Simple (=stupid?) RC Filter Question

[lars-musik]

Dear Forumites,

after many years building pedals I am often reaching the limits of my knowledge.

Here's a question (I asked myself and google from time to time) that I'd like to have answered by you insightful and sympathetic lot.

Are these two RC lowpass filters the same? Both have their cutoff frequency at 28kHz. But are there real life advantages of one over the other?


The background: I am planing on tinkering with Tom's "Vintage Crack, Hiss and Pop" circuit and would like to adjust the high- and lowpass filters in the tone shapig segments by dual-potentiometers. As they are virtually impossible to get with different values in one potentiometer, I'd go for a dual 100K.

https://electricdruid.net/wp-content/uploads/2017/04/HissCracklePop.jpg

Thanks, Lars

[ElectricDruid]

The significant difference between them is the load they present to the circuit feeding them. The second one is a 100K impedance (ish) so it won't load things too much. The first one with a 5K6 resistor is going to present a much lower impedance, and losses might be proportionally higher.
In many situations, it won't matter. For example, if you were using one to smooth the signal from an op-amp distortion stage. Most op-amps will happily drive a load of a few kilohms, so there's no problem. On the other hand, if you were sticking it after your favourite passive tone stack, you might well find that even 100K would be a considerable load and the 5K6 version would be a disaster.

[lars-musik]

Thanks for your answer, great to have the creator of that circuit on the line.

If I understood you correctly the 100K would be equally OK, or even better.

So, this will work (with the two 100K connected via the arrows as a double potentiometer?)

[antonis]

Additional to what Tom said, you might face them as Money vs Noise..
(bigger caps are more expencive than smaller ones..)

P.S.
Never dealt with Common Base configuration so, I can't tell you if source impedance driving TR3 is OK or not..

[ElectricDruid]

If I understood you correctly the 100K would be equally OK, or even better.

Yes, in most situations that's probably true.

So, this will work (with the two 100K connected via the arrows as a double potentiometer?)

Yes, it should work. But imagine it without the caps for a minute and look at what you have left. It becomes a voltage divider with 100K/100K. So at the point that the mixer resistor connects to it, you only have half the signal. Of course, that can be compensated, but that was partly what drove my original choice of 5K6. A divider of 5K6/100K drops a lot less level than 100K/100K.

The general rule of thumb when cascading passive filters like this is to multiply the resistor values up by ten for each one (so 1K for the first filter, then 10K for the next, and 100K for the next). That way, each filter doesn't load the previous one too much. In my circuit, I went for "x20" instead, because I could.

I can't remember the answer to Antonis' question about the common-base mixer. It isn't a configuration you see much, and I only did it because I wanted something simple that would run at 5V. On a 9V supply, I'd probably swap it for a proper op-amp mixer.

[GibsonGM]

I am curious how increasing the R values 10x or 20x impacts the noise floor of the circuit? 

I know that at higher (tube amp) voltages, it's usually desirable to scale DOWN the R's, and up the cap values, within reasonable limits to avoid loading problems, with the goal of making these networks quieter.  Maybe ( or probably) noise increase is only a 'big' factor at much higher operating voltages?  We see this a lot in interstage coupling, where "mods" are to scale R/C's this way to improve performance with respect to noise...

[PRR]

In general, you pick the filter resistor (or the circuit around it) to be greater than the source impedance and less than the load impedance. The more/less it is, the better the reality matches a simple-calc.

This is sometimes tough.

In THIS case, Tom's source is 100K one way and <1K the other way. A simple single impedance does not capture reality.

Yeah, just try it. Much simpler than brain-torture, and quicker than implementing buffers. You may in the end find buffers are desirable, but KISS for a first-try.