transistors are semiconductors
once you have learned how to deal with transistors
and put them into circuits
then designing with transistors is much more flexible
there are 1000 of different types of transistors
with all possible and impossible characteristics
operational amplifier and other integrated circuits, IC,
are a bunch of semiconductors, transistors put together in a fixed way
with usually only small adjustments possible from the outside
from the IC PINS
there are however some general purpose IC
that have features that makes them very useful in most any kind of applications
and there are some Audio IC that are special purpose
and with some very good data
that this makes them outperform most any diy transistor circuit
transistors are a tiny bit better performers, in 90% of all situations,
when put together by the hands of a circuit designer with good knowledge
Finally:
99% of all IC can not do without some transistors inside
99% of all transistors can do without IC
typical op-amps can house 30-50 transistors, or even 100
lineup has become a transistor man - was an op-amp man in beginning
once you have learned how to deal with transistors
and put them into circuits
then designing with transistors is much more flexible
there are 1000 of different types of transistors
with all possible and impossible characteristics
operational amplifier and other integrated circuits, IC,
are a bunch of semiconductors, transistors put together in a fixed way
with usually only small adjustments possible from the outside
from the IC PINS
there are however some general purpose IC
that have features that makes them very useful in most any kind of applications
and there are some Audio IC that are special purpose
and with some very good data
that this makes them outperform most any diy transistor circuit
transistors are a tiny bit better performers, in 90% of all situations,
when put together by the hands of a circuit designer with good knowledge
Finally:
99% of all IC can not do without some transistors inside
99% of all transistors can do without IC
typical op-amps can house 30-50 transistors, or even 100
lineup
has become a transistor man - was an op-amp man in beginning
The simple answer is yes and nojsa_ind said:
You have to know a part in order understand what it do for you. In IC's you can't make certain parts like good resistors, big caps and certainly not inductors but what you _can_ do is super matched transistors at speeds of 3 GHz or more for the cost of almost nothing.
Bgt said:LM4562 is getting very close
LM4562 is a good operational amplifier IC.
Handled properly it may do good audio.
Now, very (... read excessive fast for Audio ) fast Op-Amps can have impressive data,
but some extra care in PCB layout
and other things are often required to avoid oscillations (improve stability in some real loads).
LM4562, and the likes of it, are no challengers to
for example one of my current discrete pre amplifier projects:
+-20 Volt RMS output, using discrete regulated +-75V power supply.
At highest HiFi quality ( ... read 0.005% THD distortion )
With the specified limit of max +-17V supply
LM4562 is very far from it,
it doesn't even come close to my discrete ..
And my discrete amp uses standard low price TO92 higher voltage transistors.
By the way, I do not use, if can be avoided in low power amplifiers
at any time Class B or Class AB designs.
not when it is so easy to setup a True Class A for some mA output at line level.
*****************************************************
I don't know, but ..
what does such a limited wonder,
almost handicapped when comes to compare voltage supply
as LM4562 cost?
Not that I have any plans or needs to buy any Op-Amp.
Not for the moment.
But would interesting for comparing to the cost of
my 150 Volt HiFi amplifier.
lineup Lineup Audio Discrete Class A Designs
http://lineup.awardspace.com/
Ok, this is getting really interesting... I'm starting a group buy of the LM4562 opamps for my NewClassD amps and some active crossovers, which are all made for DIP chips. However, I've got a new CD player brewing in the back of my mind, which obviously has no limits in that respect.
Being relatively inexperienced with circuit design and PCB layout, but very "fiddly" and up for a challenge, what are your tips? Concrete examples are much appreciated.
Edit: I'll most likely be using TI's PCM1794 DACs in dual mono for my CD player.
Being relatively inexperienced with circuit design and PCB layout, but very "fiddly" and up for a challenge, what are your tips? Concrete examples are much appreciated.
Edit: I'll most likely be using TI's PCM1794 DACs in dual mono for my CD player.
Hi lineup,
Why do you need +/-75V rails for your preamp and 20V RMS output? Most commercial, high-end, and even tube amps can be driven to full output with just 2V RMS of input.
When in the 2V RMS world, one would seem very hard-pressed to beat the LM4562 op-amp for noise, distortion, low Z-out, etc. And, at <$3 each, wow, what a performer.
As far as class A vs. class A/B for signal level applications, why would one think the LM4562, or virtually any op-amp, has cross-over distortion or issues with output drive integrity? With distortion and noise being so exceptionally low, I just can't imagine worrying about the output class. If there were improvements to be made, I'm pretty sure their engineers and scientists would be all over it.
Other advantages to monolithic ICs over discrete designs include near-flawless device matching, laser trimming, minimal inter-capacitance and trace inductance, perfect temperature tracking, and the ability to fab very unusual current sources and multi-function circuits/devices.
I have no problems whatsoever with opamps and feel they are near perfect devices (when implemented properly and matched to their intended duties). Naturally, I fully understand op-amps cannot perform at high voltages, high currents, high powers, etc. Plus, most likely 99% of all recorded music you've ever heard was run through op-amps while recording, mixing, mastering, or duplicating.
Why do you need +/-75V rails for your preamp and 20V RMS output? Most commercial, high-end, and even tube amps can be driven to full output with just 2V RMS of input.
When in the 2V RMS world, one would seem very hard-pressed to beat the LM4562 op-amp for noise, distortion, low Z-out, etc. And, at <$3 each, wow, what a performer.
As far as class A vs. class A/B for signal level applications, why would one think the LM4562, or virtually any op-amp, has cross-over distortion or issues with output drive integrity? With distortion and noise being so exceptionally low, I just can't imagine worrying about the output class. If there were improvements to be made, I'm pretty sure their engineers and scientists would be all over it.
Other advantages to monolithic ICs over discrete designs include near-flawless device matching, laser trimming, minimal inter-capacitance and trace inductance, perfect temperature tracking, and the ability to fab very unusual current sources and multi-function circuits/devices.
I have no problems whatsoever with opamps and feel they are near perfect devices (when implemented properly and matched to their intended duties). Naturally, I fully understand op-amps cannot perform at high voltages, high currents, high powers, etc. Plus, most likely 99% of all recorded music you've ever heard was run through op-amps while recording, mixing, mastering, or duplicating.
DCPreamp
I thuoght your name, DCPreamp was more like Discrete Pre Amplifiers
.. but we can never know, can we
Why I can't use most any Op-Amp
and anything but my own discrete transistor +-75 V DC regulators for my idea and circuit?
Because sometimes, you do experiments and try to see
... if for example transistor voltage amplifiers will benefit from tube like voltage levels as supply.
It has not been done much research in this field.
I know only of a handful of pre-amplifiers with very high voltage operatin.
Besides, to design a 20 V RMS stand alone voltage amplifier
you can definitely not use +-18-20 volt Op-amps, = max 40-45 total voltage.
...
Because, for as anybody know:
20 V RMS requires: +28.3 Volt and -28.3 Volt
and for transistor to work linear, it is not at all bad in my very special amplifier, to add a few voltages
to avoid beginning clipping behavior and also improve other parameters.
This is why the extra added +46.7 and -46.7 volt,
making the supply in total be 150 volt.
It is a tube like transistor voltage amplifier, if you like.
lineup has got good reasons, for stay Discrete... this time
PS.
Apex high voltage Op-Amps are too expensive for this experiment
as well as I can not buy even op-amps with 50% of needed volt supply
at least not for normal money ( like I used normal standard 150V and 300 volt small transistors )
and not in my country.
I do never buy from abroad. Not from Other countries.
DS.
I thuoght your name, DCPreamp was more like Discrete Pre Amplifiers
.. but we can never know, can we
Why I can't use most any Op-Amp
and anything but my own discrete transistor +-75 V DC regulators for my idea and circuit?
Because sometimes, you do experiments and try to see
... if for example transistor voltage amplifiers will benefit from tube like voltage levels as supply.
It has not been done much research in this field.
I know only of a handful of pre-amplifiers with very high voltage operatin.
Besides, to design a 20 V RMS stand alone voltage amplifier
you can definitely not use +-18-20 volt Op-amps, = max 40-45 total voltage.
...
Because, for as anybody know:
20 V RMS requires: +28.3 Volt and -28.3 Volt
and for transistor to work linear, it is not at all bad in my very special amplifier, to add a few voltages
to avoid beginning clipping behavior and also improve other parameters.
This is why the extra added +46.7 and -46.7 volt,
making the supply in total be 150 volt.
It is a tube like transistor voltage amplifier, if you like.
lineup
has got good reasons, for stay Discrete... this timePS.
Apex high voltage Op-Amps are too expensive for this experiment
as well as I can not buy even op-amps with 50% of needed volt supply
at least not for normal money ( like I used normal standard 150V and 300 volt small transistors )
and not in my country.
I do never buy from abroad. Not from Other countries.
DS.
Ouch, Lineup, stop hitting us in the eyes!!
My addition would be that there is such a thing as headroom, explained before. The volume control is usually at the end of a pre-amp (perhaps only followed by a unity-gain buffer).
What happens before it? Particularly with the wide variation in CD outputs as well as CDs (at least in my experience), I have encountered levels up to 20dB above "normal" (whatever that might mean). If I cater for an output of 2,8Vp, that means a possible 28Vp before the volume control - plus some extra. We then arrive at total 70V odd rail voltage.
I do not think the cost is really a factor, neither is drift. I am not a dc coupled fan - more problems than advantages (if any). And lowest noise (pnp) transistors do have a lower noise factor than i.c.s I know - result of the larger number of manufacturing cycles involved. Then I also do not like the early phase correction in many i.c.s (also as said before) but that may be subjective - I have encountered the odd problem with h.f. overload there. Furthermore I put high and low steep-cut filters in my pre-amps (reason and design for another day), where the +/- 18V lot can definitely hit the rails.
In the end, I find lower distortion (particularly high order) with discrete transistor design under worst conditions, mainly because of the rail voltages. But all this may be a little subjective. One certainly acknowledges the advantages of i.c.s. It depends on the application.
Regards.
Edit: Spelling!
My addition would be that there is such a thing as headroom, explained before. The volume control is usually at the end of a pre-amp (perhaps only followed by a unity-gain buffer).
What happens before it? Particularly with the wide variation in CD outputs as well as CDs (at least in my experience), I have encountered levels up to 20dB above "normal" (whatever that might mean). If I cater for an output of 2,8Vp, that means a possible 28Vp before the volume control - plus some extra. We then arrive at total 70V odd rail voltage.
I do not think the cost is really a factor, neither is drift. I am not a dc coupled fan - more problems than advantages (if any). And lowest noise (pnp) transistors do have a lower noise factor than i.c.s I know - result of the larger number of manufacturing cycles involved. Then I also do not like the early phase correction in many i.c.s (also as said before) but that may be subjective - I have encountered the odd problem with h.f. overload there. Furthermore I put high and low steep-cut filters in my pre-amps (reason and design for another day), where the +/- 18V lot can definitely hit the rails.
In the end, I find lower distortion (particularly high order) with discrete transistor design under worst conditions, mainly because of the rail voltages. But all this may be a little subjective. One certainly acknowledges the advantages of i.c.s. It depends on the application.
Regards.
Edit: Spelling!
My friends here at diyaudio never 'get hit' by my posts.Johan Potgieter said:Ouch, Lineup, stop hitting us in the eyes!!
If you feel you get hit, it tells more about you, Johan, and your thinking, thoughts in case.
than it does tell anything about my posts.
Nicht waar?
Have a closer look at yourself, from a distance .... and you might see a bit further than your own nose-tip
------------------------------------------------------------------------------------
By the way, lineup friends or not,
take a look into this interesting TOPIC!!!!! we have goin':
started by suzyj
www.littlefishbicycles.com
diy audio projects division
Discrete Guitar Preamp - by Suzy J.
Several interesting discrete JFET circuits,
already in the above little thread!
Regards
to anyone that are willing to take my regards
and all the other ones
lineup
December 2006
i agree that op amps have less noise, but a well designed discrete circuit doesn't have any noticeable noise either. anyway, to me a well-designed discrete circuit sounds better, and i've rolled a lot of op amps. they all sound different from each other, so i couldn't imagine how they don't degrade the sound. to me the difference between various op amps even within the same manufacterer is like night and day. people with half-decent ears (like myself) should do a blind test comparing a cascade of op amps to straight wire. that will effectively demonstrate that op amps suck.
lineup said:
obviously you didn't follow up on my eariler post in this thread http://www.diyaudio.com/forums/showthread.php?postid=1041001#post1041001
the described amp uses a "cascaded" op amp output stage - 2 +/16 V rated op amps run from gnd and +/-30 V respectively, lifting the +/- supply rails of the output op amp for very close to 20 Vrms swing - essentially doubling the V swing capability of the op amps used, in principle 2 of these could then be used with a bridge tied or balanced load for another 2x swing
lineup said:
I don't detect any signs of humour in your post, nor anything but emotion in a technical thread -
So, come on, please Lineup!!
If you paid any attention at all to my previous contributions, you would by now have seen that I often use humour! My remark was simply about your using an increasing script size in order to, "sound louder" - or what? ... and my reply quite light-hearted! Is your own perception such that you required twenty little smileys after that in order to see it for what it was??
Your cryptic CV leaves me at a disadvantage, so let me just say that I have quietly been round the block of both science and life a few times. I certainly still have many shortcomings, but did not deserve your rude personal closing remark above, so please. And if I am not your friend, you qualified that.
Now can we carry on with science while recognising humour where it is intended (especially when followed by an emoticon). You might just possibly have noticed that my design approach agreed with yours.
Regards to all (including Lineup).
Hi Folks,
DCPreamp = battery powered for ultra-low/zero PS noise. I also run AC coupled as I too don't like DC coupling and don't have irrational fears of caps when they are in their low Xc state.
lineup: I understand using HV for tubes and trying to simulate a tube-transfer function with semiconductors. I'm not questioning that. Tubes do require high supply voltages because of their nature, but semiconductors do not. So I don't think running semiconductors at high voltages automatically means they will sound like tubes, but I digress.
Johan P.: I also understand having plenty of headroom and the requirements of adequate supply rails to allow for it.
However, if you run a 250WPC power amp with 8 ohm speakers, the amp must have approximately 90VDC rails to accomplish the output of 45VRMS. And if your amp has approximately 28dB of gain, it only requires 1.8VRMS to drive the amp to full power output. So now, you drive 20VRMS into the input of the amp with your "high-headroom" preamp, do you think the amp magically produces a 1,000V peak output? Nope - it just severly clips. There's no mystery here - just amp rails and fixed gain. Pretty straightforeword.
But, with the same amp running at a casual 1 or 2 watts listening level, and only requiring 160mV of preamp signal, it can reproduce those 20dB peaks. So again, why does your preamp need to be able to produce 20VRMS? It quickly starts to stink of subjectivism where is "just sounds better and is suposed to be that way" with no scientific reasoning.
Paul
DCPreamp = battery powered for ultra-low/zero PS noise. I also run AC coupled as I too don't like DC coupling and don't have irrational fears of caps when they are in their low Xc state.
lineup: I understand using HV for tubes and trying to simulate a tube-transfer function with semiconductors. I'm not questioning that. Tubes do require high supply voltages because of their nature, but semiconductors do not. So I don't think running semiconductors at high voltages automatically means they will sound like tubes, but I digress.
Johan P.: I also understand having plenty of headroom and the requirements of adequate supply rails to allow for it.
However, if you run a 250WPC power amp with 8 ohm speakers, the amp must have approximately 90VDC rails to accomplish the output of 45VRMS. And if your amp has approximately 28dB of gain, it only requires 1.8VRMS to drive the amp to full power output. So now, you drive 20VRMS into the input of the amp with your "high-headroom" preamp, do you think the amp magically produces a 1,000V peak output? Nope - it just severly clips. There's no mystery here - just amp rails and fixed gain. Pretty straightforeword.
But, with the same amp running at a casual 1 or 2 watts listening level, and only requiring 160mV of preamp signal, it can reproduce those 20dB peaks. So again, why does your preamp need to be able to produce 20VRMS? It quickly starts to stink of subjectivism where is "just sounds better and is suposed to be that way" with no scientific reasoning.
Paul
Hi Paul,
No, sorry - I think there is a misunderstanding. I did say that I was referring to before the volume control, which is usually somewhere near the end of the pre-amp. Thus most of the pre-amp is always subject to "maximum" signal level, if you understand. Obviously as you state you will "output" (from the volume control) no more than the required signal level to clip the main amplifier; your ears will tell you that that point has been reached, so.
The full signal is always present before the volume control, except in rare cases where there is also a ganged pot somewhere earlier in the circuit (thus 4 pots on a spindle for stereo). Thus, you do not drive your power amp with that "20Vrms" - that is only the safety factor for before the volume control. (In fact, I hate these amplifiers with too much gain. Folks mostly seem to think that the amplifier must be powerful if it roars as soon as one touches the volume control - never mind that it runs out of steam a little further up the knob scale. Cheating.) But as I encountered there can be such a variation in level there from different input devices/signal sources, that overdrive can occur - you seem to understand that, you said. Other designers (e.g. Douglas Self, Cherry, Linsley-Hood et al) have stated that for safety they prefer somewhere above 18dB headroom. In my view, once one goes for higher d.c., one might as well go quite safe as higher voltage transistors and transformers do not cost that much more. When you measure distortion (and I mean a full high order analysis/intermod, not the outmoded thd), it does appear to help to stay away from the rails. It is one of the reasons why tube pre-amps are quite "clean", what with usually over 150V of signal swing being available.
Your first paragraph is acknowledged (I was not picking on you because of your pseudonym, just a general comment!). I further agree with your last sentence - there is a lot of "it must sound better because [urban legend]" - more's the pity. I would certainly hope that I am never guilty of that line of thought! But some things call for a judgement call (oops, poor grammar) - people will make different decisions. I have certainly also used op-amps, depending on the application.
Regards
No, sorry - I think there is a misunderstanding. I did say that I was referring to before the volume control, which is usually somewhere near the end of the pre-amp. Thus most of the pre-amp is always subject to "maximum" signal level, if you understand. Obviously as you state you will "output" (from the volume control) no more than the required signal level to clip the main amplifier; your ears will tell you that that point has been reached, so.
The full signal is always present before the volume control, except in rare cases where there is also a ganged pot somewhere earlier in the circuit (thus 4 pots on a spindle for stereo). Thus, you do not drive your power amp with that "20Vrms" - that is only the safety factor for before the volume control. (In fact, I hate these amplifiers with too much gain. Folks mostly seem to think that the amplifier must be powerful if it roars as soon as one touches the volume control - never mind that it runs out of steam a little further up the knob scale. Cheating.) But as I encountered there can be such a variation in level there from different input devices/signal sources, that overdrive can occur - you seem to understand that, you said. Other designers (e.g. Douglas Self, Cherry, Linsley-Hood et al) have stated that for safety they prefer somewhere above 18dB headroom. In my view, once one goes for higher d.c., one might as well go quite safe as higher voltage transistors and transformers do not cost that much more. When you measure distortion (and I mean a full high order analysis/intermod, not the outmoded thd), it does appear to help to stay away from the rails. It is one of the reasons why tube pre-amps are quite "clean", what with usually over 150V of signal swing being available.
Your first paragraph is acknowledged (I was not picking on you because of your pseudonym, just a general comment!). I further agree with your last sentence - there is a lot of "it must sound better because [urban legend]" - more's the pity. I would certainly hope that I am never guilty of that line of thought! But some things call for a judgement call (oops, poor grammar) - people will make different decisions. I have certainly also used op-amps, depending on the application.
Regards
Hi Johan,
You had mentioned the pot after the amp, so I missed it and blew that part. I guess I'm not super familiar with that topology, either from a lack of experience, or lack of "true" audiophile knowledge.
I typically use the signal-pot-amp path. I can see some benefits to the signal-amp-pot-buffer idea, but it's double the active circuity. I've seen preamp designs with the pot on the output with no buffer, but with a hi-Z-out, one might as well go passive.
I do agree dicrete can be outstanding if not near flawless when done well. Believe me, I have tons of respect for many, many topologies so long as they don't cross into the subjective realm of woowoo BS and stay in-line with reality. You are definately right there based in reality, and I appreciate your details.
So back to the original "are discrete better than op-amps" question, seems a double-blind showdown between topologies would be would be loads of fun. Some day perhaps. 'Til then, I have an LM4562 I'm just dying "not" to hear. ; )
You had mentioned the pot after the amp, so I missed it and blew that part. I guess I'm not super familiar with that topology, either from a lack of experience, or lack of "true" audiophile knowledge.
I typically use the signal-pot-amp path. I can see some benefits to the signal-amp-pot-buffer idea, but it's double the active circuity. I've seen preamp designs with the pot on the output with no buffer, but with a hi-Z-out, one might as well go passive.
I do agree dicrete can be outstanding if not near flawless when done well. Believe me, I have tons of respect for many, many topologies so long as they don't cross into the subjective realm of woowoo BS and stay in-line with reality. You are definately right there based in reality, and I appreciate your details.
So back to the original "are discrete better than op-amps" question, seems a double-blind showdown between topologies would be would be loads of fun. Some day perhaps. 'Til then, I have an LM4562 I'm just dying "not" to hear. ; )
AH! There it is then Paul.
You would have seen that one often uses the volume control as late as possible, so that any pre-amp noise gets attenuated with the signal at low settings, etc. The hi-Z out could be tolerated, depending on the power amp input conditions.
In one of my more extensive pre-amps I use all of 10 transistors up to the volume control (2 in unity gain buffer after). Rail voltage is 70V.
On the original question I quickly checked whether I was "safe"! I saw in my post #13 that I did mention the right application. But see what you have done now. I will go right back to my data books and make a proper study of that LM4562, and get hold of one when nobody is looking .......
Thanks for the correspondence Go for it!
You would have seen that one often uses the volume control as late as possible, so that any pre-amp noise gets attenuated with the signal at low settings, etc. The hi-Z out could be tolerated, depending on the power amp input conditions.
In one of my more extensive pre-amps I use all of 10 transistors up to the volume control (2 in unity gain buffer after). Rail voltage is 70V.
On the original question I quickly checked whether I was "safe"! I saw in my post #13 that I did mention the right application. But see what you have done now. I will go right back to my data books and make a proper study of that LM4562, and get hold of one when nobody is looking .......
Thanks for the correspondence Go for it!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.