Are Discrete components better than Opamps ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello Again Johan,

Feel free to check out: http://www.national.com/pf/LM/LM4562.html

The LM4562 a very impressive device, but there are many impressive op-amps out there and when properly implemented, following very close attention to detail, and utilizing excellent components and topology, I'll bet many op-amps will produce fabulous audio results. My current OPA2604 preamp sounds very good to me.

Let me ask: When you design your own circuit, or copy one, build it, fully measure and implement it, how do you know its true performance? Do you have the facilities for testing IMD, THD, noise floor, etc.? If so, then you can truly have confidence in your designs and have the comfort of using top-notch gear.

Unfortunately, I do not have those tests at my disposal and am limited to bandwidth, noise, Z-in/out, VRMS, and a few others limited by my test equipment and my scrutiny.

So, what I do is use the best parts (within cost and "reality" reasons), use proven designs (typically from data-books and industry related books and journals), use solid and quality construction methods, and ultimately, by listening. Other than that, I could be listening to 5% THD for all I know (I doubt it, though). By following what has been presented by Engineers and Scientists at National Semiconductor (if Bob Pease is an indicator of their skill levels, I feel pretty good about their stuff), Burr-Brown, Linear Technology, and others, I'm pretty sure I'm producing some very good results.

What I also know, is that I have tested and listened to several ADCOM preamps, high-end tube preamps, and my own designs, and mine are at least an order-of-magnitude better on the bench as well as in listening tests. It's enough for me to make some bold claims about my gear and give me pretty good confidence, but I still have a hint of doubt about the ultimate performance. Many would say listening tests are all that is required for DIY audio, as a devout non-subjectivist, I always want to know everything that's going on within the design.

Anyway, I'd enjoy reading your theories about producing a good design and how you test your products.

Regards,
Paul
 
Administrator
Joined 2004
Paid Member
I have never understood the massive gain or head room approach in preamps. Certainly have owned and heard preamps that had massive headroom, and they sounded great, but how much of that was due to headroom? I don't know.

Johan was talking about signals that went up to 20dB above average. That's pretty typical on CDs, depending on the type of music. But it might be more useful to see it from the other side. The average signal is 20dB below peak. And what is that peak? About 2V rms or 2.8 peak. That's all you get out of most DACs, about 3V peak. Many have lower values.

So no need to worry about the average music level, if the absolute max is 3V. I would think that +/- 15V rails would be plenty, if you don't need much gain. Say less than 12dB.

Analog sources are not so well defined, but you should be able to figure out the approximate max levels.

As Mr. DCPreamp said, most power amps can run into clipping a 2 volts RMS - so how much line level gain does your preamp really need? 6~9 dB would seem to be more than enough to make up for low signal sources. Meaning 8.5 volts peak output. Or a little more than 1/2 rail voltage. I can't see where you'd ever need 8 volts to drive your amp, unless it's some strange high power/low gain thing.

Gain structure is of huge importance in pro audio where there are so many gain stages, often of high gain. Getting gain structure right in a pro rig can make a big difference to the overall result.

Just my thoughts....
 
panomaniac said:

Headroom Voltage.
----
So no need to worry about the average music level,
if the absolute max is 3V.
I would think that +/- 15V rails would be plenty,
if you don't need much gain.
Say less than 12dB.


Headroom voltage is the reason for my high +/- 75 V DC regulated supply.
For my discrete tube-like & transistor based
voltage amplifier with +/- 28.3 V minimum specified output.
(see my previous posts in this topic!)


This is a headroom of only ~ 3:1
And at same time, consider that my amplifier can be used for no global feedback,
or by choice very low feedback factor.
All in true Class A operation and bias.
-----------------------


Compare this to Operational Amplifiers.
Like panomaniac said:
So no need to worry about the average music level,
if the absolute max is +/- 3V peak.
I would think that +/- 15V rails would be plenty


This is a headroom of ~ 5:1
This is considerably more than my amplifier.
Especially when we consider op-amp has a high global negative feedback factor!
And having an open loop gain of 10.000, or more ....
As most Op-amps have.
This while running at low Class AB bias.



Regards to panomaniac
from
lineup
 
Administrator
Joined 2004
Paid Member
Hej Lineup.

I see why you need such high voltage rails if you want +/- 28V out. But why such a high output voltage for a preamp? I run power amps with a lower output voltage than that. :)

Did I miss something? I have run tube pres that were capable of over 100V out, clean. But I doubt I ever got anywhere near that - nowhere close.

BTW, some opamps and even VCAs can be run in Class-A, if it helps.
 
panomaniac said:
Hej Lineup.

I see why you need such high voltage rails if you want +/- 28V out.
But why such a high output voltage for a preamp?
I run power amps with a lower output voltage than that. :)


Very right you are again, panomaniac.

If you read my posts carefully,
you see I call not my amplifier for 'preamp'
but 'standalone voltage amplifier'.
If I said anything else, it would not be 100% correct.

The idea is this:
.. as is not very hard to find good power amplifier buffers = voltage followers for loudspeaker loads,
we would need some good voltage amplifiers for those.
A separate input and VAS stage, if you like. (VAS= Voltage Amplification Stage)

Such a voltage amplifier would have output
into a Power Follower buffer, VoltageGain ~ 0.90-0.99
This is why I decided to specication:
Min 20 V RMS into something like >=2kOhm
with very low distortion.


Now put 1 + 1 together!
And you will find you have a very good power amplifier.
Although in two separate stand-alone units.


Regards
lineup

==============================================
* fotnote:
Even myself, lineup, have published at forum such a power buffer,
in True Single End Class A, giving out 15-23 W RMS in whole span 4-8 Ohm,
with very low distortion.

I even have attached, along with my schematics and description,
test data from my simulation into different impedances
and I can guarantee stability even at normal capacitive loads!

How many transistors?
For this 20 W RMS(5 Ohm) follower?
+ two MJ21194
+ one BD139, TO126
+ one 2SA970, TO92 for low noise input.
Makes it 4T circuit, only.
 
Administrator
Joined 2004
Paid Member
Ah, so...

I thought you might be going in that direction. You are rolling the preamp and power amp voltage gain stage into one circuit, so to speak. Then putting it in one box and the current output stage in another box. Sure, why not? As long as the stages couple well.

Can you send me the files for you SE Class-A power buffer, or point me to the post with them? I'd like to see what you've done.

Thanks!
 
DCpreamp (your post #41)

Yes, guess I am overlooking the priviledge of facilities at my disposal. At work all those were available, but lately only for final checking, courtesy of the local university. My main design tool has become Spice, which I have experienced to get pretty close to reality. What I look at are high-order harmonics (or rather the absence of them). One can do worst case design and find where they start to emerge, and try for a happy medium (THD hardly of value there). This is of course more important in power amplifier design. I also try to keep the phase angle low over the whole audio range. [This is a kind of secondary thing and a subject on its own. The gist is the poor behaviour of amplifiers in out-of-sonic ranges where, should anything come along, artifacts can be generated with audible effects. Filter sharply either side of the audio range, and keep your amp in order where there might still be stuff, etc.]

The problem with listening tests is to what and for how long. You would know about something called listener fatigue, sometimes only established after days. The probable source of that is immediately recognisable in e.g. spectrum analyser tests. But as said, not everybody has that priviledge. In that case, tried designs by known designers, absolutely!

Panomaniac, your post #42,

Taking the inverse logic, quite true. But my CD has an average output of about 1,3V. It has an amplifier after the DAC - which general practice, incidentally, I have seen several comments on, with ideas for improvement. Apparently some CD designers are not so good in the analogue area.

You bring up the point: What input sensitivity? My line-ins are 0,5V, and there are also still tape- and phono-ins (the need for which, I accept, will diminish "as we speak"). But also the difference in CD signals (I mean the discs) can be quite dramatic. Then back to what must be tolerated before the volume control, etc. I do, in fact, provide measures for rough adjustment of inputs at the pre-amp input, so that folks can at least set source outputs at the loudspeaker more or less equal.

I would re-iterate that I do not condemn op-amps per se, just prefer discretes in certain cases. Also stating that my power amp maximum input is a high 600mVrms, will further clarify.

Regards
 
Johan Potgieter said:

-------------------
I would re-iterate that I do not condemn op-amps per se, just prefer discretes in certain cases.
Also stating that my power amp
maximum input is a high 600mVrms,
will further clarify.


600mVrms
-------------------------------------
This is to me a good value.
A value that I have found we can count on,
that output level from
an Average CD-recording, or most CD, will live up to.

Of course can be very different for different kinds of music.
Classic music and Rave, Rap music has got different contents of amplitude.
The relationship between Peak Music outputs and Average output level, I mean.

But sure is, we can often not design and use Max 2.0 V RMS, the CD standard
as input sensitivity
at least for normal power amplifier and if we want to play LOUD.
Not even count for 50% 1.0 V RMS will be good always.
-----------------------------------


A target value for input sensitivity I often have used
when make an initial specification of my HeadPhones Amplifiers
is input sensitivity (input voltage level required for close to maximum output, before clipping)
of 0.707 Volt RMS ... which give vPeak input +/-1.0 Volt.

I have got this value from measuring sound level output from some of my typical records.
Typical for the music I listen to most often.


lineup
 
Simple SE Class-A power buffer by Lineup

sorry
my friend,
panomaniac

I missed your post asking.
( you know I involve myself in so many many topics.
3.600 posts since 2002
once i have said what I think I can say ina subject
i move on to something else
-- that's the way i am :) a bit restless perhaps)


Of course I post the link here.
To me and my other friend's topic.
This topic was started by
a very fast learning! beginner in your own amplifier designs!
Like me he is using Simulation software to check up what looks good.
Before he actually build.

I took his idea - found it interesting!
And we discussed and improved it to very good data.
Very good data, if you consider the simplicity of circuit.

As told it is SINGLE Supply 100% Class A (both mine and jerluwoo amplifiers)
And though simple, it is using two NPN Power Transistors in PUSH-PULL
for better efficiency (=a bit less heatsinking).


Here is jerluwoo topic
diyAudio Forums > Top >Amplifiers >Solid State > My simple class a approach


Be sure to have a look at jerluwoo later version.
He did post several. One after one better than the other
until they were very good Class A power amplifiers.

Hard to say if my posted design is better than his.
They are both okay, I would say.


panomaniac,
can you post a response post in that topic, Please.
Say something, to comment.
This is all I ask from you
when I give you my schematic I attached there.
;) I think jerluwoo will be happy see someone read his posting;)


Many Regards
and thanks for showing interest in some of my works
lineup
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.