Sonic differences - discrete vs. integrated?

[mlihl]

Hi fellow DIYers,

I have a somewhat silly question...Well, here it goes:
I'm thinking of building a Pass balanced line stage preamp
or even a preamp based on a Borbely circuit. But my
problem is before I actually jump into this costly adventure,
has anyone had the chance to actually compare a discrete
FET linestage to a well-designed preamp utilizing top-of-the-line
OPAMPs such as AD8610 or OPA627?

Last year I finished my battery-powered preamp with OPA627
/BUF634 project and PCM1704 DAC so I'm extremely curious
whether going all-discrete could be worthwhile? Does anyone
have experiences?

Looks like the DIY virus has seriously caught me...

Thanks for any input,

Mike

 

[GRollins]

There are numerous ways to argue this, but let me approach it this way. Supposing that you were, in some manner, able to arrange the absolute same circuit both as a chip opamp and as a discrete circuit. Exactly the same circuit. Same values. Same everything.
Except...
The components.
Quoth you,"Say what? Grey, you just said they're the same. Now you're saying they're different!"
Let's even go so far as to postulate that the transistors are exactly the same. For various reasons that's not really likely, but let's just say that they are, in fact, identical.
Quoth you,"Huh? Man, you're talking in circles."
Ah, but you neglected one thing. Take a 100 Ohm resistor. Get a nice, 1% low noise metal film Dale resistor. Now compare that to the little dinky etched zig-zag track in the chip.
Or worse yet, consider the difference between the capacitors in the chip and a decent mica or film cap.
As the man said,"The miracle isn't that the pig sings so well...it's that it sings at all!"

Grey

 

[tade]

but maybe sonsider the fact that all those little bits are etched onto the exact same crystal of ultrapure silicone, and also that because the origin of the circuit is a picture, every one will be the exact same. That and the fact that every component you have to match yourself is only complicating the whole venture. I would say opamp.

 

[m@]

not to bump this thread, but...

having recenlty started to make the transition from opamps (small ones, big ones) to discrete -- forget performance differences, digging out the old EE text and building from the transistor up is just a hell of a lot more fun.

 

[lineup]

Even if there are some pins into a chip
and you can do external tweaks of an OP-amp,
they do not come within a mile from the tweaks
you can do with a discrete circuit.

You can change resistor values, exchange transistors, alter currents
and reduce/increase compensation caps.
Increase voltage supply!
Make a true Class A output with more than 40 mA drive.
Decide open loop gain and change amount of feedback.

In short, it is like get your suit taylormade
or buy a massproduced suit in a warehouse.

OP-amps are good. They are reasonably flexible.
They can do a good job in most situations.
If you go for discrete, you wont be sure you can get better performance.
But you will get a performance that you have created.
And it will match your needs.


And building a discrete circuit is, in my opinion,
more interesting and fun DIY.
More fun than fit a few resistors around one IC.

 

[lineup]

I'm thinking of building a Pass balanced line stage preamp
or even a preamp based on a Borbely circuit.
..... has anyone had the chance to actually compare a discrete
FET linestage
to a well-designed preamp utilizing top-of-the-line
OPAMPs?
Mike

If you go into the more interesting DIY of discrete,
I think you should read this tutorial by Mr. Pass.
If you havent already?

Nelson Pass:
Simple high-performance op amps are easy to make out of as few as six discrete components whose total cost is less than a dollar. It is the purpose of this article to show how easy it is.

Besides the satisfaction of doing it yourself,
are there any other reasons to construct your own op amps?
Yes.
First you may want some special characteristic such as very high voltage, or ultra low noise, or high output current, that might be available commercially but at very high cost.

DIY Op Amps - by Nelson Pass 1998

You are free to use FETs or Bipolar or any combination of them.
These two figures are nice starting points to very good preamps!
Be sure to read Mr. Pass comments to these small circuits.
Figure 13
http://cygnus.ipal.org/mirror/www.passlabs.com/images/diyopf13.gif
Figure 14
http://cygnus.ipal.org/mirror/www.passlabs.com/images/diyopf14.gif

 

[tmblack]

It depends on how you obtain satisfaction and how you like to spend your time (and money).

A discrete design will destroy an old IC like NE5532 but the tables may be turning with state of the art IC's.

I have yet to see and hear a discrete with lower distortion than the OPA627.

Tom

EE

 

[lineup]

It depends on how you obtain satisfaction and how you like to spend your time (and money).

A discrete design will destroy an old IC like NE5532 but the tables may be turning with state of the art IC's.

I have yet to see and hear a discrete with lower distortion than the OPA627.

Tom

EE

This is true, Tom.

Same as what I meant by:
If you go for discrete, you wont be sure you can get better performance.
But you will get a performance that you have created.

 

[steenoe]

I have yet to see and hear a discrete with lower distortion than the OPA627.

I guess you would have a hard time, finding that I have built 2 preamps with the OPA627 and the BUF634. I even listened to them, to the surprise of some of you I also built the BoZ, the BosoZ and the CC-CCS-X-BosoZ.
I also listened to all of those Well, if we talk about Noise and nothing else, the CC-CCS-X-BosoZ takes the price cleanly! No doubt. This is not measured in a scope or something, its my ears put into the speakers And I really do that for real!! The X-boz is just clean, whatsoever Noisewise the OPA's come in after the X-boZ! The BoZ and the BosoZ, both have a bit of circuit noise. Call it "Mosfet hiss" I guess I guess we have the noise thing settled now? Well, when it comes to listening to music the X-BoZ is still a clean winner! No doubt about that. It just sounds that great! Funny thing is that both the basic BoZ and the BosoZ both are way better than the OPA's when it comes to listening to music! I cant remember one time when I thought about putting a OPA preamp back in! Whenever I do put the OPA in, I just want to put my BoZ back in Thats the reality for me, call me sentimental or whatever you fancy, but the mosfets win everytime!

Steen

 

[lineup]

If anybody see my post above, I recommend Nelson Pass article on DIY Op-amp.

I will not change my opinion that these basic amplifier circuits by Mr. Pass
will make a very good start of excellent custom designed discrete pre amplifiers.
No matter if you use FETs or BJTs.

His examples uses +-32 Volt supply.

An externally hosted image should be here but it was not working when we last tested it.

 

[steenoe]

They are beauty's, both of them I think the 2sk389 variety could be permanent in my livingroom Someone, design a board

Steen

 

[Zen Mod]

one important detail is not easy to achieve with OP;
that detail is simple-current .

good stage must have lots of Iq,or must be work of genius ,to sing without lots of I.


hehe-NS10,for instance;

just don't tell Nelson for this post,or he will never ever reply to any of my silly questions

 

[steenoe]

Come to think of it, my comparison of those preamps maybe wasnt that fair after all. I run the X-BoZ with balanced outputs, which might account for the lesser noise? All the others was with singleended output. And I even forgot the NS10, how could I

Steen

 

[MIKET]

Another intersting preamp

I have followed this link as well and may be a good choice for a driver. I actually have the PCB's and the parts to build these, time seems to be the issue. http://mysite.verizon.net/res75okq/sitebuildercontent/sitebuilderfiles/bigbloak_v2.pdf

 

[AndrewT]

Hi,
I think one can build a very good discrete for less cost than a very good audio ICopamp.

If the ICopamp also needs an ICbuffer then discrete is virtually bound to be cheaper.

The Borbely designs are based on lots of expensive FETs.
If value for money is the issue then FET front end and BJT the rest.

Both types (discrete/ICopamp) will benefit from excellent to superlative PSU.

 

[davidallancole]

For the price of one OPA627 (Digikey ~$20) you could make many discrete op-amps that may very well out perform the OPA627.