Hello Everyone,
Is it true that discrete components would sound better then IC's ?
I mean there is so much of advancement in IC's I thought they would at least would be as good as discretes...am I mistaken to have such a perception ?
Certainly http://www.bursonaudio.com/Why_Burson_HDAM.htm thinks otherwise !
Thanks,
Regards,
Junia.
Is it true that discrete components would sound better then IC's ?
I mean there is so much of advancement in IC's I thought they would at least would be as good as discretes...am I mistaken to have such a perception ?
Certainly http://www.bursonaudio.com/Why_Burson_HDAM.htm thinks otherwise !
Thanks,
Regards,
Junia.
There are certain type of components that have no IC equivalent. Transformers, for example. It is hard to duplicate or exceed transformer performance with op-amps, and of course, op-amps require power and other components to function.
IC's are good at making circuits that are flexible, and consistent. The matching of transistors in an IC can be very nearly perfect which can lead to better performance than discrete parts which can't be matched as well. The manufacturing techniques used in IC yield parts that behave pretty consistently from one device to the next.
OTOH, you can make a class A power amp with an IC output stage.
Both discrete and ICs have their uses.
I_F
IC's are good at making circuits that are flexible, and consistent. The matching of transistors in an IC can be very nearly perfect which can lead to better performance than discrete parts which can't be matched as well. The manufacturing techniques used in IC yield parts that behave pretty consistently from one device to the next.
OTOH, you can make a class A power amp with an IC output stage.
Both discrete and ICs have their uses.
I_F
I_Forgot said:
OTOH, you can make a class A power amp with an IC output stage.
I_F
That was supposed to say "CAN'T".
I_F
Hi,
even for line drivers, ICs are at their limits. Anything more than a few mA and they are into ClassAB.
600ohm and 75ohm driven to 20dbu is hard going and to 24dbu is nearly impossible for a single IC.
That's why composite topology is being used but then the cost doubles.
Discretes can overcome these current/temperature effects with careful design.
even for line drivers, ICs are at their limits. Anything more than a few mA and they are into ClassAB.
600ohm and 75ohm driven to 20dbu is hard going and to 24dbu is nearly impossible for a single IC.
That's why composite topology is being used but then the cost doubles.
Discretes can overcome these current/temperature effects with careful design.
If there is not much output current demand like higher- order active filters then you must be very good in analog design to beat good op-amps.
Someone once claimed on this forum that he wasn't able to her the difference between a straight wire and a chain of op-amps used as voltage followers (which is even the worst usage of an op-amp anyway).
Regards
Charles
Someone once claimed on this forum that he wasn't able to her the difference between a straight wire and a chain of op-amps used as voltage followers (which is even the worst usage of an op-amp anyway).
Regards
Charles
AndrewT said:Hi,
even for line drivers, ICs are at their limits. Anything more than a few mA and they are into ClassAB.
600ohm and 75ohm driven to 20dbu is hard going and to 24dbu is nearly impossible for a single IC.
That's why composite topology is being used but then the cost doubles.
Discretes can overcome these current/temperature effects with careful design.
acutually with today's hotter dsl driver op amps a few 10's of mA of Class A current is available from a few parts without external biasing
and you can of course bias pairs of op amp outputs against each other for push-pull Class A up to the power/current limit of the packages
[6x TPA6120 seen edge on, 440mA Class A for real op amp output stage]
My prototype headphone amp is biased to give 440 mA peak in push-pull Class A with TPA6120 output op amps ~ 5 Wrms into 50 Ohms - with nothing but op amps!
http://www6.head-fi.org/forums/showthread.php?t=190991
scroll down to post #6, #13+ for explanation, pics, sims
some newer dsl driver op amps are faster and handle as much current as many of the classic audio driver transistors although op amp V is pretty limited - common source/emitter output stage could be driven by these op amps to just about any practical power level with the output Qs being the only discrete transistors
Any reply to this will need to be qualified. Economy, mass-production, consistent preformance, all go very much in favour of the right op-amp.
But I mostly use discreet transistors for the following reasons;
In pre-amps, the headroom with +/- 18V maximum rails is often not enough. Many such commercial ampllifiers go to 12 dB; some struggle up to 15dB headroom. This is often not always sufficient. I use 60 - 80V rail voltage on many occasions, to get some 26 - 30dB headroom. I also understand that the best discrete transistors can have lower noise than any i.c. (fewer manufacturing cycles).
Then the high impedance/current topology, especially around the Vas in power amplifiers, can lead to early Cdom influence cutting into the audio spectrum, resulting in unacceptable high-order harmonic presence. To get an open-loop response up to > 20KHz is well nigh impossible with power chips (caveat: I am not knowledgable with the newest ones). But that is required in order to get sufficient NFB stability/low phase angle necessary in excellent amplifiers. Also different compensation techniques than just a single Cdom is required, but the latter is often a given inside the i.c.
Regards
But I mostly use discreet transistors for the following reasons;
In pre-amps, the headroom with +/- 18V maximum rails is often not enough. Many such commercial ampllifiers go to 12 dB; some struggle up to 15dB headroom. This is often not always sufficient. I use 60 - 80V rail voltage on many occasions, to get some 26 - 30dB headroom. I also understand that the best discrete transistors can have lower noise than any i.c. (fewer manufacturing cycles).
Then the high impedance/current topology, especially around the Vas in power amplifiers, can lead to early Cdom influence cutting into the audio spectrum, resulting in unacceptable high-order harmonic presence. To get an open-loop response up to > 20KHz is well nigh impossible with power chips (caveat: I am not knowledgable with the newest ones). But that is required in order to get sufficient NFB stability/low phase angle necessary in excellent amplifiers. Also different compensation techniques than just a single Cdom is required, but the latter is often a given inside the i.c.
Regards
Getting back to the thread from all this blood (and before a moderator causes more to flow):
Nordic, I do not have a single universal circuit - that is part of the point. For a near unity-gain circuit I simply use the uni-gain complimentary pair, i.e. what is sometimes called a composite emitter follower. That could be given a gain of 1 - 4. Thus perfexctly stable. For gain circuits I in fact mimic a simple op-amp: Long-tailed input pair, Vas, followed by complimentary emitter-follower. Circuit values and compensation are chosen for the application. Open-loop bandwidth is never < 20 KHz, and 100 - 200V transistors are used. If that seems unnecessarily complex, the advantage is one of being free to chose operational parameters best suited to the application.
If you need a specific circuit I suggest you contact me privately; I do not want to hi-jack the thread. Just for completeness (since this thread is not about power amplifiers): There I also start with a long-tailed pair, though NFB not via that - straight to the input. The Vas is a totem-pole because of its linearity and large collector swing capability, with some refinements. This is followed by an emitter-follower (Vas output impedance is very high) driving a full-complimentary output pair. Again the advantage: I can tailor values for best response/phase characteristics/stability, etc.
Regards
Nordic, I do not have a single universal circuit - that is part of the point. For a near unity-gain circuit I simply use the uni-gain complimentary pair, i.e. what is sometimes called a composite emitter follower. That could be given a gain of 1 - 4. Thus perfexctly stable. For gain circuits I in fact mimic a simple op-amp: Long-tailed input pair, Vas, followed by complimentary emitter-follower. Circuit values and compensation are chosen for the application. Open-loop bandwidth is never < 20 KHz, and 100 - 200V transistors are used. If that seems unnecessarily complex, the advantage is one of being free to chose operational parameters best suited to the application.
If you need a specific circuit I suggest you contact me privately; I do not want to hi-jack the thread. Just for completeness (since this thread is not about power amplifiers): There I also start with a long-tailed pair, though NFB not via that - straight to the input. The Vas is a totem-pole because of its linearity and large collector swing capability, with some refinements. This is followed by an emitter-follower (Vas output impedance is very high) driving a full-complimentary output pair. Again the advantage: I can tailor values for best response/phase characteristics/stability, etc.
Regards
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Member Areas
- The Lounge
- Are Discrete components better than Opamps ?