Eq Pedal Weird Response

[ElectricDruid]

@ElectricDruid It describes the theory of operation, wherein the pots "pan" between the inverting and non-inverting inputs. So their value doesn't factor in, within reason. Gain is defined by the gyrators' series R and the input and feedback R of the "mixing" stage. This article for example shows the math: https://www.ti.com/lit/an/slyt134/slyt134.pdf

Thanks, that TI article is very complete. I don't think I've seen that circuit with the equations for it before.

It mentions that the pot value is *not* completely unimportant, even if it doesn't affect the overall boost/cut value. Larger values produce the "weird taper" effect we're talking about worse than smaller values, and I've seen this effect in a simulation. It recommends 10K as the best compromise. Certainly the weird taper effect is worse with 50K or 100K than with 10K.
 

[Rob Strand]

Thanks, that TI article is very complete. I don't think I've seen that circuit with the equations for it before.

It mentions that the pot value is *not* completely unimportant, even if it doesn't affect the overall boost/cut value. Larger values produce the "weird taper" effect we're talking about worse than smaller values, and I've seen this effect in a simulation. It recommends 10K as the best compromise. Certainly the weird taper effect is worse with 50K or 100K than with 10K.

There's no way it's complete.   It's a good working recipe.

Think about what you expect from an equalizer:
- We start with the maximum boost/cut and the pots set for max/min.
  Out of everything that will be least affected by the pot values.
  However it's not entirely immune to the pot values when other bands are present.

- Now set the pot to 25% or 75%  (presume a G taper in order for it to have a chance of working.)
  What we expect is the boost/cut to be half of maximum.
  That's only going to be true for one pot value.
  That pot value might not be practical or might impact other aspects of performance.

It's not the pot value in absolute terms that is important parameter.  It is the ratio of the pot value relative to the input/feedback resistor.    The TI example uses 10k pots and 3.3k input feedback resistor; which just happen to be what Boss uses.   If you do survey of equalizer schematics you will find this ratio varies a lot (not all equalizers make a good choice).   10k/3.3k is a fairly low ratio.

The maximum boost/cut is related ratio of the feedback resistor to the gyrator resistor.   In simple terms max gain = (1 + Rf/Rs) where Rf = feedback resistor, Rs = gyrator series resistance.   However with multiple bands  the actual gain is much less.   The actual amount of gain achieved depends on the number of bands, the Q of the sections and the ratio of the feedback resistor to the pot values - a whole lot of interactions.  Off hand the National Semiconductor Audio/Radio handbook had some tweak formulas to help get the maximum boost/cut a little closer.   Somewhere I've got a few formulas of my own.   The truth is the interactions are complex and beyond simple formulas.
[https://archive.org/details/bitsavers_nationaldaAudioRadioHandbook_17034677/page/n72/mode/1up]

The conclusion in the thread which preceded this one is it's not too hard to design a simple EQ for a pedal with one to three bands but a full graphic EQ requires quite a bit of work. (Something to think about is we think of single bands boosted or cut but what about multiple bands - a whole new set of headaches which really only gets fixed tinkering with the Q of the bands.)
As far as changing the pots goes.   The fact we don't know what the feedback resistor is on the current unit is why I said earlier we don't even know if the 50k pot value is the best choice.   The taper change will help make it sort of work.
This paper solves some of the problem by side-stepping the whole gyrator topology,
https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=3292c31cdc4df5fdb09f55800c767be249930e89

[antonis]

I don't think I've seen that circuit with the equations for it before.

Check in article's Reference 2..
(Audio/Radio Handbook, National Semiconductor (1980))

[ElectricDruid]

I don't think I've seen that circuit with the equations for it before.

Check in article's Reference 2..
(Audio/Radio Handbook, National Semiconductor (1980))

I said *I* hadn't seen them before, not that they weren't *there to be seen* before!!

Maybe some dude worked them out on a vacuum tube EQ circuit in 1950, but if they did, I didn't know about it!

[Rob Strand]

Here's a way to approach the design.  Notice I just started with some general ideas on how to make reasonable choices for the parts.  Then from that you get a reasonable ratio for the feedback resistor (Rf) to pot value (Rp).

If you compare my values, the TI values and the Boss values you will notice the TI values corresponds to low Q's, Q ~ 1.  The Boss values corresponds to higher Q's, Q ~ 3.5.  Also the TI unit as about +12dB and Boss is +15dB.  My Q values is 2.3 and +13.4dB.

The lower Q makes the full-boost ripple lower but will screw up the half maximum boost position.

Is mine better?  No.  All it does is match a loose set of criteria.
[Basically these types of EQ have inherent flaws.  You have to find a balance between conflicting requirements.
 A good deal of EQ's remove the conflicts by not using this simple topology.   One intermediate solution
 is to make two EQ's in cascade and assign alternate bands to each EQ.]

When you have a guitar/bass equalizer you are more concerned with the sound of an individual band than how the overall bands combine.  Also, an equalizer with a high Q and high boost/cut will have a lower Q when set to a lower amount of boost/cut.  That means you can't use the circuit Q to judge the sound.  This is how these gyrator work, the Q is roughly proportional to the linear gain (not dB gain).

You rarely use all bands on full boost/cut.  So a making design choices based on maximum boost/cut are unlikely to be realistic in practice.  You might put more weight on the 6dB to 10dB region.

Schematic and text for "a" method:


Single Band:


Two Bands:

[Baran Ismen]

@ElectricDruid It describes the theory of operation, wherein the pots "pan" between the inverting and non-inverting inputs. So their value doesn't factor in, within reason. Gain is defined by the gyrators' series R and the input and feedback R of the "mixing" stage. This article for example shows the math: https://www.ti.com/lit/an/slyt134/slyt134.pdf

Thanks, that TI article is very complete. I don't think I've seen that circuit with the equations for it before.

It mentions that the pot value is *not* completely unimportant, even if it doesn't affect the overall boost/cut value. Larger values produce the "weird taper" effect we're talking about worse than smaller values, and I've seen this effect in a simulation. It recommends 10K as the best compromise. Certainly the weird taper effect is worse with 50K or 100K than with 10K.
 

Wait a second, i have already ordered the w50k's what if the reason is the 50k itself?

Can I give this a try like the old style, maybe? Connecting 6.8k between 1-2 and 2-3 lugs of each slider, so that it'll become 10k and also W taper?

[Rob Strand]

Wait a second, i have already ordered the w50k's what if the reason is the 50k itself?

Can I give this a try like the old style, maybe? Connecting 6.8k between 1-2 and 2-3 lugs of each slider, so that it'll become 10k and also W taper?

The taper is on the top of the list of problems.  The issue will improve with the new taper pots.

The question is could you do better using a different value pot (and G/W taper).

The parallel resistor mod does not work.  It only really works for voltage dividers.   The graphic equalizer is using the pot more as a variable series resistance.   However, at the boost end it also needs to disconnect from the cut end - that's where you need the overall pot resistance as well.  The G/W taper is only 6% to 7% of the pot value at 25%/75% positions, around 3.3k. 

[Baran Ismen]

Wait a second, i have already ordered the w50k's what if the reason is the 50k itself?

Can I give this a try like the old style, maybe? Connecting 6.8k between 1-2 and 2-3 lugs of each slider, so that it'll become 10k and also W taper?

The taper is the top of the list problem.  The issue will improve with the new taper pots.

The question is could you do better using a different value pot (and G/W taper).

The parallel resistor mod does not work.  It only really works for voltage dividers.   The graphic equalizer is using the pot more as a variable series resistance.   However, at the boost end it also needs to disconnect from the cut end - that's where you need the overall pot resistance as well.  The G/W taper is only 6% to 7% of the pot value at 25%/75% positions, around 3.3k. 

Well, I got no other option but to wait for arrival of the sliders and try them, then.

[Baran Ismen]

New sliders will arrive tomorrow Hopefully, but i noticed something that they have 4 lugs, whereas there are 3 on the board. Placements are the same, but theres 1 extra on one side. I think 2 of them share a common ground/wiper and like 1 2 2 3 even though they are labeled as 1 2 3 4, but i'll check it once arrived. What do you think it would be?

[ElectricDruid]

What do you think it would be?

It could be a centre connection for the track. You see that sometimes, since it can be used to ensure you have "zero" in the middle. Or it could simply be a duplicate, which can help make layout easier.

[Rob Strand]

It could be a centre connection for the track. You see that sometimes, since it can be used to ensure you have "zero" in the middle. Or it could simply be a duplicate, which can help make layout easier.

Typically the two "middle" connections are joined. Occasionally on old amps with sliders you see some wires connecting to the top and some connecting to the bottom.   Not a bad thing because the connections for large sized sliders are sooo far apart.

[Baran Ismen]

Changed the pots with the new ones, and response is much better now.

[ElectricDruid]

Changed the pots with the new ones, and response is much better now.

Hey, good work!