My belief is based into observation about "what others do".
Many factories decided to discrete into some critical functions... you see this function generator, not so old to use discrete because old..in the same circuit you see TL022 and some other chips, and those are not that bad.
Why they decided to use discrete into the main amplifier, the one amplifies the tones generated, sending them to the output connector...in other words, a very important stage and using discrete when they could use operational amplifiers.
Not only this application, where you need low distortion into the waveform, but some others you can see industries, professionals, selecting discrete components to produce critical, important circuits.
Observe the simple three stages, a differential, a very standard circuit, a voltage amplifier and the simetrical output, all them class AB ... it is very interesting, as crossover distortions are not good to appear into a waveform generator output.
If professionals, using the same voltage to the supply, having nice operational amplifiers, if they prefere discrete.... of course this is a clue that reinforce my belief that discrete is better.
There are technicall reasons that makes discrete better, bigger physical distance (space) between input to output, not needing too much neutralization against oscilations.... less interactions between parts because heat and magnetic fields..and some others too.
Do you think the opposite?... can you please explain your ideas to uncle Charlie..i will be very happy to read you and to learn some new things...please, destroy my belief!, this will turn myself more wise..each belief dell down to ground will drive me more wise.... increasing my distance from the disgusting ignorance.
regards,
Carlos
Many factories decided to discrete into some critical functions... you see this function generator, not so old to use discrete because old..in the same circuit you see TL022 and some other chips, and those are not that bad.
Why they decided to use discrete into the main amplifier, the one amplifies the tones generated, sending them to the output connector...in other words, a very important stage and using discrete when they could use operational amplifiers.
Not only this application, where you need low distortion into the waveform, but some others you can see industries, professionals, selecting discrete components to produce critical, important circuits.
Observe the simple three stages, a differential, a very standard circuit, a voltage amplifier and the simetrical output, all them class AB ... it is very interesting, as crossover distortions are not good to appear into a waveform generator output.
If professionals, using the same voltage to the supply, having nice operational amplifiers, if they prefere discrete.... of course this is a clue that reinforce my belief that discrete is better.
There are technicall reasons that makes discrete better, bigger physical distance (space) between input to output, not needing too much neutralization against oscilations.... less interactions between parts because heat and magnetic fields..and some others too.
Do you think the opposite?... can you please explain your ideas to uncle Charlie..i will be very happy to read you and to learn some new things...please, destroy my belief!, this will turn myself more wise..each belief dell down to ground will drive me more wise.... increasing my distance from the disgusting ignorance.
regards,
Carlos
Attachments
Last edited:
The likes of AD797, OPA827 and OPA627 are really hard to beat. At the moment I'm enjoying the combination of OPA827 and OPA627 in a headphone amp. It leaves very little to wish for.
I think one of the problems with opamps is that they sound a bit thin, lacking some fullness. This could be perceived as being detailed and give the impression of a grander sound stage. When measured, most opamps seem to be dominated by 3rd harmonics, and there's very little 2nd. My discrete amps are dominated by 2nd and have some 3rd. Maybe this could explain some of the differences.
On the positive side, it's very easy to upgrade an opamp based amp or DAC when newer and better devices arrive, or when you think it's worth spending some money on those higher grade opamps.
I think one of the problems with opamps is that they sound a bit thin, lacking some fullness. This could be perceived as being detailed and give the impression of a grander sound stage. When measured, most opamps seem to be dominated by 3rd harmonics, and there's very little 2nd. My discrete amps are dominated by 2nd and have some 3rd. Maybe this could explain some of the differences.
On the positive side, it's very easy to upgrade an opamp based amp or DAC when newer and better devices arrive, or when you think it's worth spending some money on those higher grade opamps.
Last edited:
Carlos,
Good thread you are starting here. I think you will have as many opinions as they are members of this forum.
My take is that it depends what you define as being 'better'. One thing for sure, do not trust these so called professional to make the 'better' choice because they are not necesseraly looking for the same goals. And I am an EE so I know what I am talking about! Who knows, they might simply have chosen a discrete design to simplify legal issues, or because they already have the design so it is cheaper to reuse it. Or the procurement office has a beef with Sanken! You'd be surprise!
For me, using discrete designs allow me to play with the sound. For example, I can change local feedback caps, I can change polarization, I can implement dual supplies, etc. If I'd use an amp IC, simply attaching a supply to it wouldn't be enough to sustain my 'pleasure' and get me the feeling that 'I've done something'.
After, alls we want is to get the same adrenaline rush of exquisite sound as when we heard our first crappy sound project way back when !!
Ciao,
Patrick
Good thread you are starting here. I think you will have as many opinions as they are members of this forum.
My take is that it depends what you define as being 'better'. One thing for sure, do not trust these so called professional to make the 'better' choice because they are not necesseraly looking for the same goals. And I am an EE so I know what I am talking about! Who knows, they might simply have chosen a discrete design to simplify legal issues, or because they already have the design so it is cheaper to reuse it. Or the procurement office has a beef with Sanken! You'd be surprise!
For me, using discrete designs allow me to play with the sound. For example, I can change local feedback caps, I can change polarization, I can implement dual supplies, etc. If I'd use an amp IC, simply attaching a supply to it wouldn't be enough to sustain my 'pleasure' and get me the feeling that 'I've done something'.
After, alls we want is to get the same adrenaline rush of exquisite sound as when we heard our first crappy sound project way back when !!
Ciao,
Patrick
No.
The output must be capable of withstanding a short-circuit at the max amplitude, and at 2MHz. In this case, the output impedance is a mere 600ohm, unlike the usual 50ohm, but even that would have been too much for "normal" op amps of the time
A function generator has a pretty high distortion figure anyway, typically 1%, 0.3% for very good ones (I'm not talking about "modern", synthetised F.G's, or AWG's).
An AD797 would easily outclass this output stage on every aspect, and even an humble NE5534 could do a good job too... but not a TL022, or a µA741, LM301, etc.
Note that I am not myself an IC "pusher": if you look into the circuits I have proposed, you will see little or no opamps: I am perfectly capable of designing without them, but can also recognize their true value.
One last point: function generators are not designed to be listened to. Or maybe I am missing a point in some new audiophile trend...
Hello Gaetan, Nelsonvandal and Elvee
Thank you by the contribution..
I see Elvee, op amps cannot withstand with so low impedance for long time...but, if they want op amps because superiority they could use it with a simetrical output transistor pair to support the low impedance job.
regards,
Carlos
Thank you by the contribution..
I see Elvee, op amps cannot withstand with so low impedance for long time...but, if they want op amps because superiority they could use it with a simetrical output transistor pair to support the low impedance job.
regards,
Carlos
Hello
There is also the THS4031 and LM4562, but the LM4562 are less good for the soundstage.
I use myself a OPA637 for a phono preamp, the sound of this op-amp are quite close to a discrete circuit amp.
Have you try biasing an op-amp into class A ?
Bye
Gaetan
Last edited:
Thank you Patrick, i think we wrote same time and i could see your message
latter.... just now.
about definition...better option to engineers decide to put a chip in place of discrete or the opposite...in the reality there's a question hidden into my afirmative, and the question is:
Why they have used discrete instead of chips, as they are using chips in other subcircuits in the same generator, why this option, why not a chip into the output if chips are good enougth to output (having some discrete stage to help face low impedance and shorts of course)
In my belief discrete is much better, and all i want is someone to change my mind about..or to confirm, as i am sure, but not 100 percent sure, maybe 95 percent sure.
One reason is the low impedance, but a simetrical output can fix that and the Generator designer could use a chip instead of discrete amplifier, so, the decision was not because of that... i would like to enter the designer head to understand his reasons, in my mind is because discrete is much better, but may exist other reasons and this thread will clarify the issue.
Those chips Nelson is talking about can hold a headphone and may work into 600 ohms short too, observe the generator output, even shorting the connector, the amplifier will see minimum of 600 ohms, because the two resistance in series.....so, chips that can work with 600 ohms loads can withstand the short into the generator output.
The main reason makes me suspect the chip performance is the fact that we cannot have small distance, small gap from output to input, and chip has this small gap because the size, so, if the hell thing is not oscilating is because has circuits to avoid, and these circuits (or parts), for sure kill sonics.
thank you,
regards,
Carlos
latter.... just now.
about definition...better option to engineers decide to put a chip in place of discrete or the opposite...in the reality there's a question hidden into my afirmative, and the question is:
Why they have used discrete instead of chips, as they are using chips in other subcircuits in the same generator, why this option, why not a chip into the output if chips are good enougth to output (having some discrete stage to help face low impedance and shorts of course)
In my belief discrete is much better, and all i want is someone to change my mind about..or to confirm, as i am sure, but not 100 percent sure, maybe 95 percent sure.
One reason is the low impedance, but a simetrical output can fix that and the Generator designer could use a chip instead of discrete amplifier, so, the decision was not because of that... i would like to enter the designer head to understand his reasons, in my mind is because discrete is much better, but may exist other reasons and this thread will clarify the issue.
Those chips Nelson is talking about can hold a headphone and may work into 600 ohms short too, observe the generator output, even shorting the connector, the amplifier will see minimum of 600 ohms, because the two resistance in series.....so, chips that can work with 600 ohms loads can withstand the short into the generator output.
The main reason makes me suspect the chip performance is the fact that we cannot have small distance, small gap from output to input, and chip has this small gap because the size, so, if the hell thing is not oscilating is because has circuits to avoid, and these circuits (or parts), for sure kill sonics.
thank you,
regards,
Carlos
Last edited:
Another factor to think about ... when you build discrete you will never run out of parts. Today's "hot" driver IC for power amps might be discontinued next year.
With a full discrete circuit a sub can always be found and usually a small component change can compensate for the newer, faster device (transistor).
OS
With a full discrete circuit a sub can always be found and usually a small component change can compensate for the newer, faster device (transistor).
OS
I haven't heard THS4031 but LM4562. I don't think it belongs in the top segment, rather in the middle. Better than most but not quite good enough.
Regarding class A biasing of opamps, I enjoyed it for a while until i did a blind listening test and could hear no improvements.
Carlos, I most often use a discrete buffer with the opamps.
Yes Nelson, i use to see your ideas.
Ostripper said makes sense too, one more reason to use discrete.
There's another, when a discrete amplifier burn you have not to substitute the entire amplifier (entire chip).
Soon we gonna have someone thinking the opposite,, and would be very interesting to know that guy's reasons.
Maybe i am comparing apples and bananas, because the discrete is an amplifier, and audio amplifier, when the chip is an operational amplifier, that can be used to audio but has other functions, other applications... maybe i should understand that a dedicated, customized chip for audio could be a better option to be compared with a discrete... so, why they have not used an audio amplifier chip in the place of discrete amplifier made of a function generator that will work, mainly, into audio frequencies, and sometimes will generate above the spectrum...well, have other uses too, but they decide to use an audio amplifier, not a RF amplifier that use to be class A.
I have spent a couple of days studying and watching several schematics and service manuals from function generators, and they use more discrete than chips...is this because discrete is better the way i think and imagine..better performance and more reliable.
Can we can trust (more) in discrete transistors, when working with power?
Maybe discrete produces less distortion?
A chip is less expensive, and factories are always searching for cost reductions to increase profit..why they have not used a chip instead a discrete?
Also, now a days, audiophiles and engineers (majority) believe the amplifiers, to be good, needs CCS, sinks, mirrors, voltage regulators, protections, and a lot of things...the circuit uses nothing!>..the amplifier used for an "instrument", to professional use and purposes, use nothing special, a naked, simple, standard and lovely amplifier.
regards,
Carlos
Ostripper said makes sense too, one more reason to use discrete.
There's another, when a discrete amplifier burn you have not to substitute the entire amplifier (entire chip).
Soon we gonna have someone thinking the opposite,, and would be very interesting to know that guy's reasons.
Maybe i am comparing apples and bananas, because the discrete is an amplifier, and audio amplifier, when the chip is an operational amplifier, that can be used to audio but has other functions, other applications... maybe i should understand that a dedicated, customized chip for audio could be a better option to be compared with a discrete... so, why they have not used an audio amplifier chip in the place of discrete amplifier made of a function generator that will work, mainly, into audio frequencies, and sometimes will generate above the spectrum...well, have other uses too, but they decide to use an audio amplifier, not a RF amplifier that use to be class A.
I have spent a couple of days studying and watching several schematics and service manuals from function generators, and they use more discrete than chips...is this because discrete is better the way i think and imagine..better performance and more reliable.
Can we can trust (more) in discrete transistors, when working with power?
Maybe discrete produces less distortion?
A chip is less expensive, and factories are always searching for cost reductions to increase profit..why they have not used a chip instead a discrete?
Also, now a days, audiophiles and engineers (majority) believe the amplifiers, to be good, needs CCS, sinks, mirrors, voltage regulators, protections, and a lot of things...the circuit uses nothing!>..the amplifier used for an "instrument", to professional use and purposes, use nothing special, a naked, simple, standard and lovely amplifier.
regards,
Carlos
Last edited:
Carlos, design decisions are often less straight and less rational than we expect them to be.
Usually these flaws are cured, if the equipment does its job as it is intended to do.
Sometimes it is not good to poke to much in these solutions.
Regards
That may make sense , but there are other reasons why each (IC vs. discrete)has it 'place".
If you look at a popular IC driver ... (LME49810 - for example)
It's VAS is nothing special -
An externally hosted image should be here but it was not working when we last tested it.
Just a current sourced one with a self type 2 driver stage. Where we get different is the input stage (attached). Here we have 7 CCS's, mute circuit, 3 current mirrors , and baker clamps (only in the 49810).
They do this for 2 reasons :
1 .The consumer offerings will be using cheap power supplies (IC has super PSRR).
2. They don't want any failures , even if abused by the end user (idiotproof - almost
)They can do this because having 200 transistors on 1 piece of sand is easy.
As far as the sound , seeing the wide variety of outputs one can drive with these (BJT , darlington , mosfet) , one can not modify the base circuit to give the optimum sound for each type. One COULD do this with a full discrete. Not to say these IC's COULD NOT sound good , but they are made for general high power amplification (OEM) when mated with the right output stage.
OS
Attachments
Interesting your analisis Ostripper, but remains my central question
and i am feeling my faith reinforced those days, my belief that discrete is better.
All you said, the circuit description, the features, all that stuff makes me think the operational amplifier is a stronger competitor related the simple audio amplifier.
The hell question, that made me be a simple man, producing cheap and easy analisis, is that discrete is better.... they have not used chips to the output, when they had some nice ones in 1990, not this modern you have explained us,but they had some good ones...but they have not used!
When they produce instruments it is interesting to save space, in special this kind of generator that use to be powered by batteries that send energy to a switching power supply and them to a voltage regulator, also they use the optional transformer power supply...well, all that needs room, also audio amplifier, the discrete three stages, needs room......better to use integrated circuit to save space...but they have decided for discrete.
This confuses me searching the reasons, the obvious answer to me is that discrete is better... i know this is an enormous simplification about the problem, but everything is showing me the IC is a better idea, also bigger appeal to customers, saving space and reducing circuit density of parts and cost....but even this way, knowing that good ICs exists, they decided this way.
All those CCS, the baker clamp, the stabilizers, the sinks, well... the operational amplifier seems to be much more precise to measurements, well, more precise to produce a cleaner waveform, and this what we want into a test equipment, clean waveform, low distortion.... the integrated circuits had much better chance to match the engineer needs.... but they decide by discrete!
So, discrete is better...how is better and why is better in such kind of applications is the main question that may give me some inputs.... to destroy the belief or to reinforce, both are good results!
Power and impedance was interesting..... maybe more reliable the discrete is also another input, someone remembered me the signal is 1 percent distorted...well.... would be better to be less distorted to an instrument, and some chips provides lower distortion to measurements despite sonics produces an audible "signature"
I thank you all, including my friend Juergen of course, but i am still waiting for more inputs.
Please boys, try to destroy my belief discrete is better.....i have so many lovely chips here, and i am resistant to use them, as i do not believe them... a terrible waste of good chips here at home....makes me sad to watch them.
My local friends hate chips, i cannot made them a gift.
regards,
Carlos
and i am feeling my faith reinforced those days, my belief that discrete is better.
All you said, the circuit description, the features, all that stuff makes me think the operational amplifier is a stronger competitor related the simple audio amplifier.
The hell question, that made me be a simple man, producing cheap and easy analisis, is that discrete is better.... they have not used chips to the output, when they had some nice ones in 1990, not this modern you have explained us,but they had some good ones...but they have not used!
When they produce instruments it is interesting to save space, in special this kind of generator that use to be powered by batteries that send energy to a switching power supply and them to a voltage regulator, also they use the optional transformer power supply...well, all that needs room, also audio amplifier, the discrete three stages, needs room......better to use integrated circuit to save space...but they have decided for discrete.
This confuses me searching the reasons, the obvious answer to me is that discrete is better... i know this is an enormous simplification about the problem, but everything is showing me the IC is a better idea, also bigger appeal to customers, saving space and reducing circuit density of parts and cost....but even this way, knowing that good ICs exists, they decided this way.
All those CCS, the baker clamp, the stabilizers, the sinks, well... the operational amplifier seems to be much more precise to measurements, well, more precise to produce a cleaner waveform, and this what we want into a test equipment, clean waveform, low distortion.... the integrated circuits had much better chance to match the engineer needs.... but they decide by discrete!
So, discrete is better...how is better and why is better in such kind of applications is the main question that may give me some inputs.... to destroy the belief or to reinforce, both are good results!
Power and impedance was interesting..... maybe more reliable the discrete is also another input, someone remembered me the signal is 1 percent distorted...well.... would be better to be less distorted to an instrument, and some chips provides lower distortion to measurements despite sonics produces an audible "signature"
I thank you all, including my friend Juergen of course, but i am still waiting for more inputs.
Please boys, try to destroy my belief discrete is better.....i have so many lovely chips here, and i am resistant to use them, as i do not believe them... a terrible waste of good chips here at home....makes me sad to watch them.
My local friends hate chips, i cannot made them a gift.
regards,
Carlos
Attachments
Last edited:
Discret is better
Very interesting topic Mr Carlos
I am everything but an expert but I know at least one case, a particular one, well let's say two when it can be argued that discret is really better
1) Sanken multiemitter bjts - these are quite exceptional from what I read. And they can outperform a parallel of similarly speced bjts
2) Those matched fet/bjts on the same chip
I would say that, at least in these cases, discrete is not surely better.
Another consideration is that op-amps or chip amps have usually a very high number of active components inside, someone would say too many.
Usually discrete topology are simple, more straightforward, with better "dynamic" properties ? I am guessing here, as always of course
A lot of respected audio designers admit that simpler schemas have the potential for best sound
This is not the case of op-amp where 30-40 and more bjts can be found
Nevertheless I cannot help noticing that a number of so called high end preamp and DAC output stages use op-amp for buffering and voltage gain purposes.
Very interesting topic indeed, as always your posts Mr Carlos
Thank you very much
Kindest regards,
gino
Very interesting topic Mr Carlos
I am everything but an expert but I know at least one case, a particular one, well let's say two when it can be argued that discret is really better
1) Sanken multiemitter bjts - these are quite exceptional from what I read. And they can outperform a parallel of similarly speced bjts
2) Those matched fet/bjts on the same chip
I would say that, at least in these cases, discrete is not surely better.
Another consideration is that op-amps or chip amps have usually a very high number of active components inside, someone would say too many.
Usually discrete topology are simple, more straightforward, with better "dynamic" properties ? I am guessing here, as always of course
A lot of respected audio designers admit that simpler schemas have the potential for best sound
This is not the case of op-amp where 30-40 and more bjts can be found
Nevertheless I cannot help noticing that a number of so called high end preamp and DAC output stages use op-amp for buffering and voltage gain purposes.
Very interesting topic indeed, as always your posts Mr Carlos
Thank you very much
Kindest regards,
gino
Monolithic ICs are cheaper than discrete, but its limitation stop it from keep moving. The TDA20xx (for example) got its nice sounding with TDA2030A and TDA2050, seems going worse when they increase their wattage on TDA2052 then series stopped. The same thing happened to TDA729x, this one has many reliability problem. DMOS are strong and reliable, but now become as fragile as BJTs. The other (says LMxxxx) applied many protection circuit inside the IC, but the protectors are active frequently.
Discrete design are much more free. Place this here and there, good thermal maintenance, less fabrication failures, the designers are free from creating something. Discrete means, good design = good goods; bad design = worse woorses.
That just what I know about why discrete, hope it help a little
Discrete design are much more free. Place this here and there, good thermal maintenance, less fabrication failures, the designers are free from creating something. Discrete means, good design = good goods; bad design = worse woorses.
That just what I know about why discrete, hope it help a little
Hello !
What I intended is something like this
http://www.chinaicmart.com/uploadfile/ic-circuit/20097663926987.gif
Sometimes I see signal bjts/fets I think of a differential pair put one against the other closely
More closely than in the same chip I think is difficult
I expressed myself badly I am afraid but this what I intended
Another point is that of the layout
Let's take a phono preamp, high gains small signals
With discretes I believe that lay out issues are more critical
With smd construction and opamps maybe you can get extremely short signal paths so less risk of RF noise ?
Good evening
gino
What I intended is something like this
http://www.chinaicmart.com/uploadfile/ic-circuit/20097663926987.gif
Sometimes I see signal bjts/fets I think of a differential pair put one against the other closely
More closely than in the same chip I think is difficult
I expressed myself badly I am afraid but this what I intended
Another point is that of the layout
Let's take a phono preamp, high gains small signals
With discretes I believe that lay out issues are more critical
With smd construction and opamps maybe you can get extremely short signal paths so less risk of RF noise ?
Good evening
gino
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.