J M Fahey 6/18/2017 11:55 AM
Simulated Log/Audio potentiometers: a bad idea
This is to show why the popular "Log pot simulation" using an easier to find Linear Pot and a lower value parallel resistor is actually a bad idea in 90% of cases, and just acceptable on 10% (if that much).

This way itīs easier to understand:
[IMG]http://i.imgur.com/p9XwtS1.png[/IMG]

Suppose a pickup loaded by a 1M potentiometer.

We use a true 1M Log pot and a 1M Lin one with 100k strapped from wiper to ground.
We calculate 3 positions to make it easy and in any case are more than enough to show the heavy loading effect and wildly varying load impedance shown by the fake/simulated one.
A terrible idea I might add, sadly wildly popular because it looks "easy" .
Iīll simplify some values to closest round number, the idea here is not to get numbers accurate to 6 decimal places but to understand the concept of whatīs happening.

Ok, letīs start with wiper on 0

True Log shows pickup load=1M , signal at wiper 0
Fake Log shows pickup load=1M , signal at wiper 0

Hey!!! it looks good!!!!

Repeat with wiper on 5


True Log shows pickup load=1M , signal at wiper 11%
Fake Log shows pickup load=590k , signal at wiper 18%

Mmmmhhhh, load has varied significantly, maybe it wonīt hurt pickup sound that much, now if pot were after a small coupling cap (as in VOX and Marshall) I would be losing almost the full lowest octave.
And Log simulation is not that good either, we are almost 6dB above what a true Log would offer.

Repeat with wiper on 10


True Log shows pickup load=1M , signal at wiper 100%
Fake Log shows pickup load=90k , signal at wiper 100% .... IF pickup or earlier stage can happily drive 100k , 10X smaller than expected.

If it were a volume control after a small coupling cap, a common trick in respected Guitar amps, equalization would be a mess.

Same if following a tone control stack.

If following a triode stage, you would lose at least 6dB gain, because plate load resistor (typically 100k to 220k) would now be in parallel with 90k.

If following a pentode stage, which has high internal impedance and typically drives a high value load (220k and 470k are common values) effect will be a mess, easily losing 15dB signal.

Not enough time now to show the effect of a fake Log Pot used as variable NFB gain control in an Op Amp circuit, but believe me itīs even worse than this.

But is it absolutely useless then?
No, there is *one* very limited case where it works, sort of, at least as a Saturday afternoon stopgap until shops open on Monday and you can buy a proper Log one: IF you use it as a passive Volume control after a low impedance driving stage, say an Op Amp, which can easily drive the much reduced impedance shown on 10 , then itīs acceptable.
Not accurate but human ear is so imprecise that it wonīt complain.

Not even those who "can hear the influence of power cable Oxygen content" or whatever
 
d95err 6/20/2017 10:55 AM
For a gain control, this behavior can be desirable. Fat for clean sounds, tight distortion when you turn it up.

Incidentally, this is the opposite of the common "bright" cap used on so many amps, which I find useless (harsh and thin for low gain, farty and muddy for high gain).
 
pdf64 6/20/2017 11:59 AM
For use as a channel gain control in a marshall type amp (ie not immediately after the tone stack) that's a fair point; in conjunction with the right coupling cap value, the loading effect can help to roll off unwanted low end as the control is advanced.
Though it seems a rather different thing, and doesn't detract from, the thrust of JM's argument.
And I think it seems to work better when used with a audio / log pot anyway
 
R.G. 6/20/2017 2:36 PM
There are several asterisks needed here. As I noted in The Secret Life of Pots way back when, you're exactly correct, the loading is very variable. This is a problem in the case where the source impedance is not constant with frequency. That's almost never a problem - except when the source is a bare guitar pickup. Then it's a real problem, for all the reasons you note. There are also issues with sources having impedances of near the impedance of the pot, as the varying load of the faked pot causes volume variations on top of the taper variation.

But it's important not to throw the baby out with the bathwater. There are many instances where you have a fixed-impedance source and a high impedance load to be driven where you just want a voltage-divider pot with a log taper. In that case, the tapering resistor approach works in an exemplary manner. And it gives you the freedom to tinker your taper, which you don't get every day.

Then there's the sorry state of log pots in general. Most modern pots are simply two sections of linear taper in a crude approximation of an actual log taper. I'm not aware of any smoothly varying taper log pots at all. In that view, the log nature of a faked log pot is much better than a commercial real pot.

Here we are with the devil in the details again!!
 
loudthud 6/20/2017 5:34 PM
The Log taper tracking is very poor at the low end of the pot in the "loaded linear" example above. What works much better is using a dual linear pot where the output of the first section feeds the second section. The load variation seen by the source is only two to one and there is much more attenuation at the low end of rotation.
 
Mike Sulzer 6/21/2017 12:29 PM
Quote Originally Posted by loudthud View Post
The Log taper tracking is very poor at the low end of the pot in the "loaded linear" example above. What works much better is using a dual linear pot where the output of the first section feeds the second section. The load variation seen by the source is only two to one and there is much more attenuation at the low end of rotation.
Yes that works well, and not so hard to analyze. If we define a rotation variable "a" which has a value of zero when the sliders are at the bottom and one when they are at the top, the response ls shown in the attached plot.

[ATTACH=CONFIG]43864[/ATTACH]