Solid state question

[Stevem]

I am building some new preamps using IC chips to replace certain sections of older SS amps for noise reduction reasons.
Part of what needs to be done due to the distance these chips are from the power supply to keep oscillation at bay is to bypass the + and - suppy inputs at the chip with a small Cap.

In looking over a few schematics I have seen .1 caps used for this and .01 caps used for such.
So my question is which is right and or which is better?

Do the cap values also effect the frequency reponce of that gain stage?
In many amps for better tone that I mod ( both SS and tube ) I bypass the first main filter( s) with a .22 and a .022 in parallel.

[Tony Bones]

Those bypass capacitors are meant to provide a low impedance path for high frequencies; they bypass the inductance of the PS conductors. The value is not especially important (I use 0.1uF) but they do need to work at high frequencies, so ceramic is typically used. C0G/NP0 type are better, but probably not strictly required (they cost more.) C0G/NP0 specifies a low temperature coefficient. If the precise value of capacitance isn't important, then why do we care about the tempco? We don't. It turns out that C0G/NP0 caps are made with a better dielectric that has other advantages over the usual multilayer ceramic caps.

[R.G.]

Correct again, T.B.
It's hard to find 100nF in COG/NPO, or was the last time I delved into low tempco caps. One may be stuck with other tempco insulators by availability.

You're right, the whole idea is to provide a local bucket full of charge that the IC can use without trying to suck changing AC-over-DC through the inductance of long leads, with the corresponding voltage sag and ringing of the combined inductance and capacitance of the traces.

The whole business of bypassing otherwise fine bypassing caps comes from the fact that all capacitors have some internal inductance, and the bigger the capacitance, the bigger the inductance. Since the impedance of a capacitor goes down linear with frequency, and the impedance of an inductor goes up linearly with frequency, every capacitor has some frequency where the inductive impedance equals and exceeds the decreasing impedance of the capacitance; above that frequency, it's not a cap, it's an inductor. "Bypassing" the bypass cap with a smaller cap also gets you a smaller inductance, so you still have declining impedance up to a higher frequency.

In general, linear amplifier ICs are fine with one 0.1 or 0.01 cap from its power supply pin to ground, although nearly any value works, as T.B. correctly points out. Ceramics are used for this because they're both low ESR, low ESL if you get planar multilayer ceramics, and even more importantly, low cost per unit.

I put spaces for a 0.1 from power to ground per IC on all my layouts. When I try out the boards, I will leave some of them out until it's proven that they're needed. On a PCB, it's so much easier to leave off a part you don't need than to jam in a new part you don't have a place for.

If this sounds a bit slapdash, that's because it is. Modern ICs are vastly improved in terms of power supply immunity than the used to be. Early opamps would oscillate and sometimes die if you didn't decouple them perfectly. I learned most of this from laying out IC memory arrays. When I got my first engineering job, the hot news in memory chips was the one ... KILO-BIT (not even -byte) memory. All one thousand bits of it. It needed three different power supply voltages and would die if the three supplies were not brought up in the correct sequence.

And we had to walk to school. Barefoot. In the snow. Uphill both ways...

[Ten Over]

And we had to walk to school. Barefoot. In the snow. Uphill both ways...

Bill Cosby?

[Stevem]

Thanks for the responses folks!

[martin manning]

We don't. It turns out that C0G/NP0 caps are made with a better dielectric that has other advantages over the usual multilayer ceramic caps.

100n COG MLCC: https://www.mouser.com/ProductDetail/TD ... wJcYncM%3d