Discrete Opamps

[dano12]

Enjoy!

http://www.audio-gd.com/enweb/pro/diy/OPA.htm

[Mark Hammer]

Some time back, I scanned an article in Performer's Audio (a short-lived publication by the folks who brought you Glass Audio, Speaker Builder, and currently Audio Xpress) for Jay Doyle, that was written by Nelson Pass, concernng the design of all-discrete Op-amps.  Jay certainly made VERY productive and inventive use of the information, coming up with a surfeit of elegant circuits constructed around discrete circuits.  I'll have to take a gander on my hard-drive and see if I still have it.

On the other hand, the unit shown simply plugs into an 8-pin DIP socket.  Hard to get simpler than that!

[R.G.]

Cute.

Also see:
http://www.dibsplace.com/design/CIRCATS/je-990.pdf
http://sci.tech-archive.net/Archive/sci.electronics.design/2004-07/3898.html

For opamps, there are good reasons to question whether discrete offers any performance advantage. It is by no means clear that discrete is better, given that matching of active devices is an area of huge importance.

The devil makes me want to make up similar PCBs and fold them up that way to hide the LM833 in the middle, then ask some "Golden Ears" to tell me which is better, the old Earth/Sun/Moon modules or the new, improved "Jupiter" module. 

You do have to go discrete when you need more than +/- 18V power supplies or higher power drive (that is, higher voltages into 600 ohm balanced inputs) just because the package and power limitations mean that IC opamps can't stand that much heat.

[Sir H C]

gack, I don't like the schematics.  What is going on with those output stages?  Looks like you can not swing on the output more than 1/2 rail either way (minus a Vbe).  Otherwise a two stage, folded cascode op-amp, not even close to rail to rail in.  (comments on Earth op-amp).

The sun looks weird, don't like that second stage setup, looks like it would be a real pain to keep matched and properly biased. 

Moon is another odd one, follower output stage with current source, and why that transistor in the middle? 

I hope they don't show the compensation capacitors, interesting designs, not sure that I would jump for these before a Burr-Brown/TI or ADI op-amp.

And why jFET input?  That will be higher noise and lower gain than bipolar, to me the only reason to think of going discrete.

[R.G.]

gack, I don't like the schematics.  ...
What is going on with those output stages?  ...
The sun looks weird...
Moon is another odd one...
I hope they don't show the compensation capacitors ...
And why jFET input?  ...

You forget that these are obviously for hifi tweekos, given the verbiage on the site. Reason plays little or no part in whether they "sound good" or not. In fact, the quirkier they are, the more likely a hifi reviewer is to say that they "have an inner clarity and detail, as though a veil has been lifted from the sound, allowing the notes to stand out from each other with an almost etched quality" and so on for about five pages.

[Andi]

["an inner clarity and detail, as though a veil has been lifted from the sound, allowing the notes to stand out from each other with an almost etched quality" and so on for about five pages.

That is some beautifully written parody. 

[Mark Hammer]

Once again, I hasten to remind people that what may provide advantage in the case of multi-source wide-bandwidth signals  (i.e., mic'd orchestras or acoustic jazz combos) being reproduced in unflinchingly accurate sound systems is not necessarily what will provide advantage when cranking a Les Paul into a Marshall stack.  I don't see any guitar players rushing to implement electrostatic speakers in their amps, despite the well-known superb reproduction those can provide.  The reason?  Electrostatics work their magic onthe top end; a top end that few, if any, electric guitarists will ever produce.

As such, I will not pooh-pooh these units.  But would/could/should such a module make your Distortion+ or Bluesbreaker sound "better" in some manner?  I'll get back to you on that.  Right now I'm bent over waiting for monkeys to fly out my.....

It's all a question of matching the equipment to the detectable/audible properties of the signal it is intended to process.

[Ben N]

RG, if the Visual Sound gig doesn't work out (heaven forbid), I think you may have the makings of a fine audio reviewer. 

[MarcoMike]

[petemoore]

  So does that make it easier to adjust/slow down the slew rate ?

[Sir H C]

  So does that make it easier to adjust/slow down the slew rate ?

Partially, since the output is followers, they are not really current limited the same way that some other op-amp circuits are, and there doesn't seem to be any current limiting in the output stage other than the current source pull down on the third circuit.  You could do the third circuit, include a PNP current source above the follower device and then you would have current limited slewing to that current limit.  Could do cool things with it.

[aziltz]

i was just thinking of this today...  but using the Diode-Compression Op-Amp posting in the other forum, this looks alot more complicated.

I was hoping to build something much smaller that would fit inside an OD pedal.

[Sir H C]

Look at the discrete op-amps that are in some of the Boss pedals, simple, effective, can do cool manipulations with it.

[aziltz]

Look at the discrete op-amps that are in some of the Boss pedals, simple, effective, can do cool manipulations with it.

are you thinking of the OD-3?  I have the schematic of that... at first glance it seemed like alot of Fets for a simple overdrive...

[MikeH]

And now you know why... 

[WGTP]

http://www.aronnelson.com/gallery/main.php/v/KHE/Boss_ROD10_Schematic.jpg.html

That's what I'm talking about.   

[Jarno]

Will read all the comments here tonight, but on DIYaudio.com there has been a groupbuy of discrete opamp PCB's. These dual opamps used a couple of 2sk389's (now obsolete, or expensive, or both) and supposedly sounded rather good (but measured not as great as a OPA2604, OPA2134, 5534, AD825, AD797 or a host of other opamps).
Even though I have a handful of the PCB's, I haven't tried them yet, but am planning to.
I think I also have a synth VCF schematic somewhere which uses a state variable filter containing discrete opamps.

Best regards,

Jarno.

[Sir H C]

The Jensen 990 is probably the best out there that I know of, the API 2520 is another classic, these stand up to integrated ones, and as RG says, can use higher supplies.  Neves are cool but not as easy to use or build.

[R.G.]

There are a few critical places in the typical Linn-architecture opamp where matching is *critical*, a few more where thermal linkage is critical, and these tend to be, but are not always, the same places.

For a look at what's good for a discrete opamps, look at the designs for audio power amps. These are essentially discrete opamps scaled up to the tens and hundreds of watts levels. It's in discussions of these amps that you get down to what is critical for making discrete opamps. I highly recommend Doug Self's book on audio power amps.

The input stage of a discrete opamp is *critical* to getting low offset, low drift, good bandwidth, low distortion, yada, yada. Simple resistive loads on the collectors of the input diffamp are not suitable to accuracy; you need a current mirror load. You typically need emitter degeneration in the input pair, you usually need two pole compensation in the voltage amplifier.

Input JFETs may be nice-sounding, but be aware that "nice-sounding" instead of "accurate" means that there are distortions there that you happen to like, not that the amp is universally good performance. JFETs have a much lower transconductance and produce measurably more distortion than bipolars in diffamp input stages, as well as higher thermal drift. But then the bipolars have a much lower input impedance.

It's been about forty years now that really really smart analog design guys have been paid big bucks for figuring out how to make opamps better. If I were a betting man, I would not bet that I could walk into that snake pit and design one better than the state of the art unless I was doing that for a living.

[Ben N]

Input JFETs may be nice-sounding, but be aware that "nice-sounding" instead of "accurate" means that there are distortions there that you happen to like, not that the amp is universally good performance.

Which happens to also be the gist of the hifi argument for tubes, right?