It depends.
The Sonic Imagery Pro Discrete opamps which can sink and source an order of magnitude in Class A are in a league of their own when used in a mixing Bus. They are SMT with all the cool gadgets that monolith Opamps have. You can’t come close to the noise figures.
https://www.sonicimagerylabs.com/pr.../990Enh_Ticha_Docs/990Enh_Ticha_Datasheet.pdf
All monolithic opamps only offer tiny class A before class B with other distortion cancelling mechanisms. But the main issue is shift in measurement and linearity when they get warm under load. This all leads to higher than ideal feedback impedance and noise. Monolithic opamps are subjectively better biased 600 ua class A and buffered in SE in class A with three transistors. If you must drive a 600R load just increase the buffer bias current. Then you gave a small headphone amp.
On the hifi sonics website pages there is this guide
https://hifisonix.com/wp-content/uploads/2020/02/Class-A-opamps.pdf
When applied with the right way to decouple PSU power supply line noise the results are approximately very good.
This is why chip rolling doesn’t necessarily prove anything when a holistic approach needs to be taken with PSU noise, layout and decoupling. If a particular design hasn’t been optimised for a particular op amp your really just going around in circles chip swapping.
If you calculate, simulate then measure it may take a bit of time but you will be rewarded and save a lot of time.
https://www.sonicimagerylabs.com/pr...nh_DiscreteOpAmp_LV_VariantDatasheet_RevA.pdf
You can come close to the noise performance, the AD797 definitely comes close. The high current is useful, you can reduce impedances for lower Johnson noise, it goes with the low voltage noise.
However is it that practical? - that high current consumption means heat generation, and fans aren't a great idea for audio obviously.
However is it that practical? - that high current consumption means heat generation, and fans aren't a great idea for audio obviously.
Has anyone in this thread attempted paralleling opamps with a 10R load sharing resistor and bias of the op amp?
From a theoretical perspective using a simulation paralleling the opamps in a quad or two dual packages would improve noise performance and linearity with low impedance feedback networks and loads. The cost would not be significant for a diy project.
The subjectivists could then enjoy the sweetness of SE bias in useful amounts up to 1.4 volts peak into a load of 600R (including the feedback network) with 2.4 ma SE
But in the real world a 1.4 volts peak will be clipping the power amp or headphone amp.
What’s more important ordinarily is the switching commutation of the NPN and PNP output transistors making its way into the signal ground from the supply rails at typical
The SE biasing strategy of the opamps significantly reduces this dreaded problem in practical pcb layouts.
To improve the residual noise in the real world the opamp can be locally decoupled with 220R and 220-1000 uF forming a useful low pass filter. Once the op amp is SE biased the stiff supply rails become un important because the load is constant.
Yes the overall supply current is increased however in small signal circuits this is easily dealt with 0.5 amp or 1.00 amp regulators.
From a theoretical perspective using a simulation paralleling the opamps in a quad or two dual packages would improve noise performance and linearity with low impedance feedback networks and loads. The cost would not be significant for a diy project.
The subjectivists could then enjoy the sweetness of SE bias in useful amounts up to 1.4 volts peak into a load of 600R (including the feedback network) with 2.4 ma SE
But in the real world a 1.4 volts peak will be clipping the power amp or headphone amp.
What’s more important ordinarily is the switching commutation of the NPN and PNP output transistors making its way into the signal ground from the supply rails at typical
The SE biasing strategy of the opamps significantly reduces this dreaded problem in practical pcb layouts.
To improve the residual noise in the real world the opamp can be locally decoupled with 220R and 220-1000 uF forming a useful low pass filter. Once the op amp is SE biased the stiff supply rails become un important because the load is constant.
Yes the overall supply current is increased however in small signal circuits this is easily dealt with 0.5 amp or 1.00 amp regulators.
That's a good point about class-A output biasing, reducing assymetric supply-rail currents that might pollute the signal ground (via decoupling capacitors). Not sure this point has been made before - anyone? Though opamp level currents are usually too small for such rail injection to be an obvious problem.
Mark if you refer to page five - six here and explanation with measurements is provided. He has a solid track record in small signal audio electronics.
This can’t be simulated and it’s where a though measurement analysis at the first pcb layout can identify these things.
With SMT with a four - five layer pcb there is far more flexibility to optimise and reduce loop areas around the opamp, positioning de-coupling caps most effectively while having a low impedance power supply ground plane back to the star ground of the PSU.
A separate signal ground plane then provides a pure and clean ground for ac signals.
Locslised LDR could also be used instead of RC decoupling networks in situations where class A operation
Is not appropriate.
Think this is where a diy project comes into its own if you have the time fully explore what’s going on
In the assembled pcb.
https://hifisonix.com/wp-content/uploads/2014/11/Part-1-Ovation-Symphony-Line-Preamplifier-V1.0.pdf
This can’t be simulated and it’s where a though measurement analysis at the first pcb layout can identify these things.
With SMT with a four - five layer pcb there is far more flexibility to optimise and reduce loop areas around the opamp, positioning de-coupling caps most effectively while having a low impedance power supply ground plane back to the star ground of the PSU.
A separate signal ground plane then provides a pure and clean ground for ac signals.
Locslised LDR could also be used instead of RC decoupling networks in situations where class A operation
Is not appropriate.
Think this is where a diy project comes into its own if you have the time fully explore what’s going on
In the assembled pcb.
https://hifisonix.com/wp-content/uploads/2014/11/Part-1-Ovation-Symphony-Line-Preamplifier-V1.0.pdf