Discrete vs opamps...
Jan 27, 2006 at 7:37 AM Post #106 of 176
Quote:

Originally Posted by gpalmer
Though since you didn't phrase it that way, rather you said:

So for a given pricepoint, why do people prefer opamp designs over discrete designs, if you take modding out of the equation?

and I responded to that. Nobody said the amplifiers had to be "sucky" whatever that technical term means!
biggrin.gif
And contrary to to the black and white universe portrayed here, I've heard amps that were Class A which didn't sound as good to me as amps running Class AB so I'm not buying the basic argument that ALL Class A will always sound better than ALL Class AB. I learned long ago in Physics 101 that generally such statements should be greeted with a great deal of skepticism and you know what? So far it's worked that way in the real world also!
tongue.gif
I will agree that a Class A amplifier implmented with equal technical merit and skill will generally sound better than an amplifier running in Class AB, though since I have not heard and compared the universe of possibilities, I cannot state that they ALL do. Lastly, sicne the huge majority of amplifiers sold are well under the $300 mark, I would regard that as raising the bar fairly high. *shrug* Depends on your frame of reference I guess...



My two above statements in no way contradict each other. Stating that you can build a bad sounding amp with either approach and any given pricepoint is stating the obvious. Stating that you can build a better sounding AB amp than a bad sounding A amp is also stating the obvious and ignoring the parameter of a similar pricepoint (=quality).

As for really cheap amps... I should have made that clarification I guess, but the vast majority of headphones sold are also well under $100, but I don't see much debate on the various merits of different $15 earbuds on head-fi.

From my shopping, $300 new is about the entry level in the true hi-fi (non-portable) headphone amp market. Anything battery powered is inherently severely compromised in sound quality anyway...
 
Jan 27, 2006 at 7:43 AM Post #107 of 176
Quote:

Originally Posted by rickcr42
Honest electronics are what every one says they want until they actually hear it many times.

System matching,taste,need to cover up other flawed areas in the system withthe amp,all kinds of reason why choices are made



Couldn't agree more with you, but then that's my preference and opinion and in the grand scope of things counts for little...
tongue.gif


Quote:

Originally Posted by rickcr42
I do not think I would live to see the day when someone would pay $5,000 for an "Op-Amp in a box" (OK,Maybe a headroom but that is a different creature
very_evil_smiley.gif
) but see that in the discrete arena often.Is this just elitism and having no merit ? They are being suckered because they could have purchased an opamp based amp just as good for $500 and had $4,500 left over for CDs ? My opinion is no.My opinion is you get what you pay for even though less for more as you climb the scale and that means discrete or tubes but NOT opamps.



I agree with you and have all along, to my mind there is no doubt that the discrete amplifiers are ultimately better when you're playing for all the marbles. Your commentary about Class A versus other Classes is a large part of the reason why. I would probably raise the bar higher than just portables and I would also include areas where minimal size is important but it strikes me as squabbling over details since we're probably talking a couple of hundred dollars difference which sort of misses the point in a discussion about $5000 components. Yeah, you can run Windows without a mouse but as the French say, you have to have a mind for the trivial!
biggrin.gif
 
Jan 27, 2006 at 11:48 AM Post #108 of 176
Getting away from class distinction for a bit.
If I understand correctly,one of the points about the use of opamps that
some object to is the large amount of negative feedback required.
The preference for discrete circuits designed specifically for audio is that
linearity can be addressed better by careful stage by stage design rather
than wrapping a high feedback around the whole lot as a cure all.

Opinions?


.
 
Jan 27, 2006 at 12:01 PM Post #109 of 176
I am about to replace my current phono stage (which works with ICs in the first stage and has a discrete output stage) with another model that uses a fully discrete circuit. The MSRP for both models is almost identical, but the discrete one has been the darling of audio reviewers ever since it hit the market.

I am going to give discrete a chance!
 
Jan 27, 2006 at 2:35 PM Post #110 of 176
Quote:

Originally Posted by aerius
When op-amp chips are biased into "class A", they only stay in class A up to a point. Crank the volume up a bit and they drop out of class A into class AB or B operation. Only the first 10% or so of its maximum rated output is class A.

A discrete class A design stays in class A no matter how much you crank the volume. 100% of it's max rated output is class A.

Here's the problem, class A is very inefficient, you're lucky to get 5-10% efficiency, the rest goes up as heat. In other words, for a 1W output, you'll need at least 10W of power, which means large parts with big heatsinks. It doesn't matter where the volume dial is, the amount of power it sucks is proportional to the maximum rated output of the device, not how much power it happens to be putting out at the moment. The 1W amp may only be putting out 0.01W at a given time, but it'll still suck a full 10W.



1W is high for driving headphones. The reason I mentioned about opamps that I have used in my design (and apparently rickcr42 missed my point) was that there are power opamps that can dissipate 10W. So I don't know why he keeps using examples of opamps with max output current of 35mA to demonstrate the idea that at full Class A an opamp will go thermal runaway and be destroyed be heat. There are several opamps out there that can do several Amperes of output current and they are designed to dissipate several watts if not tens of watts.
 
Jan 27, 2006 at 2:55 PM Post #111 of 176
Quote:

Originally Posted by JahJahBinks
1W is high for driving headphones. The reason I mentioned about opamps that I have used in my design (and apparently rickcr42 missed my point) was that there are power opamps that can dissipate 10W. So I don't know why he keeps using examples of opamps with max output current of 35mA to demonstrate the idea that at full Class A an opamp will go thermal runaway and be destroyed be heat. There are several opamps out there that can do several Amperes of output current and they are designed to dissipate several watts if not tens of watts.




I dont think Rick has missed any point. He has explained the differences what .... three or four times. But then someone misunderstands and he ends up addressing something further and further away from his original points. All I can say is thanks Rick. Thanks for taking the time to explain the differences .... I appreciate your efforts and I learned some info that enhanced my understanding of the subject.
 
Jan 27, 2006 at 3:46 PM Post #112 of 176
Quote:

Originally Posted by rickcr42
fully explained at the very beginning of this thread


You explained what class A is, but not how it makes headphones sound better. In order to understand that, I would need to understand how headphones work and how they behave with different class A/AB/B devices.

I dont completely buy the explanation that "fully on" is always better than 10% class A and 90% class AB. My light bulb is never fully on either, it's cycling on and off at 60 times a second, but my eyes can't see it.


Quote:

Originally Posted by rickcr42
The thing about fact is they are what they are and do not waver.to deny something is "factual" when it is and is known to be is just being argumentive.


Your explanation of class A is factual. Your statement that "full" class A *sounds* better than "hanging" class A is an assertion -- what you want is people to accept that as fact based on your experiences with them.


Quote:

the internet is a resource people and if my "facts" bother you then do your own research,compare actual amps from the various groups,listen,think,form your own opinion.that is valid.


agreed.
 
Jan 27, 2006 at 4:08 PM Post #113 of 176
Quote:

I'm going to build a class-a op-amp with a cooler from an ATX motherboard southbridge and thermal adhesive as a proof-of-concept, one of these days



well you better crack open the opamp plastric casing then figure out how you are going to use an etched substrate transistor with very little actual metal to bond to an external heat sink or you have no shot.No heat transfer and so pissing in the wind.
An audio output transistor is not a computer chip processing ones and zeros but a continiously active device,more so if full tilt boogie class A.think analog and not digital logic

Quote:

It is relevant to note, as you did, that some opamps are designed for audio. I don't know if there are others, however the 627/637 were designed with a specific use in mind. Of course, they cannot be completely customised to the circuit in which they are placed, however they do have an intended use right back to the design stage.


yet you still have no access to the front end,any of the the bias points at any stage,the current mirrors,the driver stage or the output stage so a "box" that you can hang things on,change what happens OUTSIDE the box but not a single internal parameter other than opening up one single output transistor of the output pair to class-a operation on a limited bassis.An improvent but by nature of the beast a limited one.
Is what it is and wishing does not make it any different

Quote:

My two above statements in no way contradict each other. Stating that you can build a bad sounding amp with either approach



Who has ? what examples ? Name a single "bad sounding" class-a discrete Op-Amp that is for sale anywhere,just one single example will do.
Theory is OK in its place and if it has any merit but argue some actuals.In "theory" anyrhing is possible but I live in the real world and use real world examples to illustrate points ( looks back to pet peeve rant)


Quote:

As for really cheap amps... I should have made that clarification I guess, but the vast majority of headphones sold are also well under $100, but I don't see much debate on the various merits of different $15 earbuds on head-fi.


Are they ? The vast majority of headphone amps from a legitimate outlet and with any kind of factory backing and from someone who actually did a design ? Or are we talking about DIYers stuffing boards on the kitchen table who could not put a LED in the chassis without asking in the DIY forum what size current limiting resistor to use.Obviously that person could not even do a single transistor stage without a road map but that does not mean it is better.

Anyway.At $65 per I would take one of these over any I.C. on the planet unless the target use is portable (30ma per is not very battery freindly but neither is class-a)

http://www.forsselltech.com/JFET990-2a.PDF

Toss a serious output stage on to provide drive current,say another $25,and a power supply,box it up and the total amp cost comes in at around $250,maybe $300 depending on power supply and if the person doing the build is more interested in "pretty" parts or just good sonics.If commercial would be over $1K because we all know in order to sell it would need to "ooh look how pretty inside" pictures and that means all parts must follow the approved color scheme no matter if they sound best or not and thosae parts are expensive.
So in the $K class if commercial,the $250-$350 area if built for personal use and would be a pure class-A amp in the 2 Watt range able to drive anything out there except the difficult K-1000 and some of the 600 ohm AKG cans.
BUT !
Toss another stage on after the 2W stage (which is now a class-A driver) and you can ramp the output up to whatever you want withing size and $$$$ limitations and you now have a true power amp stage capable of driving whatever it is you are looking to drive.Simple,elegant.reasonably priced and full class-a front to back which I defy any opamp to stand to up in a head to head contest even though this also is an op-amp.that is me using an real world example and not "I feel" then ramble to no purpose

Here is another I would personally choose over any I.C. Op-Amp :

http://www.welbornelabs.com/classa.htm

Again a pure class-A Op-amp at a price of $79 per channel,another $30 prebuilt and ready to drop into a box yet will blow the doors off any I.C. Op-amp as in the first example but with the difference neing this is ready to rock and needs no additional output stage.

Another example,this time a Hybrid :

http://www.welbornelabs.com/hyb.htm


I have actually built a version of this years ago and I can say from my personal experience it blows the doors again off any I.C. Op-Amp I have ever heard.Again is a class-A Op-Amp and again at a good price of $75 per channel minus tubes (the 6GM8 is not real expensive),$30 per channel pre-built.Put it up against any opamp on the planet and hands down it will whip its a*s with only the stubborn few wanting to argue being able to deny it once heard head to head.

Quote:

1W is high for driving headphones.


for an opamp it is.Not for a buffered opamp which most are capable of the 1W and others more when the buffers are stacked.My Szekeres is 2Watts,all my triode headphone amps 2-4 watts.
1W is NOT high for a headphone AMP at all and is very common

Quote:

Couldn't agree more with you, but then that's my preference and opinion and in the grand scope of things counts for little...


Most have no clue what it is they want and even less how things work but always the loudest dissenting voices.They like to argue but have no credibility if they bring nothing to the table other than blowing smoke or calling names.Experts in their own mind because they hate to admit they have no clue and must attack those who do AND that stand by their guns instead of being intimidated into silence
rolleyes.gif

Taste is valid but when that is strictly based on very limited experience or maybe second hand news only a closed mind craps on the options or a new idea just because.

Quote:

The reason I mentioned about opamps that I have used in my design (and apparently rickcr42 missed my point) was that there are power opamps that can dissipate 10W


I did not miss the point at all.This is an audio discussion and those opamps suck for audio so why use them as a counter to what sounds better ? I can stick an ADSL driver in a batery operated box,call it a headphone amp and point to the close to 500ma output as its main feature yet as soon as someone used it for music I would get "What ?"
confused.gif
because it would sound like crap even though it looked good on paper and i know this because I tried this and it did not work out
very_evil_smiley.gif


Quote:

So I don't know why he keeps using examples of opamps with max output current of 35mA to demonstrate the idea that at full Class A an opamp will go thermal runaway and be destroyed be heat. There are several opamps out there that can do several Amperes of output current and they are designed to dissipate several watts if not tens of watts




one more time then (exhales........)

the question was which is better.A discrete Op-amp or an I.C. Op-amp for driving headphones
Still with me ?
Headphones need power,power means heat,heat means either heat sinking or limited power drive or going out of class-A operation and into class AB operation to keep the heat down
following ?
Every single opamp on the planet would use class A operation if they could but since they can not comprimises are made to make it work within limitations.One method is to add an external buffer and is 90% of the headphone amps in use by head-fi members but it is still a comprimise.

NONE of the above pertains to a discrete op-amp circuit so why I initially said it is a better way and if you can not understand what I wrote then it is not my failing to explain but yours to understand since you keep referring to my statements.

Quote:

I dont think Rick has missed any point. He has explained the differences what .... three or four times. But then someone misunderstands and he ends up addressing something further and further away from his original points.


read back a few of my posts man.Where I stated it is my thought this is an intentional attempt to blow smoke on the topic so the crystal clear again becomes cloudy and misinformation can rule the day as it mostly does at headfi.Why I rarely bother trying to clear up unclear topics anymore.Some folks just don't want to know and the ones who do must deal with the fact there are some who would rather they also have no clue.
Why ? I have my suspicians but will keep them to myself in the interest of peace and harmony (lalala lala lala laaaa )

Quote:

I appreciate your efforts and I learned some info that enhanced my understanding of the subject


You are welcome Earl.I really do try to shoot for balanced and clear and fully enjoy a discussion on the topic if it is one that interests me (most don't,boring and same old) but every time I do we end up at the same place.More confusion than clarity and 25 pages of crap fest personal baiting but being thick skinned,having a serious "kiss my behind" attitude and one of the baddest headiers on the planet I WILL as always plow on and try to bash my way to clarity
very_evil_smiley.gif


"lead,follow or get the fk out of my way"


True words, hard to come by
tongue.gif
 
Jan 27, 2006 at 4:44 PM Post #114 of 176
Quote:

Originally Posted by atx
You explained what class A is, but not how it makes headphones sound better. In order to understand that, I would need to understand how headphones work and how they behave with different class A/AB/B devices.

I dont completely buy the explanation that "fully on" is always better than 10% class A and 90% class AB. My light bulb is never fully on either, it's cycling on and off at 60 times a second, but my eyes can't see it.





10% class A / 90% class B = Class A/B. You can be bias the signal more, or less, into class A .... but its still class A/B OPERATION. Hanging class A .... is a term I have never heard used before. But in reality, it definitely remains class A/B operation. If the signal is operating in class A/B there will be crossover distortion at the zero crossing point of the positive and negative waveforms. Therefore, higher class A bias = less crossover distortion ..... less class A bias = higher crossover distortion. This is fact!

How does this effect headphone operation. The headphones receive the waveform = audio signal from the amp. If the amp has more distortion compared to another amp ...... the headphone will pass on the higher distortion signal. The headphone converts the signal into output independent on the class of operation.

Can you hear this? I hear treble smearing with every op amp design I have ever listened to. I dont hear this with the discrete SS or tube designs I have owned. Is crossover distortion the cause of the smearing .... maybe. It could also be a result of the high amounts of loop feedback the monolithic designs require to be stable. But after countless comparisons do I hear equivalent sound quality from the non-discrete designs ..... No I dont.
 
Jan 27, 2006 at 5:03 PM Post #115 of 176
Quote:

You explained what class A is, but not how it makes headphones sound better. In order to understand that, I would need to understand how headphones work and how they behave with different class A/AB/B devices.


missed this but damn,easiest answer yet !
icon10.gif


In laymans Class-A is superior for music signals because it is the only biasing method where the circuit is actually always "ON" so no matter how big or how small the signal is it just rolls right on through from the input to the output intact but increased (what the word "amplifier" means)

With any other method the circuit is "OFF" until activated by an audio signal.What this means in terms of sound is you have a condition where the electronics are always trying to catch up to the "trigger" which is the music itself and must go from off to on during the music/no music stage and where the sonics are mostly effected is in the low level detail that is lost and once lost no amount of "I wish" or headphones no matter how good will bring it back.

Classes other than class-A have always been about cost savings from the very beginning of audio and never about "this is better" so from the start of audio time a known comprimise meant to save money and why ultra expensive audio gear is always class-a.No cost limitations means you go for the best you can achieve
 
Jan 27, 2006 at 7:18 PM Post #116 of 176
Quote:

Originally Posted by rickcr42
for an opamp it is.Not for a buffered opamp which most are capable of the 1W and others more when the buffers are stacked.My Szekeres is 2Watts,all my triode headphone amps 2-4 watts.
1W is NOT high for a headphone AMP at all and is very common



The 1W I mentioned is not the maximum output power an amplifier is capable of but rather it IS the power dumping into the headphone. Try sourcing 1W into Sen 580 or Sony 3000 then see what happens. My point is that you can have an amp that is capable of doing 1W or 100W, but a pair of headphones will never eat 1W or more of power by itself, at that level of power, either the drivers are blown or the sound level is too high to be safe to listen to.
 
Jan 27, 2006 at 7:37 PM Post #117 of 176
Quote:

Originally Posted by rickcr42
I did not miss the point at all.This is an audio discussion and those opamps suck for audio so why use them as a counter to what sounds better ? I can stick an ADSL driver in a batery operated box,call it a headphone amp and point to the close to 500ma output as its main feature yet as soon as someone used it for music I would get "What ?"
confused.gif
because it would sound like crap even though it looked good on paper and i know this because I tried this and it did not work out
very_evil_smiley.gif



That was what I said in post #76. If AD or TI is willing to design a power amp from either discrete opamp design that is proven to sound good, or any of the monolithic opamps you can buy on the market today that sound good and package it in monolithic form with the output power capability of power opamps. I did not miss your post either. From a practical point of view you were comparing the state of the art opamps. It is obvious that power opamps don't have the high level performance of those signal opamps like AD843 and OPA623. That's why I said in the earlier post that if you can marry the two technologies then this new monolithic opamp is in the same ball park as the discrete opamp.

I read the posts over again. What I misunderstood in the beginning was that I thought the discussion was about having a monolithic opamp and a discrete opamp and both need to be connected to a Class A output stage. My mistake was that I did not realized that the Class A output is already designed into the discrete opamp. Then it is obvious the output stage plays a major role here, and the performance benefit asscoiated with laser trimmed monolithic opamp such as lower DC offset voltage and input bias current, slew rate and BWP, etc. have less effect on the sound since it is likely more sound degradation occurs at the output stage .
 
Jan 27, 2006 at 7:45 PM Post #118 of 176
Quote:

Originally Posted by NotJeffBuckley
Pentium 4 prescott dual-core CPUs dissipate more than 200 watts of heat off of a 90nm-process core via thermal paste coupled heatsink and fan;


Just a correction. Thermal design power which is the maximum power a chip is required to dissipate is not the same as power drawn by the CPU. As an example Pentium 620, 630 and 640 all have thermal design power of 84W, even though the chip itself might need 200W of power to operate at certain load.

PS: But this is very trivial and not pertinent to discussion in this thread.
 
Jan 27, 2006 at 8:31 PM Post #119 of 176
Quote:

My mistake was that I did not realized that the Class A output is already designed into the discrete opamp. Then it is obvious the output stage plays a major role here, and the performance benefit asscoiated with laser trimmed monolithic opamp such as lower DC offset voltage and input bias current, slew rate and BWP, etc. have less effect on the sound since it is likely more sound degradation occurs at the output


Cool.That explains a lot and especially the parts where I had to cover old ground over and over beleiving you were just rying to be a pain in the butt
wink.gif


Quote:

That was what I said in post #76. If AD or TI is willing to design a power amp from either discrete opamp design that is proven to sound good, or any of the monolithic opamps you can buy on the market today that sound good and package it in monolithic form with the output power capability of power opamps


Never happen.

Monolithics have in fact moved in the exact opposite direction and to less heat,smaller size and more efficient output stages because truth be told audio in general and for damn sure we in general are so low on the food chain that we may not make the list even.The bread and butter is high volume mass production geared to portable euipment and not high fidelity sound.Even formerly available "pro audio" chips go obsolete far more than they do "new release" versions.DACs,Amps,analog switches,all headed to miniaturisation and RoHs compliance and either alone precludes high fidelity unless entirely by accident but tiogether ?
Not a good future in audio chips unless you are just looking for insane battery life combined with crap sonics and when the OPA627 goes to RoHs compliance I don't know if BB/TI will even bother.Hopefully yes but no gurantees

Quote:

PS: But this is very trivial and not pertinent to discussion in this thread.


well hell man ! I found it interesting ! i don't know crap about processors except they either work or they don't
very_evil_smiley.gif


cheers
 
Jan 27, 2006 at 8:45 PM Post #120 of 176
Quote:

Originally Posted by JahJahBinks
1W is high for driving headphones. The reason I mentioned about opamps that I have used in my design (and apparently rickcr42 missed my point) was that there are power opamps that can dissipate 10W. So I don't know why he keeps using examples of opamps with max output current of 35mA to demonstrate the idea that at full Class A an opamp will go thermal runaway and be destroyed be heat. There are several opamps out there that can do several Amperes of output current and they are designed to dissipate several watts if not tens of watts.


There's this wonderful thing in music known as dynamic range, good recording have at least a 20dB difference between the average level and the peak levels. What does this mean? Read this. It means that most of the time you only need 0.001-0.01W to drive your headphones, but if you want to reproduce those dynamic peaks without distortion, you will need a full watt, maybe more. And to have that full watt of output power available, a class A amp will have to burn at least 10-20W. There is no way around it. You're not going to find a chip-amp which can safely dissipate that kind of heat without blowing up that's suitable for headphone audio use.
 

Users who are viewing this thread

Back
Top