• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Can op amps now be as good as discrete circuitry?

Please study Sam Groner's tome: There you can see several discrete opamps vs. many classic IC opamps. The discrete opamps, including the 990 are not great. Dean Jensen's 990 actually uses IC technology by using a dual transistor LM394 as the input long tailed pair - that's one reason it works as well as it does. Groner's study is now a little dated as some of the latest SOA opamps are even better that any he measured. The main takeaway I took from Groner's work is that no matter what opamp you use, the connection with significant common mode signal is always worse that the connection with none. Inverting is always lower distortion, but higher noise, than non-inverting. I therefore always use the inverting infinite-gain-multiple-feedback topology over the Sallen-Key configuration in line level filters (as in a crossover).
 
Your ears, definitely not some paper or someone statement.
What controls did you use to make sure that it’s ears only?
 
The main takeaway I took from Groner's work is that no matter what opamp you use, the connection with significant common mode signal is always worse that the connection with none. Inverting is always lower distortion, but higher noise, than non-inverting. I therefore always use the inverting infinite-gain-multiple-feedback topology over the Sallen-Key configuration in line level filters (as in a crossover).
If, however, you do want to use to use the configuration fraught with common-mode voltage, there is quite a bit of choice among parts with excellent common-mode input capacitance linearity these days, notably from TI.
I don't think we'd be seeing the current crop of super-high-performance DACs and such without their parts. Which, mind you, tend to be priced accordingly (the bipolar OPA1612s are actually up there with JFET input parts like OPA2132). Well, I suppose OPA1688 and OPA1678 are inexpensive enough (the former being common as a headphone driver and the latter having cropped up in low-midrange audio interfaces in recent years), and 1642, 1652 and 1656 are not totally breaking the bank either.
 
Not all discrete opamps are low distortion. Below we can see Burson with a lot more distortion than Sparkos and Sonic Imagery.

 
If, however, you do want to use to use the configuration fraught with common-mode voltage, there is quite a bit of choice among parts with excellent common-mode input capacitance linearity these days, notably from TI.
I don't think we'd be seeing the current crop of super-high-performance DACs and such without their parts. Which, mind you, tend to be priced accordingly (the bipolar OPA1612s are actually up there with JFET input parts like OPA2132). Well, I suppose OPA1688 and OPA1678 are inexpensive enough (the former being common as a headphone driver and the latter having cropped up in low-midrange audio interfaces in recent years), and 1642, 1652 and 1656 are not totally breaking the bank either.
Well you are correct and common mode rejection is very important to get right for lowest distortion. It takes more transistors (to get CM distortion great) and discrete opamps are unlikely to throw those parts in. It's much better to have a good process in the hands of a good designer with essentially no parts limit - which TI, Burr-Brown definitely have. The latest generation of opamps from those guys are basically category killers, end game opamps. The discrete opamps are just snake oil for 30V systems. Of course power amps are just big boy opamps, but they run from > +/-70 v. Best to make those discrete for power dissipation and voltage reasons.
 
Ears are the only evidence
So - no evidence at all then.

Ears (or more, our total auditory system - including the part between them) are certainly unreliable as arbiters of truth.
 
So - no evidence at all then.

Ears are certainly unreliable as arbiters of truth.
Ears are. Eyes and preconceptions are the problem, which is why we use ears-only controls.
 
Ears are. Eyes and preconceptions are the problem, which is why we use ears-only controls.
A subtle but important distinction - thanks
 
  • Like
Reactions: SIY
Dean Jensen's 990 actually uses IC technology by using a dual transistor LM394 as the input long tailed pair - that's one reason it works as well as it does.
This specific long tailed pair is the circuit used for summing in an SSL4000. SSL actually uses 5534 after the long tailed pair, unlike 'discrete' 990, and people seem to prefer SSL.
 
Last edited:
The question correlates to what exactly? In terms of audio/hifi applications, OPs are easy to use and implement but can't achieve sonical performance against properly designed discrete designs. In low/mid level segments, most designers go with OP but on higler level only discrete and tubes :)
Finaly, one can't change resistors nor caps that are already in OP and most of them are influencig a sound in a bad way (for sake using cheapest poossible materials and size).
Try to get Cello Palette and compare to any modern OP exotic design pre amps, and listen :)

Hmmmm
 
The question correlates to what exactly? In terms of audio/hifi applications, OPs are easy to use and implement but can't achieve sonical performance against properly designed discrete designs. In low/mid level segments, most designers go with OP but on higler level only discrete and tubes :)
Finaly, one can't change resistors nor caps that are already in OP and most of them are influencig a sound in a bad way (for sake using cheapest poossible materials and size).
You appear completely unaware of modern semiconductor process capability, Process Design Kits, and opamp manufacturing. Please read this whitepaper describing just a few of the ways opamps achieve passive component matching that far exceeds discrete:

Characterizing these integrated components as cheap is really naïve. On-die resistors achieve much tighter tolerance, much lower within die variance, exceeding discrete by wide margins, and further benefit from the various trimming processes mentioned in the whitepaper. No amount of hand matching discrete components can reproduce the accuracy and precision. Capacitors and inductors have similar benefits. And if you think the films and pattern tolerances in an opamp or any modern semiconductor are somehow cheap, you really haven't studied. Inductors of massively high Q not feasible in discrete component, and much tighter tolerances than you know. Advanced packaging allows even high-value inductors. People make single-electron capacitors in semiconductors, all the way to ultra high density on-die capacitance with layout characteristics that cannot be reproduced in discrete. You really laid an egg, know nothing about the process technology and materials involved.

Not to mention, semiconductor processes used to manufacture modern opamps come with Process Design Kits that explicitly model the whole process including parasitic, inclusion of thermal sensors. etc. This is how we got high speed IO, with per-lane fused parameters, operating at bandwidths many orders of magnitude higher than audio. These enable designs not possible using discrete, even if you like to imagine some old wise dude sitting around hand selecting discrete stuff from parts bins can end up with a PCIe interface. Perhaps a few too many Iron Man movies for you! I liked the stone tools and bearskins episode for Star Trek:
1726637420945.png

but we need to not take this too seriously. :facepalm:

Try to get Cello Palette and compare to any modern OP exotic design pre amps, and listen :)
You hear what you want. But you know nothing the actual benefits and drawbacks of opamps and semiconductors.

edit: typo
 
Last edited:
When I was looking for an extra couple of channels of amplification, I looked at various purveyors of Class D kits and finished products. I avoided one "manufacturer" in particular, purely based on their offering of different sonic characterisitics enabled by their discrete opamp juju.

If I want a particular sonic character, I'll hit up some DSP filtration. For components, just give me measurable transparency, thanks.

All this discrete audiophoolery stuff makes me chuckle. I wonder just how many great albums have been recorded/produced using equipment based around the humble NE5532 op amp?
 
When I was looking for an extra couple of channels of amplification, I looked at various purveyors of Class D kits and finished products. I avoided one "manufacturer" in particular, purely based on their offering of different sonic characterisitics enabled by their discrete opamp juju.

If I want a particular sonic character, I'll hit up some DSP filtration. For components, just give me measurable transparency, thanks.

All this discrete audiophoolery stuff makes me chuckle. I wonder just how many great albums have been recorded/produced using equipment based around the humble NE5532 op amp?
I (and many others) have said for decades that a lot of audiophiles would go berserk if they saw the number of devices (often cheap devices), cables, and what-not are in the recording/mixing/mastering signal path...
 
Back
Top Bottom