composite opamp benefit (audible)

[donovas]

hello, i am interest in composite opamp. especially the dual opamp, where each half used for loop feedback, much like figure 3 here

http://www.ti.com/lit/an/sboa002/sboa002.pdf

ok, technical writing all great (i dont understand a lot of it, haha) but how will it sound?!

will the opamp sound more dynamic, less distortion?

so many questions, no time to breadboard

thank you

 

[billshurv]

What performance composite are you after? That is the first thing you need to know before considering a composite. http://www.waltjung.org/PDFs/ADI_1992_Seminar_Audio.pdf has some of the classic audio applications in it.

Also note that it's VERY easy to get a composite wrong.

 

[twest820]

Yep. And actually another step back is needed as the question's specific to composite in a single dual. In practical audio situations there's no benefit to composite under this constraint as the control device has the same limitations as the output device. Benefit does exist in some cases but mostly they're pretty contrived.

More generally, in line level applications I can't see benefit to composite given proper device selection and circuit design. Modern audio op amps are good enough the limiting factor on performance is usually the size of the resistor budget. 20-40 years ago that wasn't so much the case, though most historical composite applications were non-audio as the NE5532 remains remarkably competitive.

For power amps it's different as the control device can be selected to be significantly more performant than the output device.

 

[DPH]

For power amps it's different as the control device can be selected to be significantly more performant than the output device.

It might have been you or JCX (given philosophies, I'm betting you, haha), that said it best: from a composite design mentality, looking at chipamps as integrated output stages with gain makes a lot of sense.

At audio line-levels, the value of a composite opamp is largely relegated to high-gain situations, which are honestly pretty rare (microphones and phono are the closest) in audio. There have been a number of discrete+opamp designs on diyAudio in the aforementioned space that are likely to hit design targets better (composites in their own right).

 

[sreten]

Hi,

There is hardly ever a significant advantage to compositing the same op-amp.
Perhaps some point with unity gain stable op-amps at high gain/bandwidth,
but you've really picked the wrong op-amp for high gain/bandwidth.

rgds, sreten.

 

[twest820]

(composites in their own right).

Yeah, if you look closely enough everything's composite on some level---internals of three stage op amps aren't all that different from power amps---and understanding of stage to stage stability transfers well to composite amps and vice versa.

For mics the performance limits are the mic's source impedance and the amount of resistor matching effort one's willing to pursue. THAT's mic preamp chipsets and TI's integrated mic preamps both get quite close to the source impedance limit with non-composite implementations. The miss of roughly 1 dB is mainly due to noise in the gain set resistors, which the additional loop gain of composite wouldn't help. Phonos aren't something I've looked into. But if pro audio was as silly as hi-fi I would know about someone making a mic pre operating at -40C to shave off a little bit of the noise.

Perhaps some point with unity gain stable op-amps at high gain/bandwidth, but you've really picked the wrong op-amp for high gain/bandwidth.

Indeed; in the infrequent cases it comes up decompensated op amps often are an easier way to pick up the loop gain. There's also a tendency for multiple op amps to be present in sequence, so often just budgeting gain appropriately from op amp to op amp will do fine.

 

[DPH]

Nice thing is that, nowadays, precision resistors are available (in solder-friendly 1206 0612 packages) even under a modest resistor budget. Golden era for those initiated to do it!

IC's, in terms of global costs are probably cheaper, though, especially in mass production? (Haven't looked)

Always going to be hard to out-precision fabs/processes focused on precision, though.

P.S. In an old lab, we ran our avalanche CCD around -40C via peltier and were able to, albeit noisy due to dark current, get down close to photon counting. We need that kind of performance because in audio because?

 

[twest820]

Varies with price and performance point and the circuit in question. The ICs are attractive for what one might call practical hi-fi. If one knows what one's doing very nice results can be obtained with discrete resistors. But you can only do so much to design around matching needs and, as precision requirements increase, the ICs become more cost effective.

Susumu has an all round excellent resistor portfolio, Linear makes extremely nice matched quads, and TE and Vishay have certain offerings useful in certain spots. Analog and THAT have a good range of ICs. You'll see a lot here on DIY Audio about Vishay z foil resistors due to their through hole availability but for what they cost it's not hard to buy a nice temperature controlled iron for surface mount work and come out ahead.