Discreet Op Amp

[sevenisthenumber]

I want to play with replacing op amps in things like a Tubescreamer or whatever with discreet transistor opa amps... Where can i start to learn and try?

[R.G.]

Google JRE-990.

[Mark Hammer]

I want to play with replacing op amps in things like a Tubescreamer or whatever with discreet transistor opa amps... Where can i start to learn and try?

I have no objection to, and offer every encouragement to, learnng.  But I have to ask, why the hell would you want to DO something like that?  I mean, you may as well want to learn how to make your own resistors.

I'll just note in passing that Jay Doyle posted a number of discrete op-amp circuits here over the years, and that one can see some interesting discrete designs in a variety of Boss products, like the Blues Driver, and several of the subcircuits in the ROD-10.  You will note, however, that these designs are intended to serve a particular purpose in and of themselves, and not simply insert into a pre-existing circuit. 

I do think there is something to be gained by learning how to devise these things from the ground up.  There was a magazine on the stands briefly in the 90's called Performer's Audio, or something like that, from the folks who brought you Glass Audio, Speaker Builder, Audio Amateur, and who now bring you audioXpress.  One of the issues had an article by Nelson Pass or somebody like that on designing discrete op-amps, that I think I scanned for Jay.  I may have either the scans or the original somewhere.

And, um, it's "discrete".  Discreet was one of Frank Zappa's labels.

[PRR]

> um, it's "discrete".  Discreet was one of Frank Zappa's labels

dis·creet  
1.Careful and circumspect in one's speech or actions, esp. to avoid causing offense or to gain an advantage.
2.Intentionally unobtrusive.

Perfectly good word. Wrong for parts-up op-amp, and wrong for Frank (probably his point).

see also: http://www.dailywritingtips.com/the-difference-between-discreet-and-discrete/

[sevenisthenumber]

Google JRE-990.

Nothing there???

[CodeMonk]

I want to play with replacing op amps in things like a Tubescreamer or whatever with discreet transistor opa amps... Where can i start to learn and try?

I have no objection to, and offer every encouragement to, learnng.  But I have to ask, why the hell would you want to DO something like that?  I mean, you may as well want to learn how to make your own resistors.

I'll just note in passing that Jay Doyle posted a number of discrete op-amp circuits here over the years, and that one can see some interesting discrete designs in a variety of Boss products, like the Blues Driver, and several of the subcircuits in the ROD-10.  You will note, however, that these designs are intended to serve a particular purpose in and of themselves, and not simply insert into a pre-existing circuit. 

I do think there is something to be gained by learning how to devise these things from the ground up.  There was a magazine on the stands briefly in the 90's called Performer's Audio, or something like that, from the folks who brought you Glass Audio, Speaker Builder, Audio Amateur, and who now bring you audioXpress.  One of the issues had an article by Nelson Pass or somebody like that on designing discrete op-amps, that I think I scanned for Jay.  I may have either the scans or the original somewhere.

And, um, it's "discrete".  Discreet was one of Frank Zappa's labels.

I would be interested in that, if you should run across it.
No rush or anything, just if you come across it some day

[Mark Hammer]

> um, it's "discrete".  Discreet was one of Frank Zappa's labels

dis·creet  
1.Careful and circumspect in one's speech or actions, esp. to avoid causing offense or to gain an advantage.
2.Intentionally unobtrusive.

Perfectly good word. Wrong for parts-up op-amp, and wrong for Frank (probably his point).

see also: http://www.dailywritingtips.com/the-difference-between-discreet-and-discrete/

Huh.  Well, I'll be.  NEVER saw that coming.  I'd only ever seen it spelled the one way, with the exception for Frank's record label, and my tacit assumption was that it was deliberately mis-spelled as a sort of linguistic joke.  And I pride myself on good spelling (the Canadian way).

I'm just glad THAT wasn't my 20,000th post.  

[Mark Hammer]

I would be interested in that, if you should run across it.
No rush or anything, just if you come across it some day

Send me a PM with an e-mail address that allows attachments, and I'll send the package when I find it.

[Tony Forestiere]

Here: http://www.johnhardyco.com/pdf/990.pdf

[CodeMonk]

I would be interested in that, if you should run across it.
No rush or anything, just if you come across it some day

Send me a PM with an e-mail address that allows attachments, and I'll send the package when I find it.

Done.
Thanks

Oh, and I forgot to include a subject line
(Dummy me pasted the subject line in the BCC box    )

[R.G.]

Sorry - for some reason I just keep including that "r". It's the JE-990. Here's one link.

http://www.dibsplace.com/design/CIRCATS/je-990.pdf

The schemo is available other places too. The JE-990 was intended to be a no-compromise audio opamp. As such, it cannot compete with integrated ones on DC accuracy, drift, etc., but will be capable of higher power supply voltages, more current output, wider open loop bandwidth, and lower noise if you're very careful about how you build it.

[sevenisthenumber]

What are these?

[Mark Hammer]

Hah!  Once again, memory DOES serve.

I remembered it as late 90's and Nelson Pass.  Mr. Pass has a site with an archive of articles and projects, and lo and behold, he has the article itself as a PDF.

Enjoy!  http://www.passdiy.com/pdf/diyopamp.pdf

[R.G.]

Pretty good intro, OK.

Actually, most people forget that the now-standard Linn three-stage power amplifier with a diffamp in front *is* an operational amplifier. So you can often copy the format of a power amp, scaling currents down.

It's possible to build one of these pretty easily. I used to think about how I could make one with minimal parts that were NOT semiconductors.

You need an input diffamp, with a current source tail for bias. You ought to use a current mirror load. The voltage amplifier ought to be current source loaded from a current mirror. Oops - have to use a resistor for the current sources for the input stage and voltage amplifier if you don't use selected JFETs as constant current diodes. The output stage is some form of emitter follower, but you need some biasing and ballasting parts.

You need a few resistors for biasing, a cap for the dominant pole compensation, etc, but you get rapidly down to mostly all semiconductors. This could get pretty small in SMT. Maybe only a few times bigger than a DIP. Should take about an hour to do a PCB. 

[slacker]

What are these?

They're Joe Davisson's simple discrete opamp, so good it's got its own part of the forum http://www.diystompboxes.com/smfforum/index.php?board=16.0

I built a Rat type thing using it, sounds really good. You get the most out of it using it without any clipping diodes, so you hear its own distortion, this is softer than real opamps, once you add clippers it sounds pretty much like any other opamp + clipper distorter.

[Kesh]

Here are some old Stompboxology reproductions

The one called "going discrete" has a very simple discrete op-amp or two

http://moosapotamus.net/stompboxology.html

[brett]

Hi
if I'm not wrong, the single transistor gain stages in the BMP behave like inverting op-amps (ie gain ~ Rfeedback/Rin, Zin ~ Rin, etc). Not a differential input, but how often would we use both the + and - ? If you want + and -, you could add a long-tailed pair.

Probably more important than any answer to your question is the question: Why?
cheers

[PRR]

> The JE-990 ... ... and lower noise if you're very careful about how you build it.

Lower hiss *in low impedance circuits*.

The high-impedance (5K-50K) noise optimum had already been hit. Even the old 2N5087 can hit 0.5dB NF (*for* 10K-50K source). There's bunches of preamps based on such stuff.

Jensen had two input inspirations:

1) a 1:1 transformer has MUCH better performance than any step-up

2) The then-NEW LM194 had MUCH lower rB than '5087 types which lowered its OSI into the range of dynamic mikes, 200-2K.

There's much more to it. Compensating a BJT input for any-gain stability hurts slew; he found a semi-novel (and patentable) way around that (and one which chips can't do). Another trick compensates the output against load capacitance (which destabilizes emitter followers). Carrying on, the high cost suggests doing the most with the fewest opamps, so the output is BEEFy and can pump Watts. While DC error could be moot, with high-perm transformers (and few coupling caps) it isn't, so a revised model corrects a milliVolt offset tendency. (Tho as-was, the 990's offset was way ahead of any chip of that day, and many potted opamps.)
______________________

> What are these?

"Poor" opamps.

The first one has significantly worse input bias and output drive than the '741 it replaces. Grokking that left me too tired to study the second, except I "know" that 3-transisor opamps are in generally barely-adequate as "ideal-enough" opamps. And *in general* if you can only use 3 transistors, in Audio there is usually a non-opamp plan to give better audio performance.

At 8 or 9 transistors you can build very fine opamps.
______________________

> minimal parts that were NOT semiconductors.

The classic OTA has only transistors (assuming some current-source for overall bias). There used to be programmable opamps which were all transistors.

I thought you meant "without semiconductors". Electrolytic amplifiers are possible. And the whole Fluidics thing that never went anywhere. But the classic is Tube opamps. The more general-purpose Philbrick is two twin-triodes (four devices). Heath had a one-bottle opamp, but it was inverting-only (classic analog computation is all inverters; the equations are easier to write).

[R.G.]

I thought you meant "without semiconductors". Electrolytic amplifiers are possible. And the whole Fluidics thing that never went anywhere.

Being a transformer-y guy, I'd have gone to magamps before fluidics.

Yeah, OTAs are all-semiconductor, but I was trying to think of the fewest non-semiconductor amplifiers in the format of the Linn three-stage with a differential front end I mentioned a few sentences earlier.