What are these capacitors for?

Things like this always puzzle me. At one time it seemed that a circuit consisted of an active component (transistor, chip or tube), with several passive components connected to them (caps, resistors etc) to form an amplifier or oscillator, or whatever. Needing mutliple caps like this seem rather odd. Like, who decides where they need to be?

Over the years, I worked on more old tube equipment than I have the newer stuff. Things were never so critical in tube circuits. Some high frequency RF circuits needed under chassis shielding and tube shields on the top, but for the most part, wires fell where they laid and everything worked. I always knew to keep the AC power cord away from components and other wires as much as possible, but that was about all it took.

Even the early transistor qquipment did not seem as critical. Now we have what sometimes seem to be too many things to consider. Just making a PCB seems to be a major accomplishment, because you cant just make a board so all the parts fit together as needed, but the traces have to be kept away from other traces and so on.... I often wonder who and how long it takes to design a PCB to fit all these parts.

This Equalizer is around 30 years old, so I am sure it's worse now to design things. One good thing, Peavey still supports and has parts for this equalizer, that gives me a good opinion of them. We already have a glut of old electronics polluting the earth and we sure dont need to add more, especially when the stuff is still useful and works fine, with only a little care. This one works quite well, sounds great, but had a bad power switch. Somehow one of the slider pots was broken off too, and it's the GAIN control for one of the channels, so it's important.

This thing has 26 chips. I believe they are all Op amps. There are 6 transistors, and a few diodes. The rest is mostly all caps, and resistors which form the filters for the freq bands. Beyond that, it's just the power supply parts. I like the fact that all of the chips have sockets. But I sure wish they had labeled the parts on the board.

On a sunny day (Tue, 24 Oct 2017 18:09:20 -0500) it happened snipped-for-privacy@tubes.com wrote in :

As to those caps, that is a normal way to draw those for power decoupling, the caps will be placed next to opamps and other stuff, not all next to each other.

The go from the supply pins to ground as close to the IC package as possible for the reasons Phil stated. It's not rocket surgery. Forrest Mims also recommends it in his beginner books on op amp circuits:

Good grief. For the benefit of Mr. Allison & Bitrex...

1. Changes in IC output current interact with supply impedance to cause voltage variations on the Vcc pin
2. These voltage variations get into the earlier sections of the IC's circuitry via the Vcc pin
3. Gain exists from input to output That lot is a recipe for oscillation aka instability. When sufficient phase shift occurs along with sufficiently poor PSRR and enough gain, oscillation happens. This isn't good.
4. I thought this next point would be obvious to anyone that designed electronics, but perhaps it isn't. A capacitor across the rails close to the chip much reduces supply impedance at hf and cleans up the crap on the rail. Result: stability.

I don't seriously believe you don't know that, so thanks for trolling us. I'm sure we all appreciate it and respect you all the more.

NT

it means reducing voltage variations on the rail. Who here thinks it doesn't?

it doesn't of course

no

NT

but not the basics of decoupling psu rails.

It's a fairly basic skill for electronic engineers.

NT

Low frequency stuff is not critical; just worry about ground loops and hum pickup, like in the tube days. It helps now, not having a filament supply.

But some parts are fast nowadays, like logic that switches in picoseconds or GHz opamps. There, layout and trace impedances and the speed of light really matter.

A square foot of dense logic, with switching power supplies and a couple FPGAs and some fast analog stuff and thermal worries, say a thousand parts, might be a 10-layer board and could take six weeks or so to lay out manually. That becomes a team project.

I like reference designators silkscreened on my boards, but they take as much area as the parts.

HF instability doesn't have a thing to do with "voltage variations being injected into earlier sections of the IC"; the PSRR of the output stage will drop to 0dB (or even turn into gain) long before the current source/mirror structures of the voltage gain stage ever does, and will oscillate perfectly well on its own.

Oscillation from feedback from the output to the input via the rails is called "motorboating", it's a low frequency phenomena, and generally happens with single supply topologies that don't have sufficient bypassing

They also use them in digital circuitry, in which they have a different use.

If it has any parametric controls at least one is probably gain control.

design things."

Computer aided now. It was 30 years ago as well. However, before that someone had to sit there and draw all the races. For prototypes they had a etchant resist pen and stick on filets for the pads. They had a few other aids developed over te years.

I'm familiar with the term motorboating, and recall having some amplifier cause it in the past. That was years ago, so I dont recall the reason anymore, but I recall the sound.

What's weird with this equalizer, is that I looked at all the caps. I dont see one 0.1 cap on that board, but I see a .02 next to each opamp. I read that there were variations in this circuit, so I assume those are the ones shown on the schematic, but the value was changed.

---------------------------

The caps restore stability by lowering the supply impedance.

This is NOT noise filtering. ------------------------------

FOAD you arrogant, POS f*****ad.

..... Phil

Providing a low impedance path for high frequency noise, to ground, isn't "filtering"? LOL!

Most FPGA and MPU manufacturer's recommendations are absurd.

For bulk caps, long leads don't matter much. If you're worried about it, add a ceramic directly to the board.

Have you looked at the shoot-through current of the AC family (L dI/DT, you know)? Amazing stuff (in more ways than one).

And I had tube stuff that needed a #14 wire run up the side of the envelope as a decoupling capacitor.

It's all about edge speed and inductance (L dI/DT).

IC sockets are evil.

Right. Give the layout guy and subsequent service tech a hint where the caps go.

---------------------------

FOAD d*****ad.

..... Phil