Real Man Don't Use Opamps
Op-amp has never been considered as an option for building true reference quality amplifiers. Discrete parts seems to be the way to go. You only use opamp for "cheap" amplifiers destined for the masses.
Am I mistaken or is there some truth in this?
Now we have a descrete opamp from Burson Audio of Australia and is sold by Parts Connexion.
http://www.partsconnexion.com/Index/burson.php
http://www.bursonaudio.com/
Their reason for designing a discrete opamp is to address the problems we have been trying to avoid.
Op-amp has never been considered as an option for building true reference quality amplifiers. Discrete parts seems to be the way to go. You only use opamp for "cheap" amplifiers destined for the masses.
Am I mistaken or is there some truth in this?
Now we have a descrete opamp from Burson Audio of Australia and is sold by Parts Connexion.
http://www.partsconnexion.com/Index/burson.php
http://www.bursonaudio.com/
Their reason for designing a discrete opamp is to address the problems we have been trying to avoid.
Re: Real Man Don't Use Opamps
This reminds me of a book I bought a long time ago. It's title was "Real Men Use DOS".
Care must always be taken with the inevitable advance of technology when making generalizations. While it might be true that the very very best preamps are made with no op amps in the signal path, it is probably just as true that many very good preamps made with very good op amps are superior to many preamps made discretely. The only safe generalization is that generalizations are dangerous.
Cheers,
Bob
caesar148 said:
This reminds me of a book I bought a long time ago. It's title was "Real Men Use DOS".
Care must always be taken with the inevitable advance of technology when making generalizations. While it might be true that the very very best preamps are made with no op amps in the signal path, it is probably just as true that many very good preamps made with very good op amps are superior to many preamps made discretely. The only safe generalization is that generalizations are dangerous.
Cheers,
Bob
Thanks PD. Have read this before but completely forgotten. Nice review. Pretty much what i would have expected from a discrete opamp comparison especially taking into account that none of the tested opamps sound even remotely good IMO. The buffer review is more mystifying: no down side at all?
I think that a system should be judged by how it performs. Simply having opamps in a circuit certainly should not mean that it is bad or will sound bad. Real men don't judge a circuit by its components; they judge it by its sound, using measurements as a tool to help quantify how good it sounds and as an indispensable aid during design and construction. This is of course just my opinion, so you don't have to agree.
Now, as for the Burson opamps... without performance measurements, you cannot truly judge how they perform (or even what they do!).
Now, as for the Burson opamps... without performance measurements, you cannot truly judge how they perform (or even what they do!).
leadbelly said:
Everything - at the very least (in no particular order):
- input noise vs. frequency
- psrr vs. frequency
- cmrr vs. frequency
- closed-loop distortion with varying gain (inverting and non-inverting), loading, output voltage, and frequency (as spectral output, not single THD value) <--- probably the most important
- slew rate (just out of curiosity)
- unity-gain bandwidth
- phase margin / amount of ringing with varying capacitive load
Also would be nice:
- (very difficult/impossible on a real circuit) open-loop gain vs. frequency
- effect of source impedance on closed-loop distortion
And finally, complete telltale (although Burson wouldn't be happy about its publication):
- reverse engineered schematic
IC technology and topology are well developed now and serve each other well. For example, opamps use such tricks that can't be achieved with discrete elements, so when opamp is needed ICs are superior. But when audio amp is needed opamp topology is suboptimal, so I see no reason to re-implement a discrete opamp for audio.
Wavebourn said:
Interesting. I am completely on the other side of the fence. Some things that discretes have trouble acheiving that ICs are better at are:
- good component matching
- wide bandwidth
Other than those two... I think discrete opamps can have an edge, especially because passive components of arbitrary value and good linearity are much easier to come by. The DC specs are almost certainly going to suffer, due to comparatively poor component matching, but that can be dealt with.
I use opamp topologies built out of discretes because:
- the design exercise itself is gratifying and interesting
- GNFB on top of a nice, linear circuit (open-loop) == even better circuit closed-loop (your view may vary considerably here)
i worked for Apt back in the late 70's and 80's. the Apt Holman preamp was almost all op amps. the exception being the phono preamp, which was a discrete op amp. if you look up the specs for it you will see that, especially for it's time, it was hard to beat. all of the op amps were TL072's, except for the headphone amp, which was an RC4558. it outperformed preamps costing 2 or 3 times as much. even today it's difficult to get the owner of a Holman preamp to sell (maybe if you pry it from his cold, dead fingers?)
don't make generalizations about "3 legs good-8 legs bad", 'cause it ain't always so. one of the big problems with discrete designs is their PSRR and thermal drift. you can't match differential components any better than having them on the same piece of silicon.
discrete designs, on the other hand give you a lot more control over things like slew rate, open loop gain, input devices (the Holman phono stage used a diff amp that had a jfet for the noninverting input and a bipolar as the inverting input, the jfet was a very low noise device)...
don't make generalizations about "3 legs good-8 legs bad", 'cause it ain't always so. one of the big problems with discrete designs is their PSRR and thermal drift. you can't match differential components any better than having them on the same piece of silicon.
discrete designs, on the other hand give you a lot more control over things like slew rate, open loop gain, input devices (the Holman phono stage used a diff amp that had a jfet for the noninverting input and a bipolar as the inverting input, the jfet was a very low noise device)...
Re: Real Man Don't Use Opamps
Consider what the basic differences are.
Op-amps are physically much smaller.
All components are made from layers of silicon and metal.
All components are thermally linked.
transistors are better matched - they all come from the same batch
You can only access a few circuit nodes
Advantages:
- convenience; "fast food" amplifier
- small size reduces inductance, enabling good bandwidth
- consistency from one part to another, transistors matched
Disadvantages:
- resistors are non-linear
- capacitors are very small (big caps consume too much die area)
- terrible isolation between stages, both thermal and electro-magnetic
- inductors are impractical
- PSRR is poor and it is not possible to isolate supplies to different stages
- not often an ability to change internal biases
- no ability to optimise components. Reliant on the internal design choices made by the mfr, which are almost never correctly optimised for audio.
But apart from that, op-amps are great.
caesar148 said:
Consider what the basic differences are.
Op-amps are physically much smaller.
All components are made from layers of silicon and metal.
All components are thermally linked.
transistors are better matched - they all come from the same batch
You can only access a few circuit nodes
Advantages:
- convenience; "fast food" amplifier
- small size reduces inductance, enabling good bandwidth
- consistency from one part to another, transistors matched
Disadvantages:
- resistors are non-linear
- capacitors are very small (big caps consume too much die area)
- terrible isolation between stages, both thermal and electro-magnetic
- inductors are impractical
- PSRR is poor and it is not possible to isolate supplies to different stages
- not often an ability to change internal biases
- no ability to optimise components. Reliant on the internal design choices made by the mfr, which are almost never correctly optimised for audio.
But apart from that, op-amps are great.
Op-amps are designed by some of the world's best engineers. When they need a particular blend of transistor qualities, they can ask for exactly that, and they can position everything really close so that stray capacitances aren't a problem.
Conversely, op-amps (by definition) are designed to do everything quite well, whereas a discrete design only has to do one thing very well. Also, op-amps are limited by the IC processes. If an optimum design requires entirely different processes a discrete design can do it but an op-amp might not.
My opinion is that you have to work very hard to outdo the best op-amps and it's only on the low-noise application-specific front ends that you're likely to see a real advantage by going discrete. Line level stuff is usually adequately handled by op-amps unless it also carries digital crud (I/V convertors etc).
Conversely, op-amps (by definition) are designed to do everything quite well, whereas a discrete design only has to do one thing very well. Also, op-amps are limited by the IC processes. If an optimum design requires entirely different processes a discrete design can do it but an op-amp might not.
My opinion is that you have to work very hard to outdo the best op-amps and it's only on the low-noise application-specific front ends that you're likely to see a real advantage by going discrete. Line level stuff is usually adequately handled by op-amps unless it also carries digital crud (I/V convertors etc).
That's like saying Bear Stearns is managed by some of the world's best financial experts.
Clever and experienced engineers are only as good, in this context, as how well their design goals match audio requirements. This is why so many op-amps are great for many applications but are lousy for audio. These guys don't learn audio in school.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.