I am considering a high-end preamp/dac build and was thinking what would give me the best results when choosing between a high-end op-amp and discrete components like a JFET differential amplifier with a BJT push/pull output stage in class A mode.
I can see that a lot of hobby builders tend to use the discrete solution, but I can also see that some of the high-end Op-Amps are hitting distortion and noise figures that i would think is hard to beat like the OPA1612 with its 1.1nV/sqrt(Hz) and 0.000015% THD or the OPA1642 if JFET is the way to go.
In this case I would like to see some real arguments, so if we keep religion and feelings out of the discussion and keep to facts, what would be the better solution in a pre-amp/buffer stage, and would there be audible differences in performance?
I can see that a lot of hobby builders tend to use the discrete solution, but I can also see that some of the high-end Op-Amps are hitting distortion and noise figures that i would think is hard to beat like the OPA1612 with its 1.1nV/sqrt(Hz) and 0.000015% THD or the OPA1642 if JFET is the way to go.
In this case I would like to see some real arguments, so if we keep religion and feelings out of the discussion and keep to facts, what would be the better solution in a pre-amp/buffer stage, and would there be audible differences in performance?
Assuming the hypothetical discrete and op-amp circuits are well-designed, I don't think you stand any chance of hearing any difference at all, although Markw4 would probably disagree.
Yes, we assume that both amps has to deliver the same gain, with the same components in the feedback circuit, so it would only be the amplifier stage itself that are either in a premade op-amp, or constructed from discrete components.
And just to take it a step further we can assume that both are driving the same power amp, with normal gain, attached to normal mid budget speakers.
And just to take it a step further we can assume that both are driving the same power amp, with normal gain, attached to normal mid budget speakers.
This is a fairly simple question to answer, IMHO. I know of no objective reason to bother with a discrete op-amp over a modern monolithic IC op-amp. Assuming, of course, that the IC device meets the voltage and current swing requirements of your load. In either case, however, a competently engineered, feed-back based discrete op-amp or an monolithic op-amp both exceed the perception threshold of human hearing.
If you were to include a phono input, a loudness control or a balanced input, there would be a couple of interesting configurations you could make with discrete circuits and not with op-amps. Nonetheless, there would also be alternatives you could make with op-amps.
That depends primarily on your system and ears.
It is remarkable how good opamps can sound if used optimally.
Perhaps surprisingly i prefer using either passively filtered PS with chokes and LT4320 rectifiers or batteries. Regulators, even shunt types do not add anything musically worthwhile, on the contrary. The opamp itself is best biased in class A. Obviously a low offset fet input type is preferable in order to achieve dc coupling.
Can you do better using a discrete circuit? Sure, but it is far from trivial.
Discrete circuits sound distinctly different but often suffer from one sonic deficiency or another. For some reason subjective dynamics are often better using discrete and so are high frequency extension and speed. If these are really important to you discrete is the right way.
The other question when going discrete is whether to use nfb or not. This can also determine if regulation should be used.
Forums are valuable, so one can get a variety of opinions. Nothing can replace the experience of building a dozen or so circuits and see what works in your system. After all, an active line stage is not strictly required for the majority of systems. If it doesn't enhance your system musically then what is the point of having one?
Consider the noise, voltage and current and 1/f knee, the distortion, the load-driving ability, the amount of headroom and sensitivity to RFI and whether any of the parts are going to be available in 10 years time should repairs be needed.
Some opamps are internally a JFET differential amplifier and can be output-biased into class A output mode, so is there really such a distinction? Good JFET input opamps these days may the cheapest way to get decent JFET input stage anyway, and same goes for bipolar opamps - the best way to get a matched pair of bipolar transistors is as a precision opamp.
Discrete circuits are larger and this can mean more likely to pick up interference, another issue to consider, although more an issue with microphone / phono preamps than line-level.
Some opamps are internally a JFET differential amplifier and can be output-biased into class A output mode, so is there really such a distinction? Good JFET input opamps these days may the cheapest way to get decent JFET input stage anyway, and same goes for bipolar opamps - the best way to get a matched pair of bipolar transistors is as a precision opamp.
Discrete circuits are larger and this can mean more likely to pick up interference, another issue to consider, although more an issue with microphone / phono preamps than line-level.
I highly doubt that you can design/build a discrete circuit that will out-perform high-end opamps in bench tests. That being said, you may find a design that SOUNDS better to your ears. There's a reason why so many recording engineers covet the sound of old Neve mixing consoles. They're full of discrete circuits, transformers, tantalum capacitors and all sorts of poor-measuring things, but OOH---that SOUND!!
Lower voltage noise and higher power is still the domain of discrete. From 1nV/q rout to fractals of a Watt is does not make much sense any more. Massive parallel goes only so far.
An op amp circuit is always going to be smaller and tighter than a discrete one.
Loads of excellent op amps around these days.
In my day it was a terrible choice of a 741 !
Loads of excellent op amps around these days.
In my day it was a terrible choice of a 741 !
Last edited:
When it came out it was possibly the best. But thats about 6 decades ago! By modern standards the spec's are not impressive, but it set the stage for all that followed, and its USP was that it didn't self destruct if you shorted the output, unlike earlier attempts. That alone won it fame I think.
As far as I know, the chopper-stabilized Philbrick valve op-amps had much better DC performance.
Just to weigh in. I have feet in both camps. I have been running a tube preamp for years that I built, Aikido with excellent results. I have been using for the last 2 years the Self preamp that you will find in these pages. Now the Self preamp I consider SOTA. Almost no one on the market has put as much thought into each individual gain stage as far as noise, distortion and flexibility in an all opamp pre than Self. Just have to read the backing article to get all the facts. I built it with a very wary eye(ear). It is an excellent sounding pre. Also to wrap up I have been building about 5-6 different phono pre's lately and the one I settled on in the end is opamp based. For low noise and excellent sound. What I would concentrate on is the power supply for either. There are some excellent small board power supplies on this forum. A couple with active regs and CLC design that are my go to's for any pre.
Are you saying that the Aikido and the Self are both mediocre. I take it you have listened to both?
There are many very nice opamps and they sound quite nice. What mattered much more than the opamp itself for me was the psu. Opamps do like very stable voltages and as little mains noise as possible.