I just don't want to interfere as long as the schematic is not finished. Until then, HTb is in charge. Afterwards, I will change to automatic numbering (or my own) for the PCB. Unless he changes....
There is a merit to keep common emitter output stage. It can have almost rail to rail voltage swing. It could be a good amp for low voltage supply.
The more extensive, the more complex, the more high-end flair
You claim that the cause of all evil is the incorrect choice of output stage (emitter circuit, common emitter ...). I disagree with you on this point.
But the A1 is a rail-to-rail servant. That's why it likes to break down so much. The classic is not a rail-to-rail wizard, nor does it have a second thermal breakthrough.
#
We are still lacking absolute operational reliability, or will the upper high-end coating help us? Do we still need a "SoftClipper"?
Good luck,
HBt.
Well, I'm not so sure that we're treating the symptoms and ignoring the cause.
You claim that the cause of all evil is the incorrect choice of output stage (emitter circuit, common emitter ...). I disagree with you on this point.
You have to be honest, we now have a completely different concept - a different set of rules now applies.
You are confusing the terms method and topology here. There are now two completely different circuit concepts. By further improvements you mean nothing more than extensions, for example you can select an EF1, 2 or 3 for your IPS -> VAS -> OPS division. Connect up to 8 EFs in parallel, more than 8 are useless.
But the A1 is a rail-to-rail servant. That's why it likes to break down so much. The classic is not a rail-to-rail wizard, nor does it have a second thermal breakthrough.
#
We are still lacking absolute operational reliability, or will the upper high-end coating help us? Do we still need a "SoftClipper"?
Good luck,
HBt.
No.
Methods to improve THD:
Method 1: Convert to EF ops = change topology to have a better performing ops
Method 2: Keep original topology, add vertical feature (mirror) to increase olg, conceal the errors of the inferior ops to a certain extend.
You played the mirror card already, I still have it.
Last edited:
I have played the circuit based on the OP a little bit. The current sensing feedback scheme breaks when driven into class B region. It senses the sum of the voltage drops of 2 sensing resistors. When it is driven into class B, it senses a even bigger voltage drops. That results the actual bias point drifts lower and lower. That's why the distortion spikes.
If the current sensing scheme hasn't changed, I don't see any point to build another one. You will face the same issue.
If the current sensing scheme hasn't changed, I don't see any point to build another one. You will face the same issue.
The third way
Just as "wahab" suggested through the flower - pulling together and counter-coupling locally; old hat, known as CFP. It works perfectly in operating classes B, AB and especially A.
In no time at all we end up with the Elektor LFA Class A 50Wpeak Amplifier ... but the A1 20Wpeak approach is sufficient for us.
😎
HBt.
Just as "wahab" suggested through the flower - pulling together and counter-coupling locally; old hat, known as CFP. It works perfectly in operating classes B, AB and especially A.
In no time at all we end up with the Elektor LFA Class A 50Wpeak Amplifier ... but the A1 20Wpeak approach is sufficient for us.
Then it's your turn to play your card, I'm looking forward to your hand.
😎
HBt.
It very looks like all attempts to improve the original schematic will lead to a standard design.
I have another candidate that works around this issue. In the case of a standard EF2, it uses VBE multiplier to set the bias.
Similarly, we can use a fixed sum of VBE to bias a common emitter output stage.
Below, V(R1) + V(R2), stays constant. That is how the bias is set. The opamp is used to fix DC offset at the output stage.
The bias is about 200mA in my case.
The distortion @25w is much better. but it is still a little higher than an equivalent EF2.
Last edited: