating your bias voltage, and hence bias current by alowing the current charging and discharging the gates of the MOSFETs to produce a variable (and not at all linear!) drop across R12. Not only does it vary with amplitude, but also with frequency. It introduces distortion through this mechanism, where it's trivial to avoid it.
Secondly, you have no thermal compensation of bias current. Don't be fooled by literature that states MOSFETs do not need it - all but the very oldest lateral MOSFETs will exhibit thermal runaway if not thermally compensated. You might want to find the temperature parameter in your simulator and try it for yourself, you may be in for a surprise.
For the new(er) hitachi and toshiba MOSFETs, this is true for lower quiescent currents. They can be raised higher if appropriate source resistors are included. For output stages using those, a Vbe multipier will actually overcompensate for any bias current. However, not using anything still may be a problem if you wish to experiment with the amp. Forgetting that Iq should be 100mA or thereabouts can lead to 4 dead MOSFETs... and if unlucky, to dead speakers. Don't take my word as attempts to preach, it was just a voice of caution.
OTOH, trying no thermal compensation with standard IRF/IRFP parts in regular class AB outputs WILL produce runaway... learned that the hard way
BTW just noticed the gate damper resistors... LMOS have relatively low gate capacitance, but 680 ohms seems a bit high to me...
OTOH, trying no thermal compensation with standard IRF/IRFP parts in regular class AB outputs WILL produce runaway... learned that the hard way
BTW just noticed the gate damper resistors... LMOS have relatively low gate capacitance, but 680 ohms seems a bit high to me...
thank Ilimzn,
- Now I reduce output Iq to 100mA (in simulation, THD is lowwest when Iq =50mA).
- the output iq is very stable though it don't have thermal compensating.
- add two A733 to cascode the different stage.
- add 2.2uf parallel with R12.
The sound is better than before.
all comments are welcomed!
huy
- Now I reduce output Iq to 100mA (in simulation, THD is lowwest when Iq =50mA).
- the output iq is very stable though it don't have thermal compensating.
- add two A733 to cascode the different stage.
- add 2.2uf parallel with R12.
The sound is better than before.
all comments are welcomed!
huy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.