Discrete Opamp designs recommendations

[matmonster58]

I have to design a op amp for a class that I'm taking. This is an intro to ic design class so the focus is on designing something that works and not necessarily something that has great performance.

Since I'm interested in audio and preamps and such, I want to take the extra step and make something that I can use in later projects. I understand how opamps work but I don't really have the knowledge at the moment to make something that sounds good and to spec out available transistors (the class focuses on chip design so we don't really look at discrete components).

Does anyone have recommendation for opamp schematics I should check out and take inspiration from? I'm looking for something a little more modern the classic audio opamps.

I'm looking for something that:
Can be made in 2520 format
Can be made with readily available smd components
Uses minimal resistors and caps (class requires the use of current sources and mirrors for all the biasing)
Sounds good for audio and mic preamps

Not being able to use many resistors knocks out a lot of the more primitive opamps like the API 2520, Jensen 990, Neumann OA10

 

[Matador]

https://www.nanovolt.ch/resources/discrete_opamps/index.html

 

[JohnRoberts]

DOAs are pretty mature... What exactly do you mean by "design".... there are many published designs to use as a starting point (to copy).

I kind of liked the 990(?) that Deane Jensen did with inductors in series with the input long tail pair emitters.They created an interesting stability improvement.

JR

 

[matmonster58]

DOAs are pretty mature... What exactly do you mean by "design".... there are many published designs to use as a starting point (to copy).

I kind of liked the 990(?) that Deane Jensen did with inductors in series with the input long tail pair emitters.They created an interesting stability improvement.

JR

"Design" as in I am tasked with picking transistor characteristics and bias conditions for a simple opamp in a simulator from scratch. It's my final project for an intro analog ic design class. It's very much a theory/math based assignment that doesn't really correspond to a real world device. The premise of the class is for on chip fabrication, not a discrete device hence why I can't use many resistors and definitely no inductors.

I just want to tie the project into audio doas because it is what I'm personally interested in. I want to link the theoretical knowledge to something practical I plan to use in the future. I know how to make an opamp, but I don't know how to make an opamp that sounds good. That's why I'm looking for some proven designs.

You can get away with a lot in a doa because you typically have a lot of space to work with. In my case where I need to follow more "industry" style design practices for on chip fabrication, minimizing die size has to be taken into account at least a little bit

 

[JohnRoberts]

IC design indeed has many more options (like active devices with multiple junctions)..

I recall reading a good paper in the IEEE journal decades ago, about the design of the ua741. It is generally considered a low performance op amp but it was the first internally compensated op amp and a milestone IC design back in its time. IIRC that paper described a lot of the nuts and bolts about op amp design. Namely differential input stage, level shifter, and output buffer. Of course the devil is in the details.

JR

 

[ruffrecords]

Design implies a performance specification that needs to be met and that can be measured in order to verify that this has been achieved by the design. It is a lot more than picking transistor characteristics and bias conditions. If you want to learn how to design then you need to understand how the choices you make affect the performance of the design but before you do that you need to understand what performance you need. So question one is what do you think are the important performance characteristics for an audio op amp?

Cheers

Ian

 

[Dylan W]

Check out the Nanovolt DOAs in the link above (Samuel Groner). Here are a few more high performance, "modern" DOAs:

Martech
https://groupdiy.com/threads/doa-repair.80153/

GML
https://groupdiy.com/threads/gml.78045/

Great River MP2 (current mirror in the LTP)
https://groupdiy.com/threads/great-river-mp-2-preamp.48791/#post-616438

 

[tk@halmi]

Designing an operational amplifier on a wafer (IC) is really different from designing a discrete circuit. The devices available to you will have little in common if we are talking about non-theoretical circuits. Trying to make a circuit that fits both worlds will have a rather suboptimal result.

When you design discrete circuits there are few limits other than your wallet or the customary one square inch layout. On the wafer you are restricted to the process design rules. The fab that makes your chip will give you those rules because that is what the production line is capable of manufacturing.

Without understanding the design rules for the IC it is difficult to make recommendations. The restrictions of few passive components are not sufficient. What process do you have available, BJT, MOS? What are the device parameters for available transistors? What is the resistor resistance range you can use in your IC process?

Almost always the IC amplifier will be far more complex in order to work around passive component process limitations. It will be riddled with current mirrors, current sources, and attempts to replace a resistor/capacitor with several transistors. You can find many examples on the web for each and see how differently they are implemented.

Just for giggles, I attached a dorky paper I made twenty years ago for a very simple discrete opamp. It was created for learning purposes, and offers some details about the design steps.
http://myplace.frontier.com/~tamas....sitebuilderfiles/B7-discrete-opamp-7-5-04.pdf

 

[Wavelength]

I would look at some standard 2520 schematics then work around removing resistors for silicon. I remember doing that starting with the 990 like 30 years ago for a phono stage I built.
Simulators may be something you really want to get into before you start. My cousin designs CPU's now and has a massive simulator farm they use to create libraries for parts they build.
Most of my friends that design opamps are looking at folded cascode designs because they are asked to design opamps that can work at low voltages. I use folded cascodes in guitar pedal designs because it gives you a more effective use of the limited 9V supply.

 

[CJ]

Build a vacuum tube opamp for extra credit, that's where they started anyway,

Tube Tech uses vac tube opamps,