How to get the less distortion with BJT?

[homemodder]

Hi Forr

I would be very interested in those articles, thank you. Half my family live in France and will be here on holidays soon and can translate for me. Its always easier for me to understand topologies looking at simplified circuit diagram as their are sometimes discripencies as how we view a topology. Interesting you mentioning rush cascodes, have played a bit with these but didnt like the sound, the same goes for kulish that i tried earlier today. Will try some more experiments as sometimes its in the implementation. Excellent results can be obtained with the kulish, its a shame about the sound.

In my view perrots circuit does not fully address the Vbe temperature dependance. I havent tried cascomp circuit as it would require a lot of time as i think there are gonna be some difficult stability issues with it. However this new circuit ive seen does not seem to have any problems at all, the very little people that have listened to it say its brilliant, people that are familiar with high end equipment.

My email is alexmiguel33@hotmail.com

Thanks

 

[Lumba Ogir]

homemodder,
well, linearization is not quite a cleanup sweeping out the trash, it`s more about shaping the sound to the ears liking with remaining distortions. When lowering certain distortions, inevitably other ones will be created so many of those circuits do more damage than good, their sonic impact has to be investigated. I know you do that thoroughly, a rare phenomenon these days, people rather prefer to fool around with simulations to find out sound quality...a terrible mistake...

 

[lineup]

here is the schematic,
the signal level are amplitude not rms,
input generator impedance :22kohms,
output impedance :600ohms after a buffer stage not drawn.

Your EF output stage is not inside the feedback loop.
If the output swing is significant, compared to the supply rails,
then this will create some distortion.
It could be improved with a current source
- if you want to keep it outside the feedback loop.

Referng to your schematic, dododio
http://www.diyaudio.com/forums/attachment.php?s=&postid=1553746&stamp=1215109780

The distortion level at the output transistors BASE
is dependent of your degeneration resistors,
BOTH in the input pair and in those PNP.
Lower gain per transistor/stage (higher resistor values) = lower distortion.
More transistors/stages gives more gain and more feedback factor = lower distortion.


As DRC pointed out,
additional distortion is added by the output transistors.
This addition is usually very low, even compared to one Feedback corrected Voltage amplifier.
And will be even lower if using CCS loaded output
and high input impedance for your output devices.

THD gets higher the more load(lower impedance) and the higher signal amplitude.
You adjust the load of your stages + the output to a suitable, optimized load
to get the output you want/need.
And hopefully by this you have as low distortion you can get
from your used circuit.


It is quite a nice amplfier you use, actually

Lineup

 

[unclejed613]

the two PNP transistors are acting as transconductance amps, which means they're taking a current input (from the diff pair) and turning it into a voltage. the voltage waveform on the bases of the PNPs will be a very distorted voltage waveform, but since they're being driven by current, that doesn't matter. the collector resistors for the PNPs should be replaced by current sources, which would lower distortion a lot. you would also improve performance a lot by using a pair of complementary teansistors on the output stage, biased into class A operation. once that is done, this will resemble a bridged amp.

 

[lineup]

current sources
will improve most any amplifier .. at least in terms of THD distortion

sometimes they are NECESSARY .. to get a fair amplifier stage
other times we can use bootstrapping or plain resistors instead
with evenso good result

think unclejed is right
this one case is where current sources would make a whole lot of difference
for the better