2nd order filter and pot resolution

So, I've been messing around with a simple 2nd order active HPF.

Below is the schematic...



I'm using a dual antilog pot (50k 47k and 100k) in order to gain that 3.8dB of headroom and although everything seems to be working fine I have a problem with the resolution.

The pot is much more "sensitive" on the first half of its rotation (0% to 50%). I included a graph with different positions of the pot (approx. equally deggres of rotation). Green is the the 1st half of rotation, Red is approx. the centre and blue is the 2nd half.

It really puzzles me as I would expect it to behave with a much smoother resolution.

Any ideas? Is this normal?

Here's the schematic.

Hello

What are your pot ?
"simple" carbon track ?
or advanced resistive plastic with ladder print ? (which can have issue in log curve if mounted as variable resistor)

side note, you have a bump at 50 and 100Hz check your PSU...  :

Best
Zam

The pot is Conductive polymer (plastic) .It's a bit unclear to me whether is plastic conductive or not.

In the middle, the pot reads around 5K (on the 47k deck) and 90k (on the 100k deck) so I believe that's correct. I haven't measured the resistance on the first half of the rotation but I will do so.

Is there any alternative way of connecting the pot in order to avoid this issue?

PS> I'm aware of the 50Hz bump. Everything's on a breadboard so that might be the issue. I'll investigate further but at the moment my priority is to solve the resolution problem. Really annoying...

warpie said:

The pot is Conductive polymer (plastic) .It's a bit unclear to me whether is plastic conductive or not.

In the middle, the pot reads around 5K (on the 47k deck) and 90k (on the 100k deck) so I believe that's correct. I haven't measured the resistance on the first half of the rotation but I will do so.

Is there any alternative way of connecting the pot in order to avoid this issue?

PS> I'm aware of the 50Hz bump. Everything's on a breadboard so that might be the issue. I'll investigate further but at the moment my priority is to solve the resolution problem. Really annoying...

Click to expand...

Ok so conductive plastic but not a ladder print regarding your measure.

Any brand/tech spec ?

For what I see , only the 40/50 Hz area is over sensitive (let say second quarter of your pot) all other have more or less 10Hz steps at -10 dB, considering you don't turn the pot with accurate angle, as the pot curve tolerance, your not that bad ?

Best
Zam

Well, it's not ideal either. I think I'd rather have this "bad" resolution towards the higher frequencies. I mean, 40-50-60hz is where HPF are mostly used.

The pots are omeg. Not sure what other specs would be useful. 

I might have to try a 50k dual (antilog) pot and live with the 4dB less headroom although I'd much rather using the omeg...

Shame...

warpie said:

In the middle, the pot reads around 5K (on the 47k deck) and 90k (on the 100k deck) so I believe that's correct.

Click to expand...

Please ignore me if I´m on the wrong track, but doesn´t that indicate you have a log/antilog-pot, instead of a dual antilog?

L´Andratté said:

Please ignore me if I´m on the wrong track, but doesn´t that indicate you have a log/antilog-pot, instead of a dual antilog?

Click to expand...

You're absolutely right. My bad . 5K and 10K I wanted to write. I wish that was the problem...

warpie said:

In the middle, the pot reads around 5K (on the 47k deck) and 90k (on the 100k deck)

Click to expand...

Either you have a log/anti-log pot, which is extremely unlikely, unless you have it custom-ordered, or you are measuring the wrong side of the 100k deck. Both decks must be connected so their resistances decrease when turning the pot CW.

abbey road d enfer said:

Either you have a log/anti-log pot, which is extremely unlikely, unless you have it custom-ordered, or you are measuring the wrong side of the 100k deck. Both decks must be connected so their resistances decrease when turning the pot CW.

Click to expand...

Yes, 5K and 10K is what I meant to write.

Do you need R60?

Make R65 = 2x R64, or measure the pot and use the actual ratio.  It's close as is but why have an additional source of error.

Part of the frequency bunching problem is you are using 10% log pots. If you go with something higher (15%, 20% etc) you can get closer to what you want.  To avoid buying new pots you can slug the ones you have.  For example put 100k in parallel with the 100k pot,  put 50k in parallel with the 50k pot, and then scale the other caps and resistors appropriately.

warpie said:

Yes, 5K and 10K is what I meant to write.

Click to expand...

So your pots are 10% RevLog type. They are optimized for an actual 100:1 sweep, which does not offer the finest resolution.

Thank you gents, I'll try your suggestions.

I just realised though that in the OP I wrote 50k and 100k while it's 47k and 100k instead.  I edited it. Also, they are 20% tolerance as far as I'm concerned.

warpie said:

Thank you gents, I'll try your suggestions.

I just realised though that in the OP I wrote 50k and 100k while it's 47k and 100k instead.  I edited it. Also, they are 20% tolerance as far as I'm concerned.

Click to expand...

It's not tolerance that is the concern; it's the taper. A 10% taper indicates that the resistance at mid-rotation is 10% of the nominal.
There are all sorts of Log or RevLog tapers, 5%, 10%, 20%...

47k vs. 50k is irrelevant considering the possible variations of nominal.

abbey road d enfer said:

It's not tolerance that is the concern; it's the taper. A 10% taper indicates that the resistance at mid-rotation is 10% of the nominal.
There are all sorts of Log or RevLog tapers, 5%, 10%, 20%...

Click to expand...

OK I see. So the only way to find the percentage at the mid-rotation is by testing the actual pot? I don't remember seeing it on any data sheet.

warpie said:

OK I see. So the only way to find the percentage at the mid-rotation is by testing the actual pot? I don't remember seeing it on any data sheet.

Click to expand...

First hit of googling "potentiometer taper".
http://www.potentiometers.com/potcomFAQ.cfm?FAQID=29

john12ax7 said:

Do you need R60?

Make R65 = 2x R64, or measure the pot and use the actual ratio.  It's close as is but why have an additional source of error.

Part of the frequency bunching problem is you are using 10% log pots. If you go with something higher (15%, 20% etc) you can get closer to what you want.  To avoid buying new pots you can slug the ones you have.  For example put 100k in parallel with the 100k pot,  put 50k in parallel with the 50k pot, and then scale the other caps and resistors appropriately.

Click to expand...

Thanks for the suggestion .I tried it  but unfortunately with not much luck. The result is pretty similar.

R64= 2k32
R65= 4k7
C31/C31 =220nF

Did the middle position red line shift at all? I would expect at least a modest shift from around 110 to 80 Hz. You could go even further with it,  for example slug the 50k pot with 25k, and the 100k with 50k.

Another thing would be to measure the pot itself at the same points of rotation to make sure it's doing what it's supposed to be doing.

john12ax7 said:

Did the middle position red line shift at all?

Click to expand...

It did move a little bit indeed. Can't remember the exact value though.

john12ax7 said:

You could go even further with it,  for example slug the 50k pot with 25k, and the 100k with 50k.

Click to expand...

In fact, I went as far as using 12k for the 47k and 25k for the 100k. It shifted even further.

The problem is that in all cases that "big gap" is present but on a different frequency range of course, which is expected I guess (no matter how much you shift it).

I wonder if that's the case with all (or most) antilog pots and if so, how the commercial designs get away with it.

warpie said:

I wonder if that's the case with all (or most) antilog pots and if so, how the commercial designs get away with it.

Click to expand...

As I wrote earlier, standard 10% Log pots are optimized for a 1:100 ratio. Potentiometeter manufacturers offer a number of options for designers who are ready to order 1000's of pieces.
Now there are other options, like using linear pots with law-steering resistors, but they necessitate a different circuit topology (State variable filter).