Improving the R. Ambler tone-balance control

[GK.]

Way back in 1970 Wireless world published another form of tone control to complement the standard Baxandall bass and treble control:

https://keith-snook.info/wireless-world-articles/Wireless-World-1970/Tone-balance Control DCD.pdf

A single potentiometer control simultaneously provides both bass-boost and treble-cut or, alternately, bass-cut and treble-boost, depending on which direction from center (flat response) it is turned. Hence the name "tone-balance control" - the balance being between bass frequencies and treble frequencies.

I hate Baxandall bass and treble controls as they are almost universally designed with a mutual bass/treble contour crossover in the vicinity of 1kHz. A bass control in particular that has a significant influence to near 1kHz is terrible for compensating speakers with limited bass response because it is crossed far too high. What you get with such a control is an audibly unpleasant boosting and muddying of the lower mid-range.

The Baxandall tone control is much better designed so that the bass and treble controls leave the region of 500Hz to 2kHz alone.

This is where I see the tone-balance control as a addition to the Baxandall bass and treble control to have merit. The Baxandall controls can be used for correction while the tone-balance knob can then take on the role of the altering the balance between bass and treble frequencies to your subjective preference.

Quad was one of few manufacturers that adopted the tone-balance control, calling it a "tilt" tone control. The problem with the basic circuit, however, is that its tilt response is sharpest in the mid-range:

So it is more like a control that gives an alternate boost and cut to bass and treble frequencies respectively (or vice-versa) rather than something that gently and progressively "tilts" the frequency response to your subjective preference from one end of the spectrum to the other. As a matter of fact, at any setting the frequency response is very close to that of the Baxandall tone control (as it is typically poorly implemented) with the treble control set to an equal and opposite level of boost to the bass control.

I've been looking at Quad schematics. In the Model 34 control unit the tilt balance control is implemented in a single stage along with a bass boost/cut control. There is no independent treble control:

It appears that the Quad designers had similar concerns to me, as they have used parallel RC networks with stepped frequency crossovers in this variant of the tilt-tone control circuit in an attempt to broaden and flatten out the tilted response about the crossover frequency.

The improvement, however, is rather modest. I've gone back the original circuit by Ambler and have started to experiment with the same approach, but expanded with a greater number of stepped RC networks. The approach looks promising so far.

The aim is to get a frequency response curve that is a good approximation to a straight, tilted line between 20 Hz and 20 kHz, which demands a see-saw pivot at about 1 kHz for an equal level of influence at either end of the spectrum.

 

[GK.]

OK, I've now taken a slightly more rigorous mathematical approach instead of just pulling component values out of my arse. The five frequency cross-over points are properly positioned geometrically from 20Hz to 20kHz as best as possible using standard component values. Designing to standard values removes the added complication of having to parallel additional components when realising the thing to otherwise arrive at the required C and R values.

The result is much improved.

More RC networks for additional distributed frequency intercept points could be used to still better approximate the ideal straight line response, but I think this is near as good as it needs to be.

I wonder how much further the shown network of five parallel R-Cs per leg can be can be optimised. I chose the 100:1 spread between the largest and smallest C more or less arbitrarily, but this has worked out to be what appears to be fairly close to an ideal range. My overall results so far with either a significantly smaller or a significantly larger range haven't yielded as good a result.

Here is the response beyond the audio spectrum by one decade at either end. It flattens out quite quickly so there isn't any excessive ultra and infrasonic boosting being applied that could cause signal overload unless otherwise addressed.

 

[GK.]

Getting better, but now I'm using non-standard resistance values..............

Now the response is almost symmetrical with the crossover frequency at the geometric midpoint (SQRT 20Hz*20000Hz) of about 632Hz, where it should be.

I think just 5 or 6 dB boost and cut maximum should be ample, as that is accompanied by an equal and opposite amount of boost or cut at the opposite end of the spectrum.

 

[LTig]

I think just 5 or 6 dB boost and cut maximum should be ample, as that is accompanied by an equal and opposite amount of boost or cut at the opposite end of the spectrum.

Yep, this is exactly what the digital tilt tone control in my Classé Sigma SSP does.

 

[GK.]

It took some doing, but here is where I am now at using standard/stock component values throughout.
Worst case driving impedance is 1260 ohms @ 20kHz. Modern high-end audio op-amps can happily drive 600 ohms quite OK, so no real problem there.
I'm done for this evening, but tomorrow I'll start on an evaluation PCB layout incorporating a stereo version of both this tone-balance control and a properly designed Baxandall bass/treble control.

 

[GK.]

I was hoping to have this PCB project done already, but I've only just completed the schematic entry.
The simulated response of the Baxandall bass and treble section and schematics:

 

[Scrivs]

Cool, reminds me Fenders "TBX" tone control

 

[raindance]

John Broskie offers a couple of tilt control kits similar to the Quad type over at Glassware Audio. I found plus/minus 3dB to be somewhat effective, less than that is a bit too subtle.

 

[GK.]

A small amendment/refinement. These added resistors will have a negligible effect on the circuit operation, but will keep the op-amp outputs put instead of hitting the rails should the pot wipers ever momentarily go open circuit.

 

[GK.]

The PCB layout is still a work in progress, but....

In LTspice you can import a wave file for your signal source. You can also export to a wave file using the .wave directive. So you can effectively listen to the effects of your simulated audio filters and circuits.

Attached are three audio files of one song clip put through the simulation of the final design iteration of the tone-balance control. One is a "recording" with the control set flat, one with full tilt to bass and the last with full tilt to treble.

I've converted these audio clips from wave files to mp3 to reduce the file size and have had to change the file extension to one permitted by the forum. To play rename from *.txt to *.mp3.

 

[GK.]

 

[GK.]

I am going to need a power supply I suppose.