Post by z3d on Apr 15, 2015 20:27:30 GMT
Let's just say I do not share the concerns about opamps that the 'only discrete is good' lobby has.
There aren't any (?) recordings around in this world where the audio signal hasn't passed at least one opamp anyway.
Most likely an audio signals has passed tens of these 'chips' along the way to the final mastering.
Basically an opamp is an amplifier design similar to a (quite complicated) discrete design but has components in it that are already matched and work together optimally.
There are limitations to the design process such as capacitors that can only have small values and are also somewhat 'voltage dependent'
advantages of opamps:
Small (especially the SMD sized ones)
(mostly) interchangeable with other types (brands) of opamps
standardised pinning (in most cases)
low in power consumption.
Very versatile (can be used for all kinds of circuits)
Easy to design schematics around it.
Internally it can use specialised components (transistors with multiple emitters or collectors) that are not available in discrete components.
disadvantages:
low output power capabilities (not designed to drive low impedance loads, aside from a few specialised opamps).
Can become unstable when not used correctly.
opamps are usually specialised for tasks (low noise, low offset, low input current, high speed, low speed etc) but it is hard to find opamps that do well in all departments.
Usually do not handle capacitive loads that well (unless measures are taken)
People knock them because they feel have to much influence on the sound.
I do not share those concerns unless the opamp used is not really suited for that task but used anyway for some reasons.
It's not the opamps fault but the designers fault (or that of the opamp roller without decent knowledge)
This concern may be partially correct when used for C'Moy types of amplifiers.
Also they get knocked because of their 'open loop' bandwidth which some people claim influences the sound.
There is some 'truth' in this ofcourse because at high gains there is less 'feedback capabilities' available for higher frequencies.
This causes the THD to rise for higher frequencies (as can be seen in datasheets).
The question is whether or not this is of importance.
A higher THD at 100Hz will be much more audible (because the harmonics are inside the audible band) than those for say 12kHz.
The first harmonic is 24khz and the 2nd at 36kHz, hardly audible ...
The concern 'discrete amp advocates' have that circuits need to have an open loop gain that covers the whole audible range is essential for good sound thus is not shared by me nor warranted or 'proven' in well performed listening tests.
The opamps used in the Ember (and Solstice, Polaris) are specialised opamps and only serve as 'followers' (1x gain).
These opamps do not provide any gain (well in Polaris it has a gain of 2x) so there is a lot of feedback and thus output R and harmonic distortion products introduced by these parts are very low and the bandwidth is high.
The limitations of the G1217 designs are solely determined by the used tubes, except for the output current limits, these are determined by the 'opamp'.
They are opamps that are actually designed to provide higher currents than 'normal' opamps (5x more) which is essential to drive low impedance headphones.
We use them because they are small, work better than diamond circuits (although those that use diamond circuits will have to disagree) and work well within and well outside of the audible band.
When a proper layout using ground planes and correct power supply decoupling is used these opamps work great.
because some people have pre-conceived objections against opamps and prefer (single ended) class-A circuits we offer choices.
Sunrise and Horizon are both single ended full class-A and Solstice and Ember are high current op-amp output circuits.
For those wanting to roll tubes and C'Moy output opamps there is the Starlight.
In case of starlight 4 opamp sections are in parallel to increase the output current capabilities.
With 4x NJM4556 the output current capabilities match those of Solstice.
I do not have an op-amp phobia nor problems using them where it is needed or simplifies things and as long as the right op-amps are used in the right applications under optimal conditions (power supply and PCB trace routing wise as well as signal path wise) there are no reasons NOT to use them.
As mentioned before... I do not share the concerns that 'discrete amp only guys' have against these (internally very complex) op-amps.
Of course, when opamp rollers start sticking not that well suited op-amps in their circuits and then make claims based on not properly functioning circuits my eyebrows lift a little and have to suppress any urge to comment about it.
That makes sense to me; thank you for sharing your thoughs Frans!