Article on "making your own" op amps?

[amonte]

Does anyone know where I can find this article?  I can't remember the specifics, but it was something along the lines of using two transistors (and I think some resistors?) to create or "simulate" an op amp...

[Mark Hammer]

It was in "Performer's Audio" about 9 years ago.  That magazine blended with "Glass Audio" and "Speaker Builder" to become "Audio Xpress", which is still available on the stands.  There may have been other articles on the same topic, but I know I scanned that one and sent it to Jay Doyle way back when.  Not sure if I still have that scan.

[Ben N]

Well. here is something along those lines, with a twist:
http://www.diystompboxes.com/smfforum/index.php?board=16.0

You can also search for designs by Jay Doyle--he likes discrete opamps.  There are some commercial designs, too, some with schemaatics available on the net. Two of the better Boss overrives, BD-2 and OD-3, use discret opamps in their circuitry.

hth,
Ben

[amonte]

Thanks, it was the Analog Alchmey post - but I can't find the pic of the Simple Opamp-Workalike... what was that again?

[Jay Doyle]

Mark, I'll have to look around for the scan, it is probably on my drive at home somewhere...

Here are some links to help get you started:

http://sound.westhost.com/project07.htm
http://www.4qdtec.com/opamp.html
http://www.johnhardyco.com/pdf/990.pdf
http://www.dself.dsl.pipex.com/ampins/discrete/discrop.htm
http://www.allaboutcircuits.com/vol_6/chpt_5/17.html

And this goes through a 741 and explains it:

http://en.wikipedia.org/wiki/Operational_amplifier

I'll dig around and see what else I can come up with. Also look at R.G.'s Tech of the Univibe, there is a discrete opamp in that circuit too...

Jay Doyle

[Ben N]

This?
http://www.diystompboxes.com/analogalchemy/sch/diodeopamp.html

Edit: Whad'I tell you--Jay is the man!

[amonte]

Was there only one?  I was reading my own post where I referenced the diode compression op amp and something called the "Simple OpAmp Workalike"  - but I can't find that anywhere.

Jay is, indeed, the man.  Thank you, sir!

[R.G.]

A few random thoughts about opamps and why they are what they are.

We guitar effects people are a tiny flea on the back of a very small dog in the barnyard of the electronics industry. Chips are designed and built for the barnyard, and we happen to see ways to pervert them to our uses.

Operational amplifiers were invented for the purpose of analog computation. The whole point of an opamp is that you can take an amplifier that has a variable, drifting (although high) gain and turn it into a circuit whose output is proportional to a mathematical operation. So operational amplifiers can do math operations like addition, subtraction, integration (try that on your pocket calculator!), differentiation, taking logarithms and exponents, little stuff like that. That's where the word "operational" comes from in the name.

After feedback was codified in the 1930's, the use of opamps to do certain computation was a military secret in the controllers that let US battleship guns throw a one-ton chunk of explosives 20 miles with an accuracy of +/-10 feet, and the Norden bombsight to place a bomb dropped through crosswinds for 20,000 feet inside a pickle barrel. They were computers, and as such, accuracy was an issue to be pursued with great effort. The analog sloppiness of gain error, input currents, offset voltages, drift, and so on were not to be tolerated.

And us guitar freaks did what we do. We looked at a polished scientific marvel and said "Kewl. Can we make it sound like Hendrix?"

Opamps can also be used for audio, we found out. All the high precision stuff is more or less ignored for this. As a result  only a few opamps were designed and have good results for audio.  It's a side effect for most of them.

Any circuit that has inverting gain can be used as an "operational" amplifier by hooking it up with feedback. The feedback adapts the excess gain to making the circuit perform more "perfectly" - that is, more nearly like the math predicts. In fact, until integrated circuits made operational amplifiers cheap, the use of two- and three-transistor circuits to do what we'd use an opamp for today was one of the EE's most basic tricks. The Univibe was designed before opamps were cheap enough for consumer use. So that three transistor circuit at the front was designed for what we'd use a dual opamp for today. Each phase section uses two transistors where we'd use one opamp. The two-transistor feedback stage was elevated to an artform. Today, it's hard to find the material on these transistor circuits because opamps are so cheap and so universally available. It's usually more expensive to make the transistor circuits in a manufacturing setting.

You can make replicas of opamp circuits in discrete transistors simply enough. What you can't do is to replace the matching and close thermal coupling that the transistors on an IC have.

A discrete opamp can be made with as few as two transistors. In this setup, the transistors both invert the signal, and the collector of the second transistor is the output. The base of the first transistor is the non-inverting input, and the emitter of the first transistor is the inverting input.

A better opamp can be made from three transistors. The first two are set up as a diffamp, giving both inputs at the same DC level, which is not possible for the two-transistor version, and a third transistor performs the voltage gain and output functions. This version can be made to have the same DC level at input and output, and so is the minimal setup for computational style opamps.

From the three transistor opamp, what you have left is embellishments. The first is a current buffer on the output to make the internal gain independent of external loading. After that you get current source loads and current mirror loads on the input stage, multi-device input stages for higher input impedance and more accuracy, stuff like that.

The discrete opamp can be a good audio device. The discrete JE-990 in its various forms is still used in some studio setups and is a great place to start if you want high audio performance. Of course in effects we often want the non-ideal behaviours, so some of the less ideal circuits may be "better" for effects. No way to tell except by listening.

[Mark Hammer]

That's gonna make for some decent reading material on the bus tonight!  Thanks.

In a wholly other direction, the old Boss ROD-10 provides a half-dozen or so distortions, all of which conform loosely in one way or another to the basic discrete op-amp topography that Jay has used so productively, and which is also found in the Boss BD-2 Blues Driver.  I forget where it's posted, but well worth taking a look at.

[amonte]

Wow - thanks, RG!  Great stuff.

I will take a look at the BD tonight as well...

[southtown]

so is there a noticable sound difference between a discrete op amp and a chip?

[Mark Hammer]

so is there a noticable sound difference between a discrete op amp and a chip?

Not if the discrete perfectly replicates what's inside the chip.  The thing about discrete op-amps is that you can design them according to your own parameters and performance goals.  There are literally hundreds of different commercial op-amps to choose from presently.  All of the different designs are directed at specific end-users and market niches.  I'm not so sure that any of those niches are for pedal-builders specifically.

[calculating_infinity]

Wow great info in here guys!  Ill have to print out and read these links.  Thanks!

[R.G.]

Excellent reply, Mark.

I'll add one more bit of info. It's a quirk of mine that when I'm trying to figure out what something does and what it might do if I perverted it a little, I try to imagine what the person who designed it wanted it to do in the first place. If you can get your mind around what was important and not important to them when they did the design, you can often guess what it will do well and not well when you (ab)use it.

By way of illustration: the 741 opamp. When this was designed, the idea that you could put an operational amplifier with internal compensation for stability on a single chip >>at all<< was a big deal. The 741 was a blast of clear fresh air over the design of analog circuits for low frequency signal conditioning. It was in fact that although it was not designed for it, you could put audio through it and it sounded pretty good. This was in the late 60's and early 70's. Today, a serious audio designer has coughing fits when you suggest using a 741, but that's with nearly forty years of better semiconductors under our collective belts. Audio was literally a side benefit to the use of the first opamps.

The whole point of opamps was then, and mostly is now the conditioning of analog signals, both at nearly DC and at very high speeds like video. DC accuracy is a Big Deal for some opamps. High speed is a Big Deal for video opamps. In the middle, we get special chips like the LM833 optimized for audio. Mostly we use opamps that happen to have good audio performance by accident, because audio signals don't care about DC at all and have very high voltage swings at high(ish) frequencies, something not found in video signals.

The JRE-990 is a descrete opamp that is designed to work from +/- 24Vdc, and to have full power output swing (about +/-20V peak) into 600 ohm loads while remaining very quiet and stable. If you want an audio opamp with excellent performance easy, get an LM833. If you want an audio opamp GOOD, make or buy 990s. If you want good, cheap, easy, small, etc. all at the same time, you want a TL072.

Is there a noticeable difference between discrete and chip opamps?
That's a little like asking if there are differences between Nordic and Oriental women. Of course there are. Are the differences ones you'd notice or want to correct? That depends very much on what you're trying to accomplish, and exactly which ones you compare.

[southtown]

Excellent reply, Mark.

I'll add one more bit of info.

it makes me laugh when he writes this then follows up with a essay we love ya rg

[Mark Hammer]

It's taken me years of patiently training him, but he seems to have finally latched onto what a good posting looks like.

[dano12]

Is it just me, or is this forum truly magnificent?

[R.G.]

It's taken me years of patiently training him, but he seems to have finally latched onto what a good posting looks like.

I like to think of it as standing on the shoulders of giants... 

Here's a couple of references for those who want to look into it. The first is Doug Self's take on it, which is a pretty good intro all by itself. Be forewarned, he got the schematics reversed. The simple opamp schematic is where the more complicated, better one should be and vice versa.
http://www.dself.dsl.pipex.com/ampins/discrete/discrop.htm

http://sound.westhost.com/project07.htm

http://www.johnhardyco.com/990OpAmpDetails.html

And the reigning queen of the discrete opamps:
http://recording.org/users/kev/Jensen_990C_schematic.tif

[rockgardenlove]

Is it just me, or is this forum truly magnificent?

Just you...this place sucks.

[jonathan perez]

Is it just me, or is this forum truly magnificent?

Just you...this place sucks.

+1