Versatile Tone Control

Main Index

Projects Index

Tone controls are almost always a compromise.  The 'traditional' Baxandall tone control circuit used to be common for hi-fi, and a passive tone control 'stack' is most common for guitar and bass.  I've shown a few different tone control circuits in various projects, and they have also been covered in the article Equalisers, The Various Types And How They Work.  The basis of this project is shown in Figure 17 of the article, but as a project it needs more explanation.

The circuit provides a wide-range tone control system that lets you tailor the sound to get the response you want.  This is easily set up as an 'adjunct' to an existing preamp, connected between the preamp and power amp (or electronic crossover network).  While I've shown 10k frequency pots, these can be increased to get greater range.  With the other parts as shown below, I probably wouldn't use more than 50k, as that may provide too much overlap.

None of the frequency determining values is critical, and you can modify them to get the results you're after.  The easiest way to get greater range is to increase VR1 and VR2 to either 20k or 50k, which will provide more range than you're likely to need.  Like many ESP projects, this is intended to give you a starting point, and it's easy to change to get exactly what you want.

The topology chosen is the same as that for a 'constant Q' graphic equaliser - see Project 75.  This is more complex than the traditional feedback tone control circuit, but it provides flexibility that's not available with other circuit arrangements.  The feedback paths are comparatively complex, and that makes analysis harder, but it makes the realisation of filters far less restrictive than other 'simpler' tone control circuits.

It's worth making the point that although conventional tone controls are nominally 6dB/ octave, this is only approached when maximum boost or cut is applied.  Even then, it's unusual to get more than 4dB/ octave (around 3.5dB/ octave is usually the maximum slope).  While this may seem somewhat limiting, in practice it usually works very well for most listeners because it's not meant to be 'radical', but to change the response of a system to the listener's liking.  The final section in this article ('Taking it to Extremes') shows the use of 12dB/ octave filters, but even there the vast majority of filter slopes will be not much greater than 6dB/ octave.

The circuit is almost identical to that shown in the 'EQ' article, but it's been updated to include R9, essential to prevent instability when the output is connected to a shielded cable.  The input impedance is 10k, somewhat lower than expected.  It can be increased, but at the expense of noise.  The pots are also made 10k for the same reason.  The opamp isn't critical, but a TL072 will work well in both locations.  This is one of the few circuits where I don't really recommend an NE5532, because the DC offset may cause the pots to become noisy.  For 'ultra-high' performance, the LM4562 is very hard to beat, or the OPA2134 is also a good candidate (albeit at some cost penalty).

Note that the power supply connections and bypass capacitors are not shown.  Each opamp should be bypassed between pins 4 and 8 with a 100nF multilayer ceramic cap, and each supply needs a 10µF cap to ground.  If these are omitted, most opamps will oscillate.  The input and output caps are shown as electrolytics, but you can also use bipolar types, or even film caps.  The minimum recommended value is 2.2µF, which will create a -3dB frequency of 7.2Hz for C3.  The -3dB frequency for C6 depends on the load impedance.

The complete circuit is non-inverting, because both input and output stages invert the signal.  You might think you could use the non-inverting input of U1A as the input if you wanted to get a high impedance input, but the circuit will then be inverting.  While there's little to indicate that this is audible, most people prefer non-inverting circuits.  If you do happen to need a higher input impedance or more gain (it's unity as shown when the boost/ cut pots are centred) then add another opamp at the input, configured for the gain needed (see Fig. 3).

Figure 1 - Tone Control Schematic (One Channel)

VR1 changes the bass frequency from 200Hz to 740Hz at the ±3dB point.  VR2 does the same for the treble, from 460Hz to 1.4kHz, again at the ±3dB frequencies with full boost or cut.  Everything can be changed to suit what you wish to achieve.  The capacitor and resistor values shown above cover the range quite well, but for some applications there can be good reasons to make changes.  The bass range can be raised by using a smaller cap for C1, and the treble range can be lowered by using a bigger cap for C2 (and vice versa in each case).  For more boost and cut, simply reduce the values of R5 and R6.  Attempting less than 2.2k isn't recommended if you use TL072 opamps, as they will struggle to supply enough current.  2.2k will allow ±14dB at the frequency extremes.

The caps in series with R5 and R6 minimise DC offset.  10µF is more than enough for the treble control (R6, VR4) and while 10µF works well for the bass control, at least 47µF is recommended for minimum distortion (electrolytic caps will introduce some distortion if there's a significant voltage across them).  If you use 10µF with 3.3k as shown, the bass will be 3dB down at ~5Hz, which is in addition to bass rolloff caused by the input cap (C3).  Most people don't need response to 5Hz, so that will be fine for many constructors.  The overall response will still be good to below 15Hz.

The output of U2A is primarily bass, filtered by R3+VR1 and C1.  The frequency where the control has an effect is determined by the setting of VR1, with the highest value (13.3k) making the bass turnover frequency lower than with the lowest setting (R3, 3.3k).  This is either added or subtracted from the input signal by U1B, depending on the setting of VR3.  When VR3 is centred, nothing is added or taken away, so there is no bass boost or cut.  U3B does the same with treble.  The 3dB frequency is determined for both by the combined value of R3+VR1/ R4+VR2 and the associated capacitor (C1 for bass, C2 for treble).  The situation is complicated by the feedback path from the tone controls back to the inverting input of U1A, and this is essential to allow bass and treble cut.

To get a wider frequency range for both bass and treble, use higher value pots for VR1 (bass) and VR2 (treble).  If you use 20k pots, it would probably be a good idea to reduce the capacitance for both filters.  For example, with 20k for the bass control, the minimum ±3dB frequency is 175Hz with C1 at 150nF.  The maximum frequency (±3dB) is 1.2kHz.  The treble range is from 1.5kHz to 8.8kHz with a 1.5nF cap (C2).  Note that you can't use the filter's 3dB frequencies because they are within a feedback loop which alters the response.  The curves shown below use the Fig. 1 values, not the alternatives.

Figure 2 - Response Curves, 10Hz to 40kHz

The 'family' of curves was taken with VR3 and VR4 incremented by 50% (3 steps from 0% to 100%), and the same for VR1/ VR2.  For VR1/ VR2, a lower resistance means a higher frequency.  I have deliberately kept as many resistor values the same as I could for ease of construction, but you can make changes to suit your particular needs.  The most likely changes will be to C1 and C2 to obtain the response you desire.

Figure 3 - High Impedance Input Gain Stage (One Channel)

A simple opamp gain stage can be added to the tone control circuit if needed, and a dual opamp handles both channels.  The input impedance is determined by R1, which can be up to 100k with a bipolar opamp (e.g. 4558, LM4562, etc.) or 1MΩ if you use a JFET input opamp (e.g. TL072, OPA2134, etc.).  The gain is determined by the ratio of R3 and R4, and is ×2 (6dB) as shown.  With low to medium gain settings (up to 10dB or so), there's no need for a capacitor in series with R4.  There will be some DC offset (opamp dependent), but it's removed by C3 in Fig. 1.  R2 is intended to prevent RF interference, and it must be installed as close to pin 3 (or pin 5) as possible.

The response of the gain stage is completely flat, with a -3dB frequency of 7Hz with C1 at 220nF and R1 as 100k.  C1 can be increased in value if preferred, with 1µF giving a -3dB frequency of 1.6Hz, well below any audible frequency.  If you don't need any gain but still require a high input impedance, simply join pins 1 and 2 (pins 7 and 6 for the other half of the opamp), which is used for the second channel.  If you have an input level of around 1V peak, the maximum gain for the input stage is ×2.  Any more will cause the tone control stage to clip on transients if maximum boost is used.  With the suggested values, the maximum boost is 12dB, or ×4.

The frequency response (along with bass and treble boost) extend to the capabilities of the opamps used, but 5Hz to 40kHz is easily achieved.  Not everyone wants such low or high frequencies, and this is especially true if the tone controls are used for musical instruments.  The solution is a high-pass and low-pass filter that can be tailored to suit your needs.  A suitable circuit is shown below, and they are standard 12dB/octave filters.

Figure 4 - High & Low-Pass Filter Stages (One Channel)

With the values shown, the response is 3dB down at 15Hz and just over 27kHz, but only 1dB down at 20Hz and 20kHz.  The high-pass filter should be in front of the tone control circuit to ensure that infrasonic frequencies are effectively removed before bass boost is applied.  It's surprisingly easy to cause overload (distortion) if the source is a bass guitar (for example) and the strings are muted by using one's hand.  These very low frequency 'events' will mainly be reduced by the high-impedance input stage (Fig 3), but that might not be enough in some cases.  Both filters are optional.

The frequency formula is ...

fo = 1 / ( 2π × √( R1 × C1 × R2 × C2 )) = 15Hz, 27.3kHz

The values shown give slightly better results than using two 47k resistors in series for R2 or two 2.7nF caps in parallel for C3.  The frequency for each filter can be reduced (or increased) if desired.  The high-pass filter can use 220nF caps (-3dB at 10Hz) or 100nF (-3dB at 23Hz).  There's no good reason to change R1 and R2.

With C3 = 6.8nF and C4 = 3.3nF, the -3dB frequency is 22.4kHz.  For guitar or bass, I suggest 15nF and 6.8nF, giving a -3dB frequency of 10.5kHz.  It can be lower if you prefer, but I leave that to the constructor.  Keep the value of C4 slightly less than half that of C3 for best results.  Note that the input must be DC coupled to the output of U1B, and the output resistor and capacitor are moved to the output of the low-pass filter.

The topology used is very flexible, and the filters can be 'normal' high and low-pass or bandpass.  An option that might come in handy for music production (as opposed to reproduction) is to use 12dB/ octave filters instead of the traditional 6dB/ octave types.  Almost without exception, tone controls have a maximum slope of 6dB/ octave because it's the most natural way to correct tonal imbalance.  During production, far more radical EQ is often used to obtain the 'sound' the producer is looking for.  Parametric EQ is common, as it can be used to affect a narrow band of frequencies.

However, parametric equalisers are not easy to use.  Most professional types have three controls for each band, being boost/ cut, frequency and Q (width of the frequency band).  These are commonly used in groups of three, generally with overlapping frequency ranges.  This arrangement allows very precise control, and most systems will include separate bass and treble controls as well.  Whether these are variable or not depends on the designer.  A ⅓-octave graphic EQ is also an option, but these are large and unwieldy as there are 31 slide pots to cover the range.

If some really radical (but otherwise 'conventional') EQ is needed, the Fig 1 circuit can be used with second-order high and low-pass filters.  The response is unlike any other tone control systems, as shown by the response graphs shown next.  This arrangement is not suitable for hi-fi systems because parts of the response are unpredictable, especially if the bass and treble frequencies are close to each other.  'Close' in this context can mean as much as 3 octaves.  There is (deliberately) no attempt to adjust the phase of the filters, as this provides the maximum 'disturbance' in the midrange area.  This should not be affected with hi-fi tone controls, but the filters described here are designed for the maximum effect, as often used for music production.

Figure 5 - 12dB Response Curves, 10Hz to 40kHz, Both Controls Boost/ Cut

The bass and treble controls are both set to the same amount of boost and cut.  The frequencies are set for maximum bass frequency (320Hz) and maximum treble frequency (2.2kHz).  Attempting to show a complete family of curves would result in a graph that's too messy to be read and understood.

Figure 6 - 12dB Response Curves, 10Hz to 40kHz, Controls Opposite Boost/ Cut

In Fig 6, the bass and treble controls are opposite, so with maximum bass boost there's maximum treble cut and vice versa.  The frequency controls are the same as used for Fig 5.  These two graphs show how radical the EQ can be.  Whether anyone thinks this is worthwhile is unknown, but it's included because it's an interesting arrangement that I've not seen elsewhere.  The ability to vary the frequency means there is a lot of scope, so you can either get the sound you want, or ruin it completely.

Figure 7 - 12dB Tone Control Schematic

Making this circuit as a stereo equaliser is difficult because you'd need 4-gang pots to set the frequency.  The filters are low-Q, but within the feedback loop the effective Q is enhanced, and that causes the dips that accompany boost, and peaks that accompany cut.  The worst-case peaks/ dips are 3.3dB with the frequencies set as shown, but this gets more dramatic when the frequencies are closer together.  The response remains flat when the boost/ cut controls are centred.  To use the circuit with guitar (etc.), I suggest the addition of the 'Bright' switch.  The values of R10 and C7 shown should be considered a starting point - they an be tweaked to get the sound you want.

The filter section are shown next.  The switches shown are optional.  Perhaps surprisingly, they don't change the frequency, and nor do they affect the maximum slope to any major degree.  When set to '6dB', they do minimise the depth of the notch and height of the peak (as seen in Fig. 5), making the controls behave like those shown in Fig. 1, but with different frequencies due to changed capacitor values.

Figure 8 - 12dB Filter Schematics

The filters weren't included in Fig 7 to keep the drawing as clear as possible.  They are perfectly ordinary sub-Bessel (Linkwitz-Riley) filters, and they are tuned with a dual-gang pot.  The frequencies may seem odd, with bass variable from 80Hz to 320Hz, and treble from 545Hz to 2.2kHz.  However, it's not just the filter's amplitude response that affects the EQ, it's also the phase shift.  It's easy to change the treble (or bass) range by changing the capacitor value.  Higher values mean lower frequencies for both filters.  It's possible to use Butterworth filters, but the response will be too radical, with very narrow peaks and dips (setting dependent) that rarely produces a sound that anyone wants.

The treble filter uses 22nF caps and the bass section uses 150nF.  These put the frequency range into the 'sweet spot' for guitar, and the radical response may work very well (depending on your expectations of course).  The primary effect is actually the notches (and/or peaks as seen in Fig 5), rather than the response of the filters themselves.  These are very pronounced with the 22nF treble caps, and change the sound quite dramatically, allowing the circuit to mimic a traditional guitar 'tone-stack'.  The difference is the ability to get flat response, something you can't get with a tone-stack.

If preferred, the frequency pots can be replaced with fixed resistors and a switch.  There are many possibilities, but I expect that between two and five frequencies for both bass and treble would be useful and easy to work with.  The resistors can be selected to give the responses you need, and can be outside the ranges available with the pots if that works for you.  This is a fairly radical tone control circuit, so expect it do do things that aren't possible with other configurations.  The filters can be switched for 12dB or 6dB operation, making it a very versatile tone control indeed.

Some readers will notice that the topology used is (somewhat) similar to that employed for constant-Q graphic equalisers (see Project 75 and/or Project 84).  This means that you can also use one or more bandpass filters (including variable types to create a parametric EQ) in place of the high and low-pass filters shown.  If these use variable resistors (pots), then the possibilities are endless, but the system will become hard to use, and will be even harder to reproduce a setting later on.  A tone control is of limited use if you find the sound you want by accident, but can't find it again quickly when you need it.  A potential solution is to use switched resistor values instead, so every setting can be reproduced exactly.

Having tested the circuit with 965Hz treble and 141Hz bass filters, I can add that it should work really well for DJs.  The effect is similar to a so-called 'frequency isolator', but with both boost and cut, the usefulness is somewhat greater.  Applying low-frequency boost really gets the bass 'pumping', but treble boost is seriously piercing.  Not something I'd want, but I'm not a DJ.  Overall, the circuit (even with the fixed frequencies I used for testing) is much better than I anticipated, but I still wouldn't recommend it for hi-fi.

This is probably the most versatile of the 'basic' tone control circuits.  Greater flexibility is afforded by parametric EQ, but to be useful you generally need at least three bands, and they can be challenging to use.  The Wien bridge-based parametric described in Project 152 (Part 1) is easier to use than a 'true' parametric EQ, which includes not just frequency controls, but variable Q as well.  While ideal for recording or live sound, these are overkill for a hi-fi when you just want a bit more/ less bass and/ or treble.  It's become normal for modern hi-fi systems to not include tone controls at all, which is a shame and makes systems less 'user friendly'.  While the definition of high fidelity implies accuracy, not all recorded music meets that requirement.

I've described a full tone-control preamp (Project 97), but the controls are not variable frequency.  The constructor can modify the response to suit themselves of course, but the values are fixed.  There's a variable tone control shown as part of Project 152, and the treble section uses a variable capacitance multiplier.  It works very well, but the multiplier may enter an 'invalid' state (that I've not been able to reproduce on the workbench, so it's very hard to track down what causes it).  This version is easily substituted for the original shown.

Ultimately, your choice as to whether to use this circuit or not depends on your needs, and whether (or not) you're willing to use four opamps (two dual types) per channel.  The cost in real terms isn't great, but they aren't easy to wire on Veroboard.  The likelihood of a PCB isn't high, but that will change if there's enough interest.  To be useful, a PCB should have provision for either 'traditional' 6dB filters or the more radical 12dB types.  The latter (as discussed above) are good for music production, DJ consoles and instruments (particularly guitar and bass).  The flexibility offered exceeds any other tone control system published (excluding graphic and parametric types of course).

The circuit shown has performance that's almost completely dominated by the opamps you choose.  For the highest performance, it's very hard to beat the LM4562, which has come down in price to the point where it's comparable to the NE5532, but it's one of the very few opamps that surpasses the NE5532 in all respects.  That doesn't mean that you must not use NE5532 opamps of course, as they are still very good, but they do have a comparatively high DC offset.  For 'utilitarian' applications, you can use TL072 or 4558 opamps, and with a 'decent' input level (greater than 500mV RMS) it's unlikely that you'll hear any difference.  You can also substitute your favourite device if you prefer.

Overall, both versions of the circuit are very good.  They are more complex than conventional Baxandall tone controls, but you have the ability to change the operating frequency for bass and treble, and it can be variable over a wide range if that's what is needed.  While you can change the frequencies of conventional tone controls, it requires switched capacitors and is less flexible.  The topology is very flexible, making it a simple matter to have 12dB/ octave tone controls that you may not have thought possible.

The option of using switched resistor (and/or capacitor) values means that very complex tone settings can be reproduced accurately.  This is important for music production, because it's often necessary to get the same 'sound' over multiple recording sessions or performances.  The best tone control circuit ever designed isn't useful if you find a 'magic' setting for a tune, but you can't find it again.  Close perhaps, but not 'quite right'.  This could easily be a dictionary definition of 'frustration'.

The references for this project are mostly in the linked documents.  There are no specific references other than ...

Accelerated Slope Tone Controls - Dennis Bohn (Rane) - Sadly, most of the useful documents are no longer available on the Rane website

Main Index

Projects Index