Presence Pot 5K vs 25K

[roberto]

Here we are.

- 5k lin pot fully CW with sweep on 100n 220n 330n 470n 680n 1u;
- 5k lin pot with 680n;
- 4k7 plus 5k lin pot with 680n;
- 4k7 plus 10k lin pot with 680n;
- 4k7 plus 25k lin pot with 680n;

[RockinRocket]

Maybe I misunderstood ...but I thought you were talking about the 25k with the 4.7k resistor as being half..... The 4.7k // 5k version was a mistake that Marshall eventually noticed. This configuration would make the pot 2.5k...and yes it would clearly limit the pot to half ...but this does not apply for the 4.7k//25k version...as the pot is not grounded....even if it was ..it's still not half .... Just trying to keep this clear....

I don't know any more!
Looking at the 5k/4k7/100n VS 5k/100n. At about half on the dial 5 is where the 5k/4k7 starts on the old dc configuration. That could be considered half as effective I suppose as Weber said but I am unclear as o0 what circuit he was in reference to.

With all plots considered. The old 5k dc on the pot has the superior range. The 25k comes close but the sweep is not very user friendly. To make it good solution a different taper pot would be needed as Roberto mentioned.

Roberto, Thanks for sharing and talking the time to do this. Gives me a great perspective with out having to use my ears that usually play tricks on me!

[Cameron]

The other point is that the scratchy solution is linear in its range, while new ones aren't with linear pots.
That's why some builders switched to rev log pots, to linearize the effect while keeping away the scratchy noise.

The 25k is linear in most amps that use the 4.7k//25k version ....log would bunch it up at the end ..so they don't work for a presence control. I use reverse audio when can find them ..and have done it for years. The other way I like to do it is....use a 10k pot with a 10k resistor ..both grounded ..so it's basically like the scratchy pot 5k but it has a little better taper to it.

[roberto]

You are right, it wasn't clear. I meant:
"The other point is that the scratchy solution is linear in its range, while new ones aren't (linear) with linear pots."

For sure log pots would be worse, that's why I wrote rev log.
And I totally agree that most of the amps have lin pots for the 25k.

I don't understand the 10k resistor // 10k lin pot configuration with both grounded.
I'm focusing on the "both grounded". If you mean pot and resistor in parallel, it is still scratchy.
Otherwise, if it's just the same configuration as the classical 4k7 // 25k+100n, you are doubling the feedback and keeping the same situation as the 4k7 // 5k+100n in terms of excursion of the control.

Thanks for clarifying!

[Cameron]

You are right, it wasn't clear. I meant:
"The other point is that the scratchy solution is linear in its range, while new ones aren't (linear) with linear pots."

For sure log pots would be worse, that's why I wrote rev log.
And I totally agree that most of the amps have lin pots for the 25k.

I don't understand the 10k resistor // 10k lin pot configuration with both grounded.
I'm focusing on the "both grounded". If you mean pot and resistor in parallel, it is still scratchy.
Otherwise, if it's just the same configuration as the classical 4k7 // 25k+100n, you are doubling the feedback and keeping the same situation as the 4k7 // 5k+100n in terms of excursion of the control.

Thanks for clarifying!

There is no change in feedback....I'm not sure why you think that ....but maybe I'm misunderstanding what you are saying.........most of these vertions have the same amount of feedback.....the feedback is determined by the voltage divider of the 100k(47k) and the 5k pot ...or the 4.7k resistor...nothing else....the 25k pot only effects the amount of presence added to the feedback...not the amount of feedback itself......

its not like the 4.7k//25k version.....again... in that configuration the 25k pot is not grounded.......10k pot with a 10k resistor parallel with it will make the pot a 5k ...so yes... it the same as the 5k scratchy version....but doing it this way gives it a better taper...from my experience

You can use resistors to change the taper of pots in different ways....there are some good articles online about why and how....

[roberto]

No need to read the lin2logpot links, thanks.

I thought you were using like the modern 4k5 // 10k+100n. That's why I said the feedback would have been double.
I thought it because I didn't thought you place the 10k resistor and the 10k pot in parallel.

Got it now.
Thanks.

[martin manning]

Adding shunt resistors to a pot to customize the taper is less satisfactory than using a log pot. Reverse log pots are less common than standard, but there is no reason a standard log pot can't be hooked up backwards (an "Absence" control) and be very effective at speaking the effect out over the pot rotation in both the vintage and the modern circuits.

[teemuk]

There is another detail in selecting proper potentiometer taper for NFB-based controls, which is that we can't assume the amplifier is driving a purely resistive load.

In essence, the control (frequency selectively) varies magnitude of NFB, adjusting effective gain of the amplifier between open loop and closed loop extremes. We can quite easily calculate open and closed loop gains to a nominal resistive load and from that determine range of the control in dB. Then calculating in what "increments" the control works in adjusting the response allows estimating what type of taper will fit the funtion best. If increments in gain seem to have logarithmic characteristics a linear control is likely right choice for matching those to logarithmic nature of our hearing. A logarithmic potentiometer is in turn chosen for linear incremental characteristics.

Simple so far.

But a loudspeaker is not a resistive load, it's a reactive load. The load impedance increases towards high frequencies due to inductivity of the voice coil, and, due to virtual electromechanic circuit created by the diaphgram, it peaks at speaker's resonant frequency.

The effects of varying load impedance are more or less negligible as long as the amplifier has moderately low output impedance (e.g. a generic tube amplifier in fully closed-loop mode of operation). If effective output Z is reasonably low voltage at load terminals is not affected considerably due to voltage divider formed by output Z and load Z.

However, when we hook such reactive load to an amplifier with high output Z, (e.g. a generic tube amplifier in open loop mode or with reduced negative feedback), we notice that the high output Z starts to have an effect to voltage at load terminals. Basically, the higher the load impedance, the less attenuation output impedance of the amp introduces to the signal. The poorly damped reactive load therefore introduces its own equalising effect to the frequency response of the power amp: The upper higher frequencies and frequencies around resonant frequency of the loudspeaker are boosted. This created response is unique with each unique speaker. Admitted there are wealth of common characteristics.

Not co-incidentally common NFB-based tone controls are also tuned around those characteristics: "Presence" controls upper higher frequencies (at which voice coil impedance increases) and "Resonance", like name implies, controls lower frequency range that likely covers the speaker's resonant frequency. In addition to introducing their effect to overall frequency response of the amp these controls also affect damping of the loudspeaker at its most critical frequencies where insufficient damping means operation is somewhat unpredictable.

----

So where all of this leads? To this issue: In essence we can deduce that amplifier's frequency response in closed loop mode is rather "flat". This reference point, however, is only the minimum gain ratio of the amplifier.
We also know that by removing or reducing NFB the closed loop gain figure shifts towards open loop gain figure. And if both figures are known we can calculate the range of the control in dB and the increments to response it introduces at different dial settings. But what exactly is the open loop gain?

We sure know it's not a "fixed" value with reactive loads so we can't use open loop gain figure at nominal rated impedance as a valid reference. If OLG is, say, 40 dB at speaker's nominal impedance OLG can still perfectly be something like 46 dB at upper higher frequencies or at resonant frequency of the speaker. In fact, as acknowledged, we can draw OLG as a frequency response graph, and it would portray distinct boost around speaker's resonant frequency and towards higher frequencies. In addition to to some fixed-value difference between OLG and CLG, we also need to acknowledge the difference introduced by the fact that OLG is not a fixed value but interacts with load impedance.

Basically we can make few basic assumptions and estimations:
- Open loop gain at effective frequencies of "Resonance" or "Presence" controls is likely considerably higher than open loop gain at frequencies where speaker portray's its nominal impedance. When dimensioning these controls we may wish to calculate open loop gain at load impedance that more realistically corresponds to speaker's impedance at said frequencies of interest. (The actual load impedance may be several dozens of ohms at frequencies the controls operate at).
- The "shelving" -type curve dopes not portray response of these controls accurately. You get shelving response only when both open and closed loop modes have constant, fixed values of gain. Which they don't with a reactive speker load. As determined, the range of control is OLG-CLG, and if OLG varies with impedance, producing peaks in gain at certain frequencies, those peaks will also reflect to overall response as well.

In essence, due to open loop gain variance with reactive speaker loads the NFB-based tone controls likely have more range than we expect them to have, and increments to boost at different dial settings are different than they are in idealized "shelving" -type operation.