Is it not the same power dissipation at the same bias/rails for both types of devices, resulting in a same temperature increase. Using darlingtons would demand thermal compensation of some sort, laterals can handle without.
Good choice.
Yes, agree about input stage. Synergy, am I missing somehing here hehe
What if leave +/-32V and lower the bias, it would be fine to have more output swing headroom here. Distortion profile wouldn't change dramatically by lowering bias to let's say 1,5 A?
What about cascoded inputs? Maybe in this case you wouldn't need extra CCS for bias/offset, as sound would stay pretty much the same regardless input biasing.
Attachments
I set up the input stage to have constant power. Basically it is Baxandall pairs cascoded with J-Fet´s, parallel symmetric so it can be called a cascoded band gap reference. The signal path is as short. It goes now through the J-Fets to the LatFets and the feedback is sensed by the Baxandall pairs.
I just wanted to say that there are better solutions then the lone bipolar pair. Even that version plus CCSs sounds great to begin with. I really like the basic idea of post 1.
I personally think that i can not hear better then 0,1% low order distortion so to make the circuit more complicated to get PPM distortion is not so attractive to me when it looses the simplicity.
I just wanted to say that there are better solutions then the lone bipolar pair. Even that version plus CCSs sounds great to begin with. I really like the basic idea of post 1.
I personally think that i can not hear better then 0,1% low order distortion so to make the circuit more complicated to get PPM distortion is not so attractive to me when it looses the simplicity.
You still change the working point of the transistor.
Even if the hFE of the transistor is relatively flat. it will start to fall at above 2 - 3mA. So if your current starts to get above 2 - 3 mA because of offset and bias adjustment the transistor is not that linear any more.
http://www.nxp.com/documents/data_sheet/BC859_BC860.pdf
Look on page 5.
So i would still use the separate bias and offset circuit. But adding the cascode to the front will raise the bandwidth of the input pair and lower distortion due to crossover.
Even if the hFE of the transistor is relatively flat. it will start to fall at above 2 - 3mA. So if your current starts to get above 2 - 3 mA because of offset and bias adjustment the transistor is not that linear any more.
http://www.nxp.com/documents/data_sheet/BC859_BC860.pdf
Look on page 5.
So i would still use the separate bias and offset circuit. But adding the cascode to the front will raise the bandwidth of the input pair and lower distortion due to crossover.
WOW it sounds appealing, would love to see it if you can make a sketch.
Agree, we stay within two amp stages like it is now, only to convert each of them to max performance.
OK you got the point, also Gerhard with Baxandall pairs cascoded as inputs. We have to explore both options to choose best one for the input.
