In the Lab Manual for Floyd's Electronic Fundamentals, he gives this as a typical power supply for Op Amps. I don't think the capacitors are really necessary so am I missing something here? What would be the purpose of coupling capacitors here? I could understand it maybe IF there was a lot of ripple in the supply.

 

They are not coupling capacitors, instead they are supply filter capacitors to keep the opamp from oscillating at a high frequency.
Usually the main supply filter capacitors are high value to filter low frequencies but they work poorly at high frequencies and are too far away from the IC with the distance causing inductance in the wires feeding power to the IC. Therefore these high frequency capacitors are mounted very close to the IC to eliminate the high impedance caused by inductance of the wires.

 

Hello there, Sir.


 

Here's some good reading on the topic of decoupling caps. As AGa said, they're needed to keep the op amp from oscillating.

 

Indeed the capacitors are DEcoupling capacitors, as partly explained in post #3 In a simulator with zero impedance voltage sources the capacitors are not required, in the real world you may find that without such capacitors the operation may not be what you expect.

 

K, thanks. I did notice getting an oscillation when working on a non-inverting amp and wondered where it came from. I actually meant to type decoupling. Not doing this on simulator, building the circuits to test IRL.

 

I did notice getting an oscillation when working on a non-inverting amp and wondered where it came from.

You might find the attached app note interesting.

One thing that especially vexatous about op amp circuit oscillations is that they often occur at frequencies far above what we can pick up on a hobby-grade oscilloscope-- sometimes at VHF or even UHF frequencies. Although they may not show up on a scope, they often manifest as mysterious anomolies at lower frequencies-- like inexplicable DC shifts, discontinuities in the input-output transfer function, noise, gain errors, signal-dependent gain errors, and all kinds of other goofy behavior.

100 nF ceramic capacitors are cheap as cat-dirt. I have a box of thousands of them. Every power supply connection of every chip on every project gets one. It's cheap insurance.

 

I'm not sure if I can replicate the oscillation I saw. I think I had left the decoupling caps off the breadboard rails. The task was to build an amp with 47 Acl. I used a 1kΩ Ri and a 47kΩ Rf with the sig-gen set for 100mV @ 1kHz. Since I had been getting a noisy signal at that level I ran it first through a Voltage Follower 741 amp to remove the noise. What I got was a modulated signal like feeding ~100hZ into a 1kHz carrier. I didn't really measure it just thought that's odd and replaced the resistors with a 10K and 470K and there was no oscillation except the 1kHz. I have 2 sig-gens and the one that I can adjust to single-digit mVs is noisy at low voltage levels. My other one can only be adjusted to a single decimal place but is quieter. I was using the 3 decimal place one in order to tweak the signal to get exactly 100mV input seen on the scope. Anyway, thanks for the PDFs! I scanned through them and stored them for reference.

 

You might get away with omitting the decoupling capacitors if you were a hobbyist building a project for your own consumption.

If you are a professional circuit designer designing a circuit for production and resale then it is a different matter. As a professional you would already know why this is not optional.

 

I am a bit dismayed that the book didn't mention their need. So I learned something from it. Anyway, thanks again guys!

 

An antique 741 opamp is 52 years old, is noisy, has a low max frequency response, has a low slew rate and needs a high supply voltage of +15V and -15V.

A breadboard has high capacitance between its rows of contacts and jumper wires all over the place that cause an opamp to oscillate. The rows of contacts and wires are antennas that pickup interference.