Op Amp chips in the signal path - evil or not evil ?

[thylantyr]

Sorry to bug everyone, I've gotten
bit by the speaker/amplifier virus.
Right now I'm working on my ambitious
speaker project but also am compiling
data on amplifiers.

Amplifiers confuse me because
there is two audio factions that seem
to have two opposite views.

The issue is.. Is simple circuit design
sonically better? Does the use
of op-amps (chips, ie, 5534, etc.) in
the signal path contaminate the
music ?

Exotic car audio amplifier manufacturers use op-amps (chips)
in the signal path and they use them for electronics crossovers and use
them for tone controls... and they
use capacitors in the signal path.
Nobody claims sonic inferiority
by doing this.

Exotic home audio amplfier
manufacturers use no op-amps,
no capacitors, nothing in the signal
path - hehe /joke-->If they could get away away with it, they would use a
piece of wire to amplify music if they
could get it to work.
This faction claims everything is
bad in the signal path.

How can both factions be correct ?

I can understand the need for simple
circuits, it's fun to find a simple solution to a complex problem.

Example, if you were asked to build
a vehicle for 2 person transportation,
the car audio faction would design
a Corvette, 4 wheels, seats 2, goes
fast, reasonble price. The home audio
faction would build a moped because
'simple is better", it's lightweight,
only has 2 wheels... but they would
charge 2x more than the Corvette
right ?

[jwb]

Build it, and find out

The car audio example is really bad, because car audio is targetted toward audible music in a very noisy environment with an unreliable electrical utility and a distracted audience. Totally different from dedicated listening in a quiet home with conditioned power and acoustic room treatment.

That aside, I think you should use opamps if they fit in with your philosophy of design. They are terribly convenient for electronic crossovers, equalization, current-to-voltage conversion, and voltage gain. Some of these circuits can be easily build with discreets instead of opamps. For example, you can build an I/V converter or a 3rd-order low-pass filter out of discreets with no problem.

I avoid opamps in most things. I once used one as the voltage gain stage in a headphone amplifier, and was happy with the result. But I don't use them because I don't need them. The audio band extends from DC to perhaps 96kHz. I do not need an opamp capable of swinging 135V/µs. I believe that if you do use an opamp capable of driving a 1GHz signal, you might accidentally wind up with a lot of 1GHz noise on your otherwise clean audio signals. And that would be Bad.

[thylantyr]

--The car audio example is really bad, --because car audio is targetted toward --audible music in a very noisy --environment with an unreliable --electrical utility and a distracted --audience. Totally different from --dedicated listening in a quiet home --with conditioned power and acoustic --room treatment.

I spent
10 years of doing car audio, mostly
for myself and helping others,
taking amplifiers apart to see
what makes them tick, but never
reaching the stage where I could
design from scratch, the only
thing that I was able to design
was text-box active filters using some
books as guidance and it worked
well.

I'm not one of those "bumping" dudes when it comes to car audio, I always built using the "tweeter down"
approach to seek quality sound
not just "boom" "boom"

After doing the car audio think, I could
never find any happiness when
visiting the local high end audio store
auditioning expensive speakers and amplifiers, my car audio experiences
provided more audible pleasure, perhaps because it was more challening
to work inside a car using multiple
drivers in a 4 way full active speaker systems. The quailty was sweet.

Even today, I can't seem to find
home audio happiness when it
comes to speakers, not so much
amplifiers, but the good amps I like
cost way tooo much.......

The audio adiction leads to the
desire to DIY speakers and
even amplifiers if a good design
is found - hehe


I'm breaking some speaker design
rules for my "1/2" line array project,
seeking more "horsepower" and
sacrificing some ride comfort - hehe

I'm wondering if the same could
be done for my amp project - hehe
Insert op-amp here, etc...

[woneill]

Hi Guys,

The purist reason why op-amps are bad is to do with negative feedback - which is also deemed bad when it operates from the output of a high-gain system back to its input.

Any negative feedback needs to be very carefully controlled with respect to phase, group delays, and (voltage/current) clipping during transients.

The higher the feedback, the more control needed. Ideally, any such feedback should be local to each stage of the amp, to minimise the above effects.

Op-amps operate with enormous gain, and enormous feedback. They offer very little ability to callibrate the feedback to the characteristics of the amp and the signal being used, beyond a trimmer cap.

That said, on simple signals, they do a mighty fine job. On transients, and very "dynamic" signals, they are regarded with suspicion...

All said, though, if they sound good, then use them.

Bill.

[Bill Fitzpatrick]

Just don't forget that with the great majority of recorded material the signal has already passed through many, many opamps. A couple more isn't going to hurt.

If anyone want's a signal that isn't opamp <i>contaminated</i> they are pretty much out of luck.

On the other hand, there are some recordings that are opamp free. What's a purist to do? Personally, I don't worry about it.

[Keithj]

Originally posted by jwb
Build it, and find out

That aside, I think you should use opamps if they fit in with your philosophy of design. They are terribly convenient for electronic crossovers, equalization, current-to-voltage conversion, and voltage gain. Some of these circuits can be easily build with discreets instead of opamps.

Yep, they're certainly good for all those uses.

I avoid opamps in most things. I once used one as the voltage gain stage in a headphone amplifier, and was happy with the result. But I don't use them because I don't need them. The audio band extends from DC to perhaps 96kHz.

Can you justify this comment (about 96kHz)? I seek only clarification

I do not need an opamp capable of swinging 135V/&micro;s. I believe that if you do use an opamp capable of driving a 1GHz signal, you might accidentally wind up with a lot of 1GHz noise on your otherwise clean audio signals. And that would be Bad.

I think Peter Baxandall did some experiments many years ago in which the maximum slew rate coming off a record (he deliberately chose records with lots of high frequencies such as cymbals, high hats etc) was about 2 volts per microsecond at the output of a 50 watt amplifier. I haven't seen such experiments done with CDs or other media but don't really expect anything different. Me? I still use 5534s. Wonderful little beasts. If I want low noise I use OP37s or LTA1036s.

As to high feedback, what's wrong with it? I've looked at some of these so called non feedback amplifiers and even built the odd wondrous one of them. All amplifiers have some form of feedback, even if it's local. Some of the so called magic stuff is marginally stable at best with ringing on certain frequencies (frequency dependent ringing... arrgh) and poor behaviour into real world loads. (I have to say that some equipment with overall feedback does the same). So, how does a competently designed and built amp sound different in its overall feedback and non overall feedback guises? Anecdote is no evidence BTW.

regards, Keith. (tilting at windmills again)

[Aud_Mot]

I think the answer depends on which specific opamp you are comparing to which specific discrete design.

Generally speaking, a properly designed discrete circuit specifically for an audio application will outperform a generic multi-application opamp. However, opamps are designed by some of the best engineers in the electronics world. They are wonderful devices.

The math for an opamp and some resistors is much more intuative and easier to apply than the math for a handful of transistors, diodes and resistors.

To get a start on discrete opamps:

http://www.passdiy.com/legacy.htm
Download D.I.Y. Op Amps

http://www.borbelyaudio.com/index15.htm
Download JFETS: New Frontiers Part I & II

To get a handle on opamps, look up the application notes for the opamp of your choice, or try some of the opamp cookbooks.

If you have access to a computer, there are several demo SPICE programs.

http://beigebag.com/
http://www.linear.com/software/

Car electronics are used in a unique enviroment and designs are very much driven by price points, both list and dealer cost.

Aud_Mot

[jwb]

Can you justify this comment (about 96kHz)? I seek only clarification

Sure. Biggest, baddest digital audio is published with 192kHz sample rate. The maximum frequency in that system is 96kHz. Most analog formats are way, way below that. Certainly vinyl and cassette tape.

For shorthand I usually figure the audio band is 0-100kHz.

[Lisandro_P]

Opamps are like everything else electronic; they're cheap, expensive, good, very good, plain ugly, etc. I tend to use discrete components on my designs because i like being able to tell EXACTLY what's going on there. But sometimes you just can't beat a good opamp, either for flexibility or specs. Ever checked the datasheets on the best Burr-Brown stuff? (drool)

[DrewP]

Some very well respected hifi companies regularly use opamps in their kit.

AFAIK, Musical Fidelity uses them in the phono stage of their newest A3 amp, the Michell phono stages use them too as do the Sonneteer Sedley DacT and Lehman phono stages. The MF, and Lehman are regarded as good stages while the Sonneteer and Michell are considered very good by most folks.

At the end of the day, implementation will have a lot to do with the resulting sound. How well are the power supplies sorted, what standard of passive components are used etc and so forth.

I mean, in theory most of these opamps are supposed to have huge levels of PSRR. If that's the case, then why does improvement of PSU (esp in the case of the michell and DacT) and going dual mono have so much impact on the sound?

I'm in the process of further developing an MC RIAA stage I built to incorporate higher grade opamps, caps and resistors but even with the bog standard cheap metalfilms, MKT polyesters and NE5534 (common audiophile response... Blleeeeccchhh!) I'm getting quite good results.

Sure, 2 or 3 valves would be more elegant but it'd weigh and cost lots more, require higher voltages and definitely couldn't be run on 9v cells or 12 volt cells.

I would say, if properly utilised not evil.

Drew