.
and that is exactly what I said in my first post... and called John on yesterday, opamps are not magic bullets.
I love how the situation is always painted with cheap opamps and bad layout as the competition... who except you and perhaps others trying to make silly comparisons, has put forward the idea that the cheapest opamp in a bad layout will beat a well designed discrete circuit? I am simply saying that all things being equal, in the vast majority of cases, an IC version will produce equal or better results. simply replacing an opamp with a discrete opamp in a badly designed, or even competently designed circuit will not magically remove the issues. Which have been described here as blatantly obvious.
this conversation exactly was recently covered ...again.... in the blowtorch thread. thin film resistors are every bit as linear as your average metal film resistor that you speak of and they can be used on die where needed, in the positions in the circuit that actually have consequence ie. significant voltage transients or tempco related distortions etc. i'll track it down, but a paper was linked that showed that thin film on die resistors are very linear and not a source of error to look at first. you just need to look at the electronically controlled ladder type attenuators, like those from Ti and even the audiophile friendly MUSE; which contain at least 2 opamps, some logic and a quad attenuator for less than $10, a fair way short of your $100
I already mentioned the caps being a sticking point, but for 10-20pf, thin film caps are far more linear than c0g. they are limited in capacity, as I mentioned, but you would always use an external cap if a larger cap was needed. it rarely is, except in power supply or decoupling, which should not be considered part of the argument, external decoupling would still be used if need be; as would an RF filter if needed. did anyone say that no supporting ancillary components would be needed and the IC would stand alone?
Last edited:
dsl ? http://www.ti.com/lit/ds/symlink/tpa6120a2.pdf
not mentioned there, but seems is picky about pcb layout though.. (p17)
Another way to look at this is to remember how you (used to) do your best discrete designs.
You start with a need, an idea, something you feel would be nice to build. As the design develops and you try things out, you realise you really should put in a current source here, and better put in a buffer there, maybe upgrade this reference from a simple zener or LED to a real voltage reference. Before you know, you are actually getting towards the complexity - and thus the performance - of an opamp!
So there is a continium of designs, and the better performance you get, the closer you get to an IC implementation.
Linear IC's with non-linearity of just a few ppm, offset voltages of a few uV and thermal drift of less than that and BW of sveral 100Mhz just are not practical in discretes. Even the power spec gets overtaken quickly: look at the LT1210 that can deliver more than 1A from a 5-pin TO220 case with all other specs making you drool.
I'm not saying discretes and ICs cannot sound diffferently - they can and often do. I also accept that some prefer discretes and some prefer ICs for whatever reason. But as to objective performance, I'm pretty sure I can match any discrete circuit with an integrated solution with better performance for a freaction of the cost. Except, as noted, some high voltage, high power and other niches.
jan
You start with a need, an idea, something you feel would be nice to build. As the design develops and you try things out, you realise you really should put in a current source here, and better put in a buffer there, maybe upgrade this reference from a simple zener or LED to a real voltage reference. Before you know, you are actually getting towards the complexity - and thus the performance - of an opamp!
So there is a continium of designs, and the better performance you get, the closer you get to an IC implementation.
Linear IC's with non-linearity of just a few ppm, offset voltages of a few uV and thermal drift of less than that and BW of sveral 100Mhz just are not practical in discretes. Even the power spec gets overtaken quickly: look at the LT1210 that can deliver more than 1A from a 5-pin TO220 case with all other specs making you drool.
I'm not saying discretes and ICs cannot sound diffferently - they can and often do. I also accept that some prefer discretes and some prefer ICs for whatever reason. But as to objective performance, I'm pretty sure I can match any discrete circuit with an integrated solution with better performance for a freaction of the cost. Except, as noted, some high voltage, high power and other niches.
jan
Hey Counter ol' buddy ol' pal!
I'm sure I'm meant to read all that as heavy sarcasm, but just in case you were in any way serious, my first reaction would be: Why should I? Just what kind of "substantial evidence" do I need to express an opinion here? If I told you I preferred a Porsche to a Ferrari because it handles better and is nicer to drive, would you insist that I'm wrong because they have the same specs on paper and demand that I do a double-blind driving test to prove I can tell the difference (now that could be dangerous
)?!Still if you feel like a trip to Sydney to my listening room I would have no problems proving to you I can hear the preamp in a blind test. How could I not invite my ol' buddy over after all?
I don't think so. (WoW, the +21 edition)
Reminds me, how does one compensate the output offset of an opamp ?
I know little about the muse, more about the Ti, I only read of the muse attenuator when it was linked in the BT thread, I think their opamps are an expensive joke, but the attenuator spec read quite well. Why? is it not opamps and a ladder?
compensate the output offset? what makes you ask that? the idea would be to not have input offset... it really depends on what is causing the offset doesnt it? but since I have no idea what you are talking about, i'll simply answer, I have no idea... it would depend on the opamp and the circuit
Last edited:
Not the chip itself under that name, i'll see if I can find it, it was under a different name for the DSL driver.
If I told you I preferred a Porsche to a Ferrari because it handles better and is nicer to drive, would you insist that I'm wrong because they have the same specs on paper and demand that I do a double-blind driving test to prove I can tell the difference (now that could be dangerous
Nicer to drive is subjective opinion and nobody could argue with that.
Better handling is an objective statement and could even be measured or tested (take the same torture course with both cars and see which one gets trough the curves better/faster.
So you're mixing things up a bit.
jan
Theire specs are below their predecessor, the discontinued Wolfson, and MUCH worse than for example a Cirrus CS3318. But they have that 'hi-end' appeal.
jan
On the contrary, owdeo, no sarcasm is intended.
A little lighthearted ragging and a mild admonition against the use of inflammatory language, perhaps.
As to why you should undertake to deliver some proof; to advance the cause of science and because nobody else can.
Have you ever read Eric Berne's book Games People Play? It's central to the psychological model called Transactional Analysis.
It seems to me that the exchanges on many forums, particularly with regard to subjects like these, are variations on what is identified in TA as a Game called YDYB. 'Why don't you - yes but.'
In this game the central protagonist (White) airs a problem. People (Blacks) in his immediate surroundings propose solutions. Whatever solution is suggested, White finds some reason for rejecting it.
From Wikipedia:
White: I wish I could lose some weight.
Black: Why don't you join a gym?
White: Yes but, I can't afford the payments for a gym.
Black: Why don't you speed walk around your block after you get home from work?
White: Yes but, I don't dare walk alone in my neighborhood after dark.
Black: Why don't you take the stairs at work instead of the elevator?
White: Yes but, after my knee surgery, it hurts too much to walk that many flights of stairs.
Black: Why don't you change your diet?
White: Yes but, my stomach is sensitive and I can tolerate only certain foods.
"Why Don't You, Yes But" can proceed indefinitely, with any number of players in the Black role, until Black's imagination is exhausted, and he/she can think of no other solutions. At this point, White "wins" by having stumped Black. After a silent pause following Black's final suggestion, the game is often brought to a formal end by a third role, Green, who makes a comment such as, "It just goes to show how difficult it is to lose weight."
Of course the game in our case is complicated by the fact that numerous players take up the cause of White and the issue is not necessarily posited as a problem requiring a solution, but it seems to me that there is much in the simple game scenario that is mimicked in our exchanges.
Transactional Analysis (briefly) considers people as acting in 3 principal ego-states. Parent, Adult and Child.
From Wikipedia again:
The secondary gain for White was that he could claim to have justified his problem as insoluble and thus avoid the hard work of internal change; and for Black, to either feel the frustrated martyr ("I was only trying to help") or a superior being, disrespected ("the patient was uncooperative").
Superficially, this game can resemble Adult to Adult interaction (people seeking information or advice), but more often, according to Berne, the game is played by White's helpless Child, and Black's lecturing Parent ego states.
So in order for us to be reassured that this is an Adult to Adult conversation and not merely a (possibly unhealthy for us, as well as you) pastime, we really require to know whether there is some factual foundation to what we are discussing.
I suggest that you take 2 amplifiers along to the psychology department of your local university, explain the problem to them, and ask for their help in the form of an affidavit confirming your ability to reliably distinguish between the 2 amplifiers in a blind test conducted by them. We'll require some evidence too that they understand requirements such as level matching.
A little lighthearted ragging and a mild admonition against the use of inflammatory language, perhaps.
As to why you should undertake to deliver some proof; to advance the cause of science and because nobody else can.
Have you ever read Eric Berne's book Games People Play? It's central to the psychological model called Transactional Analysis.
It seems to me that the exchanges on many forums, particularly with regard to subjects like these, are variations on what is identified in TA as a Game called YDYB. 'Why don't you - yes but.'
In this game the central protagonist (White) airs a problem. People (Blacks) in his immediate surroundings propose solutions. Whatever solution is suggested, White finds some reason for rejecting it.
From Wikipedia:
White: I wish I could lose some weight.
Black: Why don't you join a gym?
White: Yes but, I can't afford the payments for a gym.
Black: Why don't you speed walk around your block after you get home from work?
White: Yes but, I don't dare walk alone in my neighborhood after dark.
Black: Why don't you take the stairs at work instead of the elevator?
White: Yes but, after my knee surgery, it hurts too much to walk that many flights of stairs.
Black: Why don't you change your diet?
White: Yes but, my stomach is sensitive and I can tolerate only certain foods.
"Why Don't You, Yes But" can proceed indefinitely, with any number of players in the Black role, until Black's imagination is exhausted, and he/she can think of no other solutions. At this point, White "wins" by having stumped Black. After a silent pause following Black's final suggestion, the game is often brought to a formal end by a third role, Green, who makes a comment such as, "It just goes to show how difficult it is to lose weight."
Of course the game in our case is complicated by the fact that numerous players take up the cause of White and the issue is not necessarily posited as a problem requiring a solution, but it seems to me that there is much in the simple game scenario that is mimicked in our exchanges.
Transactional Analysis (briefly) considers people as acting in 3 principal ego-states. Parent, Adult and Child.
From Wikipedia again:
The secondary gain for White was that he could claim to have justified his problem as insoluble and thus avoid the hard work of internal change; and for Black, to either feel the frustrated martyr ("I was only trying to help") or a superior being, disrespected ("the patient was uncooperative").
Superficially, this game can resemble Adult to Adult interaction (people seeking information or advice), but more often, according to Berne, the game is played by White's helpless Child, and Black's lecturing Parent ego states.
So in order for us to be reassured that this is an Adult to Adult conversation and not merely a (possibly unhealthy for us, as well as you) pastime, we really require to know whether there is some factual foundation to what we are discussing.
I suggest that you take 2 amplifiers along to the psychology department of your local university, explain the problem to them, and ask for their help in the form of an affidavit confirming your ability to reliably distinguish between the 2 amplifiers in a blind test conducted by them. We'll require some evidence too that they understand requirements such as level matching.
Last edited:
@ Jan:
hmm fair enough, I use fully digital volume on everything and have for a few years, so i'll just trust you on that. I do have some experience with the 4 channel Ti PGA4311, which worked quite well, but in the end I didnt like adding 2 opamps to a discrete amp just for the volume control, which led me to pursue digital more fully.
yeah the cirrus sounds interesting, caught mention of that a couple months ago, but again I dont build things with volume controls generally now.
agreed about that 'hi-end' appeal, not sure if the volume control has the OFC leadframe.... I find that quite hilarious
just a reminder, I only brought these volume control chips up as an example of on chip thin film resistor linearity
hmm fair enough, I use fully digital volume on everything and have for a few years, so i'll just trust you on that. I do have some experience with the 4 channel Ti PGA4311, which worked quite well, but in the end I didnt like adding 2 opamps to a discrete amp just for the volume control, which led me to pursue digital more fully.
yeah the cirrus sounds interesting, caught mention of that a couple months ago, but again I dont build things with volume controls generally now.
agreed about that 'hi-end' appeal, not sure if the volume control has the OFC leadframe.... I find that quite hilarious
just a reminder, I only brought these volume control chips up as an example of on chip thin film resistor linearity
Last edited:
The ones I ordered straight from Japan cost me $25/pc (ex fckng taxes)
Digikey is yet to stock the parts, but list the SSOP32 part at $16/pc and the 02 opamp at $50 each, at 100 minimum order (plus taxes).
Luxman has a 72320/02 volume attenuator in it's latest $6k balanced headphone amp, P-700U. (plus others)
Esoteric also uses the combo, e.g. in their latest C-02 preamp top model ($23.5k retail in the US)
PassLabs does in their XP30, +$16K retail there, +$23k here.
As does my old acquaintance Günther Mania is his current AVM preamp top model, the modular PA-8 ($10K starters to $20K fully loaded)
Could name a few more, but you get the picture.
What all of them differ in, is the way the ladder chip is implemented.
Luxman uses a pretty FET buffer in front, Pass goes all discrete, and Günther still does BB tricks after using the OPA627 for two decades.
A decent application of the chip does a multitude of 10 bucks. (complete volume attenuator modules in Japan now go for $500/stereo, you can guesstimate what the balanced 4-decker version does)
But Hey, really cool stuff for fully balanced headphone amps.
(rotten shame the Luxman amp lacks oomph for my AKG ESL levitation flaps)
The offset compensation remark is for you to think about, I'm just the souffleur head sticking out of the floor.
Or in the words of Mr AVM, good luck with the fiddlin.
Attachments
Last edited:
Oh, OK. If you're talking about devices that don't actually exist then I have to agree, some of them are very good, and don't suffer from the problems and limitations afflicting currently available opamps.
Jacco, if you just mean that in most cases with opamps we have no access to the nodes we would use to correct for offset like we would with discrete, then sure, I presumed you meant something else. of course if its steady state DC from an external source thats easy enough, just feed the same DC to the inverting input, but not so effective for dynamic DC error signal components.
but again, there is no reason that couldnt be added if the demand was there for it, it has nothing to do with the technology and as I said, I prefer to supply no offset and use opamps that work correctly for the job at hand. thats jhust part of the correct selection and testing process, no matter what parts.
my headphone amps, my poweramps and most of my line level gear is DC coupled, except for the dac IV in my current system which uses a pass D1 variant IV stage by OPC. I dont have any DC problem on any of them and everything is balanced, so it would have to be significant
regarding the muse, right OK so $16... a far cry from $100
I get the picture, companies that sell expensive equipment often have a hard time using opamps, because of the notion in the marketplace that they are somehow musically corrupt, very common and not exclusive enough, but opamps make for generally shorter time to market, better measurements etc which are also desireable, so that creates a conundrum. it makes it easier when they can advertise more expensive OFC leaded audiophile buzzword approved opamps, with magical advertising literature for the $50 chips, rather than a $3 lme49990 for example.
yes i'm a cynic, but this speaks at least as much about the customers.
but again, there is no reason that couldnt be added if the demand was there for it, it has nothing to do with the technology and as I said, I prefer to supply no offset and use opamps that work correctly for the job at hand. thats jhust part of the correct selection and testing process, no matter what parts.
my headphone amps, my poweramps and most of my line level gear is DC coupled, except for the dac IV in my current system which uses a pass D1 variant IV stage by OPC. I dont have any DC problem on any of them and everything is balanced, so it would have to be significant
regarding the muse, right OK so $16... a far cry from $100
I get the picture, companies that sell expensive equipment often have a hard time using opamps, because of the notion in the marketplace that they are somehow musically corrupt, very common and not exclusive enough, but opamps make for generally shorter time to market, better measurements etc which are also desireable, so that creates a conundrum. it makes it easier when they can advertise more expensive OFC leaded audiophile buzzword approved opamps, with magical advertising literature for the $50 chips, rather than a $3 lme49990 for example.
yes i'm a cynic, but this speaks at least as much about the customers.
Last edited:
not only, but for that phrase yes, I was simply commenting that its not a baked in limitation of the technology like many suggest, rather the economics of the large corporations that mostly make opamps. building high voltage or high power opamps for audio is pretty niche and probably wouldnt pay for itself, so they dont do it.
besides class d solves this for the masses, as they can do both without having to handle a heap of heat. some come close, needing only a few external power devices and I think as short run fabrication becomes more accessible, we will see more high power audio specific power devices again.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.