Right, but there's always the degree of compromising in any product. AFAIK, nobody has made a compromiseless circuit so far.
Accepting that as my premise, I'd say some op amps are very, very good. And I'd add that a discrete circuit which would better them would be far more expensive and bulky.
That too is a compromise.
Optimization itself means choice of better compromises to satisfy requirements. When I design audio amp I do not need for example CMRR and similar parameters vital for analog computers, so I can use couple of vacuum pentodes and couple of MOSFETs and get results you can never obtain with opamps. If you want to speak of prices and ICs, I am hundred percent sure that an IC that contains 5-6 transistors inside can be comparable with an IC with hundred of them if to manufacture it in certain quantity. And it can be no less universal for audio than opamp that was well advertised in textbooks for engineers and in magazine articles, thanks to requirement for analog computers that does not exist anymore.
This is not leading anywhere, as too many ifs are at play.
It's a simple fact that audio is no longer in the vogue it was say 30 years ago. Therefore, most manufacturers of op amps see no particular reason to develop extremely high quality audio orientated op amps. Since the arrival of digital audio and video, even less so.
I agree that they could, if they wanted to, develop outstanding strictly audio circuits inside that small plastic, ceramic or metal package, but it's a simple fact that its price would be a problem to most manufacturers, even when buying in bulk.
ALL of us here, I think, need to understand we are are the diehards, the old guys, still salivating over a great amp or whatever. It's a fact of life that modenr kids care incomparably more about the next generation of i-Pods, cell phones and tablets than about audio. In other words, audio as we know it is rapidly dying. It will never disappear completely, of course, and who knows, it may once again become fashonable in the "in" crowd once every Joe and his dog have the latest i-Pod type device, but even so, it will more of a swan song than anything.
I know I have lived since 1966 with audio playing a very important role in my life, and I will not let that change. After 46 years with it making my life better, why the hell should I?
But op amp manufacturers have a completely different logic. To them, it's a business first and foremost, and if they don't see a profit down the road, they won't do it.
Let me ask you this - how many NEW tubes (meaning new designs, not new versions of the same old same old) have appeared over the last 20 years? THAT'S the point. Measure it however you like.
Look at how many truly NEW transistors have been introduced in what we would think of as interesting for us - now compare that with the number of very good transistors you used to have that are no longer with us, and what do you get?
Just when you think - aha, here's a new BC type transistor, you are disappointed, because it turns out that it is in fact VERY sightly reworked old type, with a new name and almost insignificatly improved parameter or two. And if they won't introduce REALLY NEW transistor types, the chances of really new high audio quality op amps appearing are way smaller. Not because of technology, but because of the money involved.
Am I right, or not?
I think you have slightly overstated it, Dvv. It is in our interest to REMIND IC designers that their product is not 'perfect', and really not even good enough for the serious listener, in many cases. IF we do not remind them, they will be satisfied with what they make, and ultimately, we will all lose. It is a bit like 'fast food'. It is OK, at times, but it is not 'haute cuisine' and a visit to a quality restaurant can 'awaken' our taste buds, just like listening to a really high quality sound system can remind us of what we are missing with our portable hi fi's.
the demands of precision op amps for instrumentation used to be largely limited to DC, to a few kHz
today, last decade or so, medical ultrasound, xDSL have driven op amp design to new levels in processes and topology
raw speed increases alone make high loop gain feedback work much better at audio frequencies
SiGe devices give a slightly better i_noise in bjt
Analog Devices linearized diff pair gives a huge reduction in high frequency distortion in feedback op amps into the MHz - that is loop feedback working into the MHz
National made some effort with the LME series – claimed to have setup subjective audio testing room in addition to getting the AP # down into the noise floor
TI/BB keeps releasing new “Sound Plus” branded op amps
Several manufacturers have released parts in the past decade aimed at the OPA627, a still prime performance device, if pricey
and of course there is my personal hobby horse - composite op amp circuits using great input op amps, buffering, adding loop gain themal/load isolation with CFA ouput op amps - smt cuts the PCB real estate required
I really don't understand complaining about single op amp performance limitations when 2 modern $2 op amps in a multiloop removes virtually all active device performance limitations at audio (other than input noise)
today, last decade or so, medical ultrasound, xDSL have driven op amp design to new levels in processes and topology
raw speed increases alone make high loop gain feedback work much better at audio frequencies
SiGe devices give a slightly better i_noise in bjt
Analog Devices linearized diff pair gives a huge reduction in high frequency distortion in feedback op amps into the MHz - that is loop feedback working into the MHz
National made some effort with the LME series – claimed to have setup subjective audio testing room in addition to getting the AP # down into the noise floor
TI/BB keeps releasing new “Sound Plus” branded op amps
Several manufacturers have released parts in the past decade aimed at the OPA627, a still prime performance device, if pricey
and of course there is my personal hobby horse - composite op amp circuits using great input op amps, buffering, adding loop gain themal/load isolation with CFA ouput op amps - smt cuts the PCB real estate required
I really don't understand complaining about single op amp performance limitations when 2 modern $2 op amps in a multiloop removes virtually all active device performance limitations at audio (other than input noise)
Last edited:
All my designs are new.
Well, some of them are more than 20 years' old (because I am almost 3 times older than this number), but some are very new. And I don't use opamps that cost 1 dollar instead of tubes that cost 1-5 dollar each, because for $3,000 product difference in few dollars (and even tens of dollars) does not matter at all.
No.
Relatively recently Zetex introduced transistors that would cause me drooling 20 years ago. And SemiSouth recently introduced devices that could not be imagined then. If to think about design of something new you will find new components, technologies available. But if to think about reproduction of old good designs however components disappear, even when they would be great for new designs.
The life always brings new problems and new resources to solve them. It is dynamic. The same as music itself. That's why we designers are still needed: to match new solutions to new problems. To create new fashions that create new problems that require new resources.
All my designs are new.
Well, some of them are more than 20 years' old (because I am almost 3 times older than this number), but some are very new. And I don't use opamps that cost 1 dollar instead of tubes that cost 1-5 dollar each, because for $3,000 product difference in few dollars (and even tens of dollars) does not matter at all.
What about the added circuitry ? Extra PSU for tube device , are op-amps used Because of economics or performance ...?
No. They are used mostly because of application examples in datasheets, books for colleges, magazines, etc...
All application examples and schoolbook schematics are free, and using them companies don't expect to be sued by somebody who dared to patent similar solution. Engineers do not need to be creative: just come to work, take schoolbook examples, draw schematic using software, push a button and generate BOM. Then go home, take a lunch, and watch your favorite football game...
Fair enough.
I agree everybody, including op amp manufacturers, needs a reality check on occasion.
While a firm believer in discrete circuitry, I still say the op amp manufacturers have come a long way. Not all of them, not with every product, but there are some REALLY good sounding op amps out there.
And there are quite good sounding op amps which have been around for ages. OP37, LF356, OP275, and so forth. Not for the High End, but can do just fine in the Mid Fi segment.
Not to mention the "service" op amps, meaning those virtually ideal for a specific function - a good example would be LF411, a Gopdsend for DC servo, given its drift of less than 1 mV PER ANNUM! Try that with discrete and you'll end up with a prohibitively expensive, large circuit board.
All my designs are new.
Well, some of them are more than 20 years' old (because I am almost 3 times older than this number), but some are very new. And I don't use opamps that cost 1 dollar instead of tubes that cost 1-5 dollar each, because for $3,000 product difference in few dollars (and even tens of dollars) does not matter at all.
A slight misunderstaning.
Let me rephrase - how many completely new tubes did you have over the last say 20 years?
A completely new, from the ground up version of say ECC83? Which outperforms most others and perhaps adds a few virtues of its own?
I'd like to see a say BC1000, which would be good for say Bvceo of 100V, 100 mA continuous and say 500 mA peak, 1W, TO-92 package, Nf <1 dB, Ft >500 Mhz, Gain 100...450, etc.
It's quite possible new production lines have been set up, or old ones revived, but they make only what has been around for decades.
I was NOT questioning your own design work, putting in so many tubes in such and such arrangement to do whatever. Even if I wanted to, I couldn't, because I am not qualified to judge anybody's tube work, except in terms of listening and speaking my mind as to how it sound to me.
I'm old enough (full 59 today) and experienced enough not to presume to judge anybody's work on basis of paper schematics. I've been surprised too many times by real life experiences, in both ways.
... and because they are extremely convenient building blocks, and because they are cheaper than equivalent discrete ciruitry often enough, and because they shrink your board no end, and because they are easily available.
I would love the people here to design a discreet op amp from easy to get parts . Lets say 10 transistors . 20 MHz GBP . Class A output option . +/- 25 V supply .
I am so incredibly lazy that by 2020 I might have an idea worth trying . I suspect others would just love this and do it in 6 weeks . It could be surface mount so as to be not too large . That would also allow special transistors . I would say not the That's range as they are a tad expensive for a mk 1 version . It could be produced and profits go to feed this forum . For me I would call it HCL 1 .
It could have JFET or bipolar depending on use perhaps ( HCL 1 or HCL 2 ) ?
The letters are the daddy of op amps Dr Lin . HCL is an acid so perhaps not ? Mr Szikai is worth a mention .
I am so incredibly lazy that by 2020 I might have an idea worth trying . I suspect others would just love this and do it in 6 weeks . It could be surface mount so as to be not too large . That would also allow special transistors . I would say not the That's range as they are a tad expensive for a mk 1 version . It could be produced and profits go to feed this forum . For me I would call it HCL 1 .
It could have JFET or bipolar depending on use perhaps ( HCL 1 or HCL 2 ) ?
The letters are the daddy of op amps Dr Lin . HCL is an acid so perhaps not ? Mr Szikai is worth a mention .
I think wires got criss-crossed here.
Methinks Nigel was not referring to TOPOLOGIES as such so much as the circuits made in discrete technology, not necessarily mimicking internal op amp topologies.
Meaning make a discrete op amp on a board, with pins to solder it in and hold it, but use any topology you see fit to use.
I seem to remember an US made preamp, around mid-70ies, which did just that - it had a motherboard and six (if memory serves) 100% discrete "op amps" soldered to it using pins. JC might remember more clearly, I just have a picture in my head from the Audio magazine test.
Actually, there is logic in this. If you are making just say two types, the phono RIAA eq/amp being the second, on a series production basis this is not a bad idea at all. You can vary the quality of parts as well as their number (to an extent), and thus provide for levels of quality.
Of course, the drawback is that ultimtely you will come to a point where limitations in space will start to get in the way.
- Status
- Not open for further replies.