Forced Class A opamp

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Now that you mention it Ted, that is indeed a good theory! However, a lot of discrete opamps do follow similar designs as IC opamps except for a few certain tidbits that are different between the parts. Discrete opamps, as we usually build them, include areas of inductance and capacitance that ICs do not have due to trace/lead lengths and proximity to adjacent parts, and also the quality of the parts used. Heat and proximity to the other parts near also affect overall quality where an IC will heat most parts within the package fairly equally, the discrete will likely heat only a few parts unless of course it's potted with a thermal compound.

so this actually leads me to wonder if the difference that we actually hear between a discrete and IC part of similar topology is not so representative of the quality of the parts but the addition of qualities of the devices that we usually don't look at? Maybe an IC is actually TOO sensitive to changing conditions and the discrete is not as sensitive do to much longer/thicker, more inductive and cacacitive traces and leads than the IC with it's microscopic traces/leads?

Just a thought that has actually been bouncing around for some time..
 
Very interesting indeed. I definitely agree that transient behavior of circuits is something that is not usually well understood. Music is full of transient and non-regular signals, but we spec nost things with fixed freq sines and squares. Big gap.

You guys should pick up copies of Jim Williams' two books for some really interesting reading on analog design:

Analog Circuit Design : Art, Science and Personalities
The Art and Science of Analog Circuit Design

Each chapter of these books is written by a different designer (with some authors having more than one chapter). There are chapters on specific circuits, thermal issues, design philosophy, mentoring, etc. Bob Pease has a chapter in one of the books where he looks at old Philbrick discrete op amps. Fascinating stuff. Of course his own book, Troubleshooting Analog Circuits, is also a gem.

Analog Packrat
 
Hmmm... I never saw the point of making descrete op-amps.... they are designed with extreme feedback to make up for the poor hfe of the devices. Simple two transistor amps with the addition of active loads, or at the most, ring-of-three amps are much more stable.

BTW... it's not really a 'theory', it's clearly demonstrable!
Years ago I designed a really simple little mixer (the Alice 828). It sounded fabulous and became a 'standard' in the film business, (and was used on some of the early Bond films!). Later, I 'improved' it with numerous modifications that reduced the distortion dramatically... it never recovered.
Similarly, The first Trident 'A-range' mixer was designed by my old friend Barry Porter (sadly no longer with us). It was made up of simple descrete amplifiers and that became a standard for rock-and roll recording in the UK. Subsequent Trident mixers never sounded so good.
I could go on... but I think it's QED.

What I really mean is, as designers we must remain aware of where problems lie.... Op-amps have their place as long as we are really sure that we don't over-drive them!!
 
Hi Ted. I'd love to hear more detail on the particulars of the Alice 828. Was it transformer-coupled? What was the topology of the amplifiers? Did it use the same amplifier type throughout? Was there EQ? Etc...
 
true, very true. I only mean that we need to figure out exactly what causes the differences between the two ways of doing the same thing in order to truly know the designs at hand. This in turn would help us design future devices with the qualities we are looking for and less of the attributes we dislike.

We sketch great designs on napkins and then build great devices only to learn that it's inferior to another design because of some subtle difference that we might have never seen on paper. At least this is my observation on many occasions.

I guess I am trying to say that I would like to investigate if the great discrete opamps are truly better by design or if they are better due to flukes in that design. I guess I'm still skeptical why we always turn to the discretes for "quality" of audio when there are IC opamps that should be far superior but always see to fall a little short.
 
they are designed with extreme feedback to make up for the poor hfe of the devices.
Could you elaborate? Most IC processes have much lower hfe than what we get with discrete transistors, and most discrete opamps I have seen have lower GBW (which in turn means less feedback for same closed loop gain) than monolithic ones...

Samuel
 
Hi Dave, the 828 was really simple.... transformer mic input, simple descrete mic amp, a Baxendall EQ with an added coil and mid section, and a really basic transistor limiter (similar to the AGC on a Philips cassette machine). The EQ was a single transistor(!!) The mixing was resistive; which suited the limiter, which needed a very low level.

Bear in mind, this was 1972 so things were a little primitive.... but it did sound good!

Ted Fletcher
www.tfpro.com
 
[quote author="TedF"]Bear in mind, this was 1972 so things were a little primitive.... but it did sound good![/quote]

You've probably noticed that most people here tend to prefer such "primitive" technology :wink:

I'd love to see schematics, but I know that's asking a lot, so I won't press...

The "Philips"-type AGC works by controlling the base bias on a BJT, doesn't it? I know I've seen that circuit, but can't remember where.
 
Hi Samuel, I apologise for sloppy terminology... I meant that IC op-amps are designed with lots of compensation to allow for the poor quality of the devices. This translates to lots of feedback so that when an anomaly occurs, the disruption is very much greater than it would be in a descrete amplifier, whether it is an op-amp, or a simple 2-transistor amplifier.
 
I just re-read some pages from D. Self and found this little circuit:

hybcct1.gif

It looks to me as if the bias arrangement for the two diodes converts the opamp output to single ended class A as discussed in this thread; this would be a very nice and simple to use variation for heavier loads.

The 4.7k could be replaced with a CCS or get some bootstrapping, but according to Self this does not change linearity with a TL072.

Samuel
 
True, the configuration does hang a bit of an assymetric load on the IC, but I doubt if you would hear the advantage.... the addition of the output stage diode drops adds even more non linearity within the feedback loop making the circuit prone to low level short term 3rd harmonic distortion.

It works fine and is a good way to get some extra current drive, and isolate the IC from nasty impedances, say driving a transformer.... I used it as such for many years.
 
The signal diodes don't match the base/emitter diodes of the output pair... so there is a significant step during the transition from the op-amp providing the drive current (at very low levels, to the class B condition when all the current is coming from the transistors (and the output of the op-amp itself goes very non-linear).
(No, I didn't explain that too well, but I'm sure you can see what I mean!)
 
TedF: just looked at your website.. wow, that's good stuff! can you give a bit of insite on your current based circuitry? I understand if you can't talk about it but it I'd like to learn a bit about it if possible. Probably a topic for it's own thread.

:thumb:
 
Hi Samuel, Yes, remaining within class A everything's fine but I suspect that the old 'descrete output' circuit will go int class B at only a few millivolts of signal... there's actually a much better circuit arrangement with the transistors in the power supply rails; they turn on by virtue of the current drawn from the rails. This arrangement can be set up so that the system holds in class A for longer.... but all this is really academic, I prefer to use one of the well known op-amp based balanced outputs, or when I want to drive more current, good op-amps in parallel.

Re 'current mode' circuits.... there's nothing particularly innovative about this, it's more a different way of looking at circuitry; I was involved with international comms systems for a while, and started looking at electronics as current loops rather than potential differences. Suddenly I could see that if you look at a microphone circuit as a current loop, and amplify the current rather than the voltage, the 'quality' of the signal at the mic head is reflected exactly at the preamp input (Kirchhoff's Law).... this is not so if it's a conventional voltage amplifier. As a bonus, the length of the cable becomes immaterial (within reason!)
A logical extension to the thinking is to isolate circuits with transformers; when transformers operate as current converters, because there is virtually no voltage generated across the windings, then the core cannot saturate... so the frequency response is extended almost down to DC!

Of course it's not all as easy as that; for a start, microphones are designed to operate into finite impedances, not into the 'short circuit' of a current input, and commercial transformers are designed as voltage converters, and it's difficult to persuade transformer manufacturers to think in terms of micro-currents (they look at me strangely!)
There are compromises, but most of the advantages can be retained.
 
Wow, what fantastic directions this thread has gone! It's become such an interesting read--maybe someone should add it to the meta, along with the other class A opamp threads.

I do feel compelled to ask a question of you Ted, and to make a comment about something you stated in an earlier post:

after all the only thing the resistor does is to alter the conditions of the output devices, it does nothing for the internal architecture.

But, as I understand it, this is exactly the goal here--to get rid of high-order distortions caused by the output stage crossing over. The typical distortions in the input and voltage amplifier stages are generally low-order (second and third harmonic) when they occur, with the exception of load-based distortions from the voltage amplifier stage. This would most likely make them less objectionable to the ear.

By altering the DC balance position of the amplifier it's entirely possible to eliminate the 3rd order products caused by instability at crossover

I was under the impression that 3rd order distortion in an output pair is a large-signal distortion, and occurs when driving a 4 ohm load. Pretty much everything else is a complex, high-order type. Am I mistaken?
 
Back
Top