I've started to doodle around with a case idea for this amp.
This is my general idea(borrowed from the old QSC MX1500 internal construction):
http://i.imgur.com/YUYEE.jpg
Those heatsinks are 20cm * 11cm * 3.3cm and are thus a bit small for this amp in its full form.
I am trying to find two heatsinks that are 24cm * 12cm * 5-6cm instead, and use these for another smaller amp.
This is my general idea(borrowed from the old QSC MX1500 internal construction):
http://i.imgur.com/YUYEE.jpg
Those heatsinks are 20cm * 11cm * 3.3cm and are thus a bit small for this amp in its full form.
I am trying to find two heatsinks that are 24cm * 12cm * 5-6cm instead, and use these for another smaller amp.
Im seriously starting to consider abandoning this all NMOS design again, been simulating some more and it seems the only way to get below 0.1% THD at 20kHz is to severly overdrive VAS way beyond its specs, choosing a beefier transistor to get back within spec just brings THD back up to 0.2-0.5%.
Atleast i seem to get the best results with the basic blameless topology.
Atleast i seem to get the best results with the basic blameless topology.
Rather than concentrating on THD, simply ensure the majority of the distortion is H2 and H3.
You might try using a buffer, a simple EF, on the VAS.
Hugh
You might try using a buffer, a simple EF, on the VAS.
Hugh
The darlington VAS does literally no difference, the best results is with a 2N5401 punished to over 1W dissipation, i think my edition of LTspice is broken, it keeps locking up every so often where the simulation halts and cant be stopped, it does allow me to save though before restarting.
In real life the amp does sound pretty decent and since many mosfet ppl prefer the quasicomplementary topology over the full complementary ones, i blame LTspice and the models used for the crappy simulation results.
Or maybe its just me missing something.
My P channel amp on the other hand happily simulates just as low THD as full complementary designs, it also sounds really good, actually rivals my lateral mosfet amp which is the best sounding amp i own, which i find odd as a full p channel output stage should be no different than a n channel output stage.
In real life the amp does sound pretty decent and since many mosfet ppl prefer the quasicomplementary topology over the full complementary ones, i blame LTspice and the models used for the crappy simulation results.
Or maybe its just me missing something.
My P channel amp on the other hand happily simulates just as low THD as full complementary designs, it also sounds really good, actually rivals my lateral mosfet amp which is the best sounding amp i own, which i find odd as a full p channel output stage should be no different than a n channel output stage.
Maybe we are talking at cross porpoises......
Use the 2N5401 as VAS by all means, but run its collector to a 330R strung between the base and emitter of an EF, such as BF469 (or 2SC2682 if you can find it).
Collector of the C2682 is taken to the negative rail, emitter to the CCS that feeds the VA stage. You then have a highly linear, low current VAS with powerful drive from a cascaded EF.
Quasi is preferred over complementary by those who like a little H2/H3 in their music. THD is much higher, of course.
Cheers,
Hugh
Use the 2N5401 as VAS by all means, but run its collector to a 330R strung between the base and emitter of an EF, such as BF469 (or 2SC2682 if you can find it).
Collector of the C2682 is taken to the negative rail, emitter to the CCS that feeds the VA stage. You then have a highly linear, low current VAS with powerful drive from a cascaded EF.
Quasi is preferred over complementary by those who like a little H2/H3 in their music. THD is much higher, of course.
Cheers,
Hugh
I ususally run the collector if the first transistor in a EF(darlington) to ground through a 470 ohm resistor because i happen to have alot of those, and 1-3k for the B to E resistor on the second transistor.
something like this: http://i.imgur.com/nFSEy.png
something like this: http://i.imgur.com/nFSEy.png
- Status
- Not open for further replies.