advertisement


Do you need to be discrete?

stevec67 said:
Maybe not from you but I bet the marketing dept would. As the question implies, "discrete" is being used in the marketing to suggest it's better.
Click to expand...
I took the OP to be asking whether there is any truth to such marketing. As a general rule, I believe you agree, there is not.

stevec67 said:
Also, as I said earlier, opamps may be available but if you want something that isn't available in an opamp then you have to build it.
Click to expand...
The OP specifically mentioned opamps.

If a particular sub-circuit is used frequently enough, someone is likely to make an IC of it. High-gain differential amplifiers happen to be very useful building blocks, and that is why opamp ICs exist in numerous variants. The same can be said about, for example, voltage regulators.

Audio devices are pretty common, and they all do the same basic job. If your design is so unique that it can't incorporate ICs (which generally simplifies the design and reduces cost), I'd like to see some good evidence that it performs better than a traditional approach in some important way. Since audio electronics can be made with no audible distortions whatsoever, that's a pretty tall order.
 
Accuphase make the best sounding DACs and CD players I have heard .They use op amp based output stages.
 
In the early audiophile years, serious equipment avoided opamps - criticism was a cold, dimensionless, hard sound.

Now, I find that my Emotiva balanced amplification chain, which is opamp based just really doesn't have those faults.
 
John Phillips said:
I don't have a complete answer since the crossover and power amplifiers are on the same board, but for the SCM100A compare.
  • Standard amplifier pack photo here.
  • Discrete amplifier pack photo here.
(I omit the Anniversary edition amp pack).
Click to expand...

They describe them as 200W/100W/50W amp packs, but they only have a single pair of TO-220 output devices per channel?

Anyway the "standard amplifier pack" appears to have an opamp front end (for the crossover?) and discrete for the back end.

One difference between discrete and opamps is that discrete will need to be designed by an engineer who can actually do circuit design; functioning opamp circuits can be thrown together almost Lego-style by any 3rd year EE student. Although there still is an art to doing opamps well.
 
Yank said:
They describe them as 200W/100W/50W amp packs, but they only have a single pair of TO-220 output devices per channel?
Click to expand...
It's difficult to see but there seem to be the more usual four output devices per channel. Clearer views are here (standard - topside outlines) and here (discrete - underside devices).

As @Arkless Electronics says it seems probable that the ATC's amplifiers, described as "grounded source", are architecturally like the James Strickland distortion cancelling patent US4467288A as used by David Hafler in at least some of his amplifiers.
Yank said:
Anyway the "standard amplifier pack" appears to have an opamp front end (for the crossover?) and discrete for the back end.
Click to expand...
Certainly discrete where the op-amps were for the crossover filters and phase correction, handling line-level signals. But it's not clear if any op-amps are used in the amplifiers in the standard amp pack. Possibly not, as you write.
Yank said:
One difference between discrete and opamps is that discrete will need to be designed by an engineer who can actually do circuit design; functioning opamp circuits can be thrown together almost Lego-style by any 3rd year EE student. Although there still is an art to doing opamps well.
Click to expand...
Yes. With op-amps you certainly need to get beyond the tempting but naïve belief that op-amps have infinite gain, infinite bandwidth and are perfect in all other ways. Much like good discrete design, after all.
 
stevec67 said:
Also, as I said earlier, opamps may be available but if you want something that isn't available in an opamp then you have to build it.
Click to expand...
And as an aside, someone who reviewed a Vincent AV amp some years ago contacted Vincent about one of the special operations that the amp performed, as it wasn't covered in the manual. The amplifier's designers had no idea the opamp had that facility! It came with the opamp, but they never knew. :rolleyes::D
 
Last edited:
Yank said:
They describe them as 200W/100W/50W amp packs, but they only have a single pair of TO-220 output devices per channel?
Click to expand...
I thought that, I was looking for the heat sinking. Are they using Class D or some other cool running topology?

One difference between discrete and opamps is that discrete will need to be designed by an engineer who can actually do circuit design; functioning opamp circuits can be thrown together almost Lego-style by any 3rd year EE student.
Click to expand...
3rd year EE? I should hope so. I'm a food technologist with an O-level and *I* can bolt together a working amp on stripboard following a read of the guidance notes. No 741 is safe. It's not as if a 7297 or 3116 is an awful lot more difficult. It's only one step up from making a light flash with a 555 timer.
 
mansr said:
I took the OP to be asking whether there is any truth to such marketing. As a general rule, I believe you agree, there is not.
Click to expand...
Absolutely. As I think I said on page 1, a good IC based circuit is better than a poor discrete circuit and vice versa.

If a particular sub-circuit is used frequently enough, someone is likely to make an IC of it. High-gain differential amplifiers happen to be very useful building blocks, and that is why opamp ICs exist in numerous variants. The same can be said about, for example, voltage regulators.
Click to expand...
Yes indeed. Is it just me or are more and more high quality amplifier ICs making their way into the world? I know they have always been in consumer electronics, after all nobody is going to recreate a Krell KSA 100 to run a TV or a kitchen radio, but they are popping up all over now. TPA 3116, TDA 7297 (?) the original T-amp from 15 years ago and so on.

Audio devices are pretty common, and they all do the same basic job. If your design is so unique that it can't incorporate ICs (which generally simplifies the design and reduces cost), I'd like to see some good evidence that it performs better than a traditional approach in some important way. Since audio electronics can be made with no audible distortions whatsoever, that's a pretty tall order.
Click to expand...
So says the engineer, but the Marketing department will have you killed if you so much as breathe that to the consumer. Audio is special, that's why a £1000 amp is worth having and nobody wants to talk about the £2 chips that only need a power supply and an old beans can as a heat sink. That, and the fact that heat sinking has got to be a problem in high power IC applications, drives the "discrete is better" marketing myth. Quite apart from anything else, how would you feel as a consumer if you paid £1000+ for an audio amplifier and then read a review that showed it uses the same chip in the same circuit as a £20 item from Fleabay? I know I would be bloody pig sick. 47 Lab, I'm looking at you.
 
mansr said:
The question doesn't make sense for functions where there is no choice.


No, but opamps are. If, say, an OPA1612 fits in terms of signal level, bandwidth, and drive strength, you'll have a very hard time getting better performance using discrete components. You don't get points for doing things the hard way just because, at least not from me.
Click to expand...

Op amps absolutely can easily be beaten on sound quality. They are a weak link in the chain for true high end equipment. Sure they can give amazing measured performance but they just don't sound as good.
 
mansr said:
I assumed we were talking about circuits that can reasonably be implemented either way. There are also things that no sane person would attempt to implement using discrete components.

My opinion is that if an IC exists with specs suitable for what you want to do, there is no (technical) reason not to use it. ICs do not possess some mysterious badness that only manifests when used to play music.
Click to expand...

You should see an "ultimate" MC phono stage design of mine! Single stage... zero NFB... more matched arrays of discrete transistors than you can shake a stick at (thanks for the samples THATS!). No one has ever attempted such a thing before to the best of my knowledge.... You'd have to be mad to try it! It's erm... a WIP... got it nearly there but put it to one side for now.
 
Arkless Electronics said:
Late 90's was roughly when they became available as normal off the shelf parts.
Click to expand...
They were off the shelf, just that nobody in the HiFi industry had heard of them. Walt Jung noted the HA2500 series in his 2005 Op Amp Applications book
Quoting:
In the early seventies, just about the only truly fast IC process was owned by Harris Semiconductor. This
dielectrically isolated process produced equal speed NPN and PNPs. and the Harris HA250O series became
popular for fast settling characteristics. In 1973 ADI released the fast AD509 op amp. a screened Harris part
 
You must log in or register to reply here.


advertisement


Back
Top