Pot tapers for filters

[MrStab]

hi guys,

this is partly me asking stuff for my own benefit, and partly to initiate a discussion on the topic.

as you probably know, reverse-log pots often make for the most satisfying response for filters. in my experience, though, there's one major drawback: it's much harder to mark the different frequency points around the knob consistently, as there's some variation between manufacturers (and even individual units) when compared to linear tapers.

so i'm just curious about what you guys and "the industry" think about all this generally. i'm considering switching to linear pots for my EQs, primarily for the marking reasons, but also for the vastly-increased availability. do you find linear pots allow for better or worse fine-tuning? does tolerance limit any accuracy benefits? any drawbacks or advantages i haven't mentioned? it's mostly on wide bandwidths that i remember linear pots messing with "pot feel", personally.

btw, about tracking on dual rev-log pots: IME this isn't really a problem for guitar work, as any drift in the Q is always well within reason. though the variation between wafers can be a bit extreme, depending on the source. if you're building for yourself, i'd recommend rev-logs.

cheers!

[ElectricDruid]

I think you're only considering the pot taper when there's possibly/probably more to it than that. "Depends what the pot is controlling", in short.

If you've got a linear pot as a potential divider feeding a CV to a V/Oct filter, you'd get an exponential response of frequency/position. But you could equally well see that as a *linear* relationship of musical pitch/position.

Likewise, you might use a audio/log pot with a linear circuit to give it the required musical exponential character, or in other cases, use a reverse-log to give a more linear character to an exponential response. I've seen this done with filter resonance controls, where the approach to oscillation can be rather abrupt, and a reverse log response can make the useful zone much wider.

So, it depends what you're doing and how you're doing it. Put another way, I think there are some background assumptions here that you haven't stated. "Filters" is an extremely broad topic, and even "EQs" is pretty big, and can certainly be done any number of different ways. What have you got in mind specifically?

HTH,
Tom

[PRR]

If you aim for less than 10:1 range, linear may be OK, though cramped at one end.

If you aim for over 30:1 range, you are forced to a LOG, and reverse is the audio standard for frequency marks. (Oddly, gas stove knobs work the other way...)

[thermionix]

some variation between manufacturers (and even individual units) when compared to linear tapers.

Pots have a tolerance for overall value (often quite HUGE) but also one for taper...conformity.  I can see how it might be easier to manufacture more consistent linear tapers than others.

[MrStab]

I think you're only considering the pot taper when there's possibly/probably more to it than that.

it is a pretty broad title now you mention it. in my OP i was thinking about basic active filters of 1 or 2 poles, but i'm all for expanding the discussion into different types of circuit & objective. when CV's are required, there's definitely more to the response than just a resistance-cap relationship. i didn't think of mentioning that at all.

If you aim for less than 10:1 range, linear may be OK, though cramped at one end.

i've tried 16:1 on SVFs and passive RC's lately, and it's a lot smoother than i'd convinced myself it'd be. 32:1 is a no. my stove knobs go clockwise! and what's weird is i think we're in the same hemisphere

but also one for taper...conformity.

i've only ever noticed tolerance as a variable like that on pot datasheets. is this something they wouldn't put in the pot product code, but further down the datasheets?

[ElectricDruid]

I think you're only considering the pot taper when there's possibly/probably more to it than that.

it is a pretty broad title now you mention it. in my OP i was thinking about basic active filters of 1 or 2 poles, but i'm all for expanding the discussion into different types of circuit & objective. when CV's are required, there's definitely more to the response than just a resistance-cap relationship. i didn't think of mentioning that at all.

Ah, ok. Sounds like I was over-complicating things then!

i've only ever noticed tolerance as a variable like that on pot datasheets. is this something they wouldn't put in the pot product code, but further down the data sheets?

Certainly "further down the data sheets" is right. On this one, it's the last thing:

http://datasheet.octopart.com/RV16B-10-15R1-B54-Alpha-Taiwan-datasheet-7275199.pdf

The information isn't very clear or very helpful either. The ranges look pretty wide to me, and there aren't a lot of boxes filled in, so I think they give themselves plenty of room for manoeuvre on the taper.

HTH,
Tom

[PRR]

> my stove knobs go clockwise!

Some of mine go both ways. Confusing.

The plain old gas stove we took out, counter-clockwise was bigger fire.

[MrStab]

here's one: are there any small frequency ranges in particular which you guys would say, because of their greatest influence on guitar or bass tone, a linear pot would respond much better? where the same ratio would feel less-responsive elsewhere on the spectrum? i'm thinking the lows to mids would respond better, based on foggy memory and understanding.

[pinkjimiphoton]

i've found that you can use little trimmers as variable resistors between the wiper and whatever lug you need to even out...
i had a 500k reverse taper pot in a treadle control. unfortunately, the pot mounts backwards from how it would in a crybaby.. pointing the opposite direction completely. so now it was a normal linear pot, and all the fine adjustment was all bunched up at the end of the pot.
so i decided i'd go with a reverse foot... heel down for fast, toe up for slow. works better for my purposes, but with size 15EEE boots its hard to get the heel all the way down for the phaser on stun thang. and the toe down was just too slow to be useful.
so i took a 100k trimmer and stuck it between the wiper and the slow side of the pot,
and monkeyed with the trimmer until it made the pot come alive as i wanted to with a good useable sweep range with the treadle.
i'm gonna go downstairs to my dungeon and see if i can reverse it back to toe down for high speed, and use the trimmer to "open up" the bunched up area. it worked great on the lower range of the pot, "condensing" the sweep more... not sure if i can "expand" the sweep on the high range yet. doubtful but hopeful.

[blackieNYC]

Then of course there are the possibilities one gets with parallel fixed resistors.  Geofex secret life of pots.   Make like a Hollywood starlet and customize the curves.

[pinkjimiphoton]

well, i found it worked good on condensing the range. not good on expanding it.

[MrStab]

i've thought about trimmers to compensate for pot tolerance where stopper resistance is needed, but never got round to it. when coming up with any original footpedal circuit, given the limited travel length, i can see trimmers being almost essential.

one thing that bugged me a while back was the shortage of rev-log guitar pots (lefty pots?) for use as right-handed tone pots. that'd be the ideal taper and ideal pot rotation, but finding one for cheap-ish was too much hassle. i never use my tone pot anyway, i suppose.

in addition to that article, there's an awesome Excel document somewhere on the forum that helps with pot-resistor paralleling, Blackie - i can't find it atm but i think the file name is potcalc1.xls . there are sometimes more caveats than can first appear when doing that though, depending on the circuit & way the pot's used.

tangent: i deleted part of my OP because all of a sudden i was insanely confused by a weird interaction i've noticed in state variable filters, which real tests, sims and (my poor) maths all confirmed: certain resistance relationships in the voltage-divider-as-frequency-pot type of SVF EQ can force a linear pot's taper to respond totally differently, despite the start & end of the frequency range being as-desired. a few common schematics online have such values! i found that if the frequency pot is around 10x greater than the max frequency resistors, it behaves more like a reverse-log pot. if either those or the stopper resistors are increased by 10x, though, the taper becomes linear again. whether it's a paralleling thing or a loading thing, i haven't figured out yet, but this has put me off of using anything greater than 10k pots in SVFs so as to avoid higher resistors~noise. hmm!