Tone Controls

[Synchrodyne]

In respect of steep-cut low-pass filters, I imagine that the Quad implementation is one of the better known examples. The first iteration, with 6 and 8 kHz turnover points, was used in the original Quad (Quad I) of 1951. Quad’s rationale is shown in its advertising of the time.



The next iteration, with 5, 7 and 10 kHz turnover points, arrived with the QCII in 1953, and basically similar filters were used in the Quad 22 (1959), Quad 33 (1967) and Quad 44 (1979). The basic rationale remained essentially unchanged, although the 44 had also introduced the Tilt tone control.



The later Quad 34 and 66 had what might be described as a “lite” version of the filter, with two turnover points and two slopes.

There were other viewpoints, though. This is from the technical data for the Radford SC22 control unit:

“The purpose of a treble filter is to reduce the response to high frequencies at a more rapid rate than is possible with the treble tone control. The value of treble filters in the high quality reproduction of sound is a contentious matter.

“It is stated that a high frequency filter of a suitable slope and roll-off frequency will improve listening pleasure on distorted programme material. If a sharp-cut filter is used to eliminate distortion products however, it will also eliminate components of the original sound, and frequency distortion will be introduced. If frequencies above cut-off are not contained in the original programme then it would appear that nothing is lost in the filtering process. Sharp-cut-off filters are also used to reduce noise, but in practice the apparent noise may be increased due to transient ringing. Transient distortion is not serious up to a slope of 10dB per octave roll-off, but it becomes objectionable as the slope is increased above 12sB per octave. It is increasingly audible as the cut-off frequency is lowered towards mid-frequencies. The effect of transient distortion in practice, as produced by a sharp filter, is analogous to a cavity resonance, i.e. listening through a tube in front of the loudspeaker.”

The filter on the SC22 had three switched turnover points, namely 4, 7 and 10 kHz, with a fixed slope of around 12 dB/8ve.


Cheers,

[Synchrodyne]

Probably as well known as the Quad variable-slope low-pass filter was that used by Leak, initially in its “Vari-Slope” preamplifier, the first version of which was released in 1952.

Quad and Leak used quite different circuitry. Quad had used an inductor (choke) as the basis for its variable low-pass filter, and this is the original (1951) circuit:



The inductor was retained for the filters in the Quad QCII, 22, 33 and 44 control units.

On the other hand, Leak used a twin-T R-C filter in the shunt feedback loop of a triode gain stage. This type had been used previously by Williamson for a fixed-slope filter with four switched turnover frequencies:



Here is the initial Leak implementation, with variable slope obtained by shunting the T with a potentiometer that varied the Q:



In its preceding RC/PU/U preamplifier, Leak had incorporated switched treble and bass controls around a triode gain stage, each with three steps of lift and three of cut. The bass control is shown here:



When the variable slope filter was added in the Vari-Slope model, the three cut steps of the treble control were purloined to switch the three turnover frequencies of 5, 7 and 9 kHz, and a slope control potentiometer added. Thus there was overlap between the treble and filter controls, whereas Quad had deliberately kept them separate. On the other hand, the Leak filter had three turnover frequencies whereas at the time, Quad had only two, 6 and 8 kHz.

Leak claimed the advantage for its circuitry that did not use an inductor, as shown in this advertisement from WW 1952 August:



Leak claimed that the more obvious advantages of the electronic feedback method over conventional choke filters included:

(a) Improved transient response characteristics (due to absence of chokes having self-capacitance) and the consequent reduction of “ringing”.
(b) Extremely low harmonic and intermodulation distortion due to negative voltage feedback action.
(c) No discontinuities in the rates of slope when the slope control is operated, an no change in signal level at frequencies below turnover. (Both these faults occur in variable-slope choke filters due to the slope control altering the terminating impedance and the insertion loss.)
(d) There are no chokes to cause magnetic hum pick-up.
(e) Smaller size, lighter weight, greater uniformity in production.

Leak was certainly claiming the “high ground” when it came to variable slope filters.

The Quad QCII/II was released in 1953 September. The QCII retained an inductor based filter, now with three turnover frequencies at half-octave intervals (5, 7 and 10 kHz) and with a slope range of 0 to 50 dB/8ve, whereas the Quad I had been 10 to 100 dB/8ve and the Leak Vari-Slope range was 5 to 50 dB/8ve.

In its sales literature for the QCII, the following was said about the filter:

“The filtering is accomplished at the output of the control unit where the signal level is high and the impedance relatively low. This enables optimum circuitry to be used and does not preclude the use of inductors by hum problems.

“For the best possible performance, the filter must produce the required curve with the minimum of deleterious effects. The effect on transients due to ringing is a direct function of the response curve together with any mis-matching in the circuit. The first is a natural law and is, of course, common to all methods of filtering, the advantage of an LC configuration lying solely in the more favourable conditions for correct matching. With the added advantages of greater range and complete freedom from distortion, a properly matched LC network becomes the obvious choice when other circuit arrangements do not preclude its use”.

It may be noted that the nominal signal level (full-scale) at the output of the QCII was 1.4 volts, as compared with 125 mV for the Leak Vari-Slope, a 21 dB difference. So certainly Quad was in a much better position to use an inductor from the hum pickup viewpoint.

I’d say that in that copy (probably written by PJW), Quad, as well as describing its filter, was addressing potential criticism because of the fact that it used an inductor. The Quad argument was less about what you used than how you used it, whereas Leak’s claims were more based upon what was used than how it was used. (I have a vague notion that the same dichotomy has been observed in respect of other human endeavours.)

The Leak Vari-Slope II of 1954 had the same tone and filter control pattern as the original. In the Vari-Slope III of 1957, the tone and filter control functions were separated, a Baxandall tone control being used in place of the previous switched controls, with its network placed around the same gain stage as used for the twin-T variable slope filter. Leak had first used the Baxandall tone control in its Point One preamplifier of 1954, but this had lacked the variable slope filter.

Leak used the twin-T filter, sometimes simplified with switched rather than continuously variable slope control, through the valve era and then for its initial solid-state amplifier, the Stereo 30. But it changed to inductor-based filters for the Stereo 70 and Stereo 30 Plus models.

As mentioned, Quad used its inductor-based filter with 5, 7 and 10 kHz turnover frequencies for the 22, 33 and 44 control units, in these cases with a 0 to 25 dB/8ve slope range. Nominal full-scale output signal levels were respectively 1.4, 0.5 and 5.0 volts. For the simpler filters with switched slope control used in the Quad 34 and 66 control units, RC filters based around op-amps were used. (I am not sure if these were Sallen & Key filters.)


Cheers,

[kalee20]

Originally Posted by Synchrodyne

Leak claimed the advantage for its circuitry that did not use an inductor, as shown in this advertisement from WW 1952 August:...

I’d say that in that copy (probably written by PJW), Quad, as well as describing its filter, was addressing potential criticism because of the fact that it used an inductor.

The parallel-T filter, as used above, seems to me more a notch filter (maybe 9kHz interference whistle filter? I've not calculated frequencies - pushed for time at the mo!) rather than a tone control as such.

As to use of inductors, well yes unfortunately it's a lot easier for capacitor manufacturers to get to near-perfection than it is for inductor manufacturers. But ringing due to self-capacitance only happens if the values of L, C, R satisfies certain conditions. With more than a critical level of R, ringing just does not happen. I've always felt vaguely unhappy about 'classic' tone controls, such as the Baxandall one, that they are basically non-symmetric for bass and treble, if one uses R and C, then the other should use R and L. Some day, I may try and make one.

[Synchrodyne]

The parallel-T filter was basically a notch filter, and I think that it was used as such in audio work, as you suggest as an AM adjacent channel notch filter (at 5, 9 and 10 kHz).

However, it could be segued into a low-pass filter, and I understand that it was quite widely used in that mode. How that was done was nicely explained by Williamson.




Cheers,

[Synchrodyne]

Preceding the Radford SC24, mentioned in post#36, were the valved SC2 and SC22 control units.

The SC2 of c.1962 had passive treble and bass controls, positioned between two triode gain stages. A single-position, switched low-pass filter was also fitted between the tone control network and the second triode. This was of the simple R-C type.



The SC22 of c.1965, which in general was quite different to the SC2, Radford moved to feedback type treble and bass controls, arranged around a triode pair non-inverting gain stage with series feedback. The bass control was in both the output arm and the feedback arm, whilst the treble control was entirely within the feedback arm. In basic form this was the same arrangement as Quad had used for its QCII control unit.



The low-pass filter was of the switched three-position type, again using simple R-C circuitry, not so surprising in view of Radford’s previously stated position on steep-cut filters, The filter was between the two triodes, so within their feedback loop.


Cheers,

[Synchrodyne]

Originally Posted by Synchrodyne

I have endeavoured to extract a simplified block schematic of the SC24 tone control section for one channel. I was working from a schematic that was in the first place not easy to follow and which was in “small print”, so I am not completely sure that I have gotten it right. Corrections are welcome.

Attachment 183362

The bass control used a one-input, two output network that was partly in the feedback arm and partly in the output arm of a two-stage amplifier (emitter follower Tr4 and common emitter gain stage Tr5). The treble control used a one-input, one-output network in the feedback arm of the same two-stage amplifier, but there was an emitter follower buffer (Tr6) between the output of the bass network and the treble network.

The output from the bass control network fed into a three-stage amplifier (Tr7 common emitter gain stage followed by Tr8 common emitter gain stage DC-coupled to Tr9 emitter follower). The feedback loop for that amplifier went via the middle tone control network (which was of the one input, one output type) to an emitter follower buffer (Tr10) to the emitter of the Tr7 gain stage.

Notable is that each of the two tone control feedback loops included an emitter follower buffer stage entirely within the feedback path. The basic form of the bass control, with a network fed from a gain stage and dividing into an output arm and a feedback arm was similar to that used in the Quad QCII and 22. Inclusion of the treble control network wholly within that same loop also followed the QCII precedent.

The original diagram attached to post #36 contained an error. I had the + and – input signs for the Tr4 stage around the wrong way. Here is the corrected version:



Tr4 and Tr5 together form a non-inverting gain stage. Given that the treble and bass control range is ± 8 dB, its gain with the tone controls flat must be at least 8 dB, and probably somewhat higher. It may also be seen that the SC24 treble and bass controls essentially followed the pattern that was used in the SC22. The latter had used a triode pair rather than a transistor pair for the non-inverting gain stage. In the SC24, an emitter follower buffer was placed in the feedback loop between the bass control and treble control networks. Presumably this was to minimize interaction between the two and to allow independent choice of network impedances.



The SC24 differed from the SC22 in having a (fairly unusual) middle tone control, providing a lift of up to 6 dB at 1.5 kHz, but no cut. It could be considered to have been a form of “presence” control. The circuitry for this was added after the treble and bass controls. As previously stated, the middle control RC network was fully in a feedback loop around a non-inverting amplifier comprised of Tr7, 8 and 9. A buffer stage Tr10 was included in the feedback loop, after the tone control network. Given that Tr10 had a bootstrapped input, one may deduce that the middle RC network needed to look into a very high impedance, more than provide by the emitter of Tr7, whence the feedback loop returned. (I am not sure what impedance would be seen looking into Tr7 emitter, but more I think than the value of the Tr7 emitter resistor.) The middle pot was 250k as compared to 25k used for the treble and bass controls, so it was a higher impedance circuit. Given that the middle control provided lift but not cut, the amplifier formed by Tr7,8.9 would not have been constrained to have a certain minimum flat gain.



Middle tone controls generally seemed to have been seldom used. Perhaps the most notable user was Tripletone, which company name (changed from the original Servio) was derived from the fact that its first amplifier, the Mk I of 1954, had three tone controls, treble, middle and bass. This feature was retained, as far as I know, on all subsequent Tripletone amplifiers, both valve and solid-state. Whether Tripletone used passive or active tone control circuitry is unknown; schematics for its equipment appear to be unobtanium. Active circuits were gaining ground by 1954, so that approach was certainly possible. On the other hand, valve count constraints in simple amplifiers may have pointed to the passive type.

Here is a generalized feedback tone control with treble, middle and bass controls:



It could be seen as a Baxandall modified by the addition of a middle control. In this circuit, C3 would provide a bypass for treble frequencies, ensuring that RV2 had but minor effect on them, and C4 would block bass frequencies, again ensuring minor effect from RV2. Thus effectively RV2 would be operating in a mid-range bandpass filter.


Cheers,

[Synchrodyne]

The Radford SC24 had a similar low-pass filter to that of the preceding SC22. That is, it was switched with selectable turnover frequencies of 4, 7 and 10 kHz, with a fixed slope of 12 dB/8ve. I haven’t extracted the actual circuitry, but I imagine that given Radford’s views on the matter, it was of the simple R-C type similar to that used on the SC22.

Also inherited from the SC22 was the “Quiet Listening” facility, a form of loudness control. The SC22 circuit is shown here:



Radford’s rationale was as follows:

“All recording and reproducing systems are equalised to a flat response and do not take into account the variations in frequency response of the ear at different listening levels. It is assumed that the normal listening conditions approximate the sound pressure of the original source. Curves of the ear response with varying sound pressures were published many years ago by Fletcher and Munson in America and more recently by Robinson and Dadson in England. At lower listening levels a higher output at low frequencies is required.”

And the means of achieving the desired outcome was described thus:

“An attenuating circuit with bass lift is interposed between the auxiliary amplifier and the output amplifier. This network provides a reduction of the sound pressure level of 15 phons with a frequency/amplitude correction for this level reduction. This facility assumes that the pre-amplifier has been set up in conjunction with the power amplifier and loudspeakers to provide maximum listening sound pressure level at position 7 to 8 on the volume control. The change in response characteristic from the level response is shown in Fig. 12 and the circuit giving the detail of the network introduced in the circuit is as in Fig. 11.”

J.E. Sugden also included a similar “Quiet” control in its C51 control unit of the late 1960s. (I think it was also included on the preceding C41 which was sold as a Richard Allan product.)

Sugden’s rationale for its “Quiet Control” was given as:

“Attenuates mid frequencies by 16dB but lower frequencies to a less extent in accordance with the subjective equal loudness curves of the ear. This facility where normal volume is set and then "quiet" selected is to be preferred to the so called "loudness" compensation of a volume control which has no relevance to the absolute value of the incoming signal level.”

I have not yet found a schematic for the Sugden C51, and the available data do not include a curve showing the effect of the quiet control, just those for the tone and filter controls. Whilst Sugden was aligned with Radford in respect of the quiet control, it evidently differed to some extent when it came to low-pass filters. The C51 had the same turnover frequencies as used by Radford in its SC22 and SC24, namely 4, 7 and 10 kHz, but offered two slopes for each, namely 6 and 18 dB/8ve, the latter steeper than Radford thought advisable.

With their quiet listening controls, both Radford and Sugden sought to address one of the several criticisms, both theoretical and practical, made in respect of simple loudness control systems, which was that their frequency bending action was related simply to volume control position and not in any way to sound pressure level. That is, when switched in, they simply changed the volume control from a frequency insensitive device to a frequency contoured device – at least at lower levels - with a rather arbitrary transition point.



Another way around the problem was to have separate loudness and volume controls. In this case, the loudness control provided variable attenuation from the volume control setting with progressively increasing frequency contouring. Here is an example, as used by Beam Echo:



The Radford and Sugden quiet controls can be viewed as being similar but with a single fixed level of attenuation rather than variable.


Cheers,

[Craig Sawyers]

Loudness controls certainly had their day, although I'm not sure how many people actually used them. They seemed to arise by a combination of engineers over thinking the problem which arguably does not really exist, and marketers looking for a USP advantage or a me-too product.

You probably know when they dropped off the audio perch in detail, but I would say offhand they were out of vogue by the mid 80's.

Craig

[Hartley118]

Originally Posted by Synchrodyne

Middle tone controls generally seemed to have been seldom used. Perhaps the most notable user was Tripletone, which company name (changed from the original Servio) was derived from the fact that its first amplifier, the Mk I of 1954, had three tone controls, treble, middle and bass. This feature was retained, as far as I know, on all subsequent Tripletone amplifiers, both valve and solid-state. Whether Tripletone used passive or active tone control circuitry is unknown; schematics for its equipment appear to be unobtanium. Active circuits were gaining ground by 1954, so that approach was certainly possible. On the other hand, valve count constraints in simple amplifiers may have pointed to the passive type.


Cheers,

A friend of mine owned a Tripletone 10W amplifier in the late 1950s. The main unusual feature that I remember was that turning bass, middle and treble to minimum had much the same effect as turning down the volume to zero. That suggested three separate filters whose outputs were mixed together. They were probably passive filters because the valve complement was pretty standard.

Unfortunately I never had the opportunity to investigate the circuit.

Martin

[Synchrodyne]

Thanks for that. I have since remembered the attached item from WW 1973 December, which briefly described the Tripletone 1818 amplifier, and which provides indirect confirmation.



Therein it was said: "Dual concentric tone controls, bass mid and treble, now operate active circuits…”

That implies that hitherto, Tripletone had not used active tone controls, but rather the passive type.


Cheers,

[Synchrodyne]

Originally Posted by Craig Sawyers

Loudness controls certainly had their day, although I'm not sure how many people actually used them. They seemed to arise by a combination of engineers over thinking the problem which arguably does not really exist, and marketers looking for a USP advantage or a me-too product.

That is a nice succinct summary. I guess one can add that some users liked them as another reason.

Originally Posted by Craig Sawyers

You probably know when they dropped off the audio perch in detail, but I would say offhand they were out of vogue by the mid 80's.

I don’t have a good fix on that part of the timeline, but mid-1980s sounds about right. A quick scan through Hi Fi Year Book 1980 shows that loudness controls were still alive and well at that time, prevalent on Japanese and Continental equipment but less so on British equipment. Loudness controls may have been caught by a double trend, one an overarching move away from “frequency bending” of any kind and the other the dying away of the loudness control concept itself.


Cheers,

[Synchrodyne]

In the development history of treble and bass tone controls, one may see that an early key objective was to obtain continuous variation without the need for any switching. The passive networks such as the Volkoff/James and Voigt achieved this, and Baxandall provided a more elegant approach that delivered reasonably symmetrical curves, and which despite some early misgivings, was found to deliver different kinds of treble curves according to individual amplifier designer preference.

Notwithstanding the subsequent dominance of continuously variable treble and bass controls, some amplifier makers preferred to use switched controls. An example of switched controls, from the early solid-state era, was found in the Revox A50 integrated amplifier of 1968. These were carried over to the succeeding A78 of 1971.

Here is the Revox A78 tone control circuitry:



As may be seen, the tone control network was wrapped around an inverting amplifier stage, partly in the input arm and partly in the feedback arm. Thus it followed the Baxandall functional form. Bass control was effected by a switched resistance network, so could be said to be a switched version of the Baxandall. On the other hand, treble control was effected by switching a capacitor network. The resultant curves were as follows:



Also of interest is the form of the tone control amplifier, using two transistors in apparent DC-series. As noted upthread, quite a bit of effort was expended in developing suitable discrete transistor amplifier configurations for the Baxandall tone control. These spanned the range from single transistor (Bailey) through single transistor with simple emitter follower, single transistor with bootstrapping emitter follower (Quad, Quilter and H.P. Walker), cascode transistor pair with bootstrapping emitter follower (Ellis) to discrete opamp (Meyer).

The arrangement used in the Revox A50/A78 tone control was also used as the second stage of the input section of the same amplifiers:



A description of this circuit was provided as follows: “The transistor Q103 in the cascade connection of Q102 and Q103 acts as the collector resistor with a high impedance for audio signals, while representing a low DC resistance.”

Nonetheless, to my untutored layman’s eye, it looks like a variation on the bootstrap theme. My simplistic analysis is that in a “regular” bootstrap circuit, the collector of the gain transistor is DC-coupled to the base of the emitter follower, whilst the emitter of the emitter follower is AC-coupled (via the bootstrap capacitor) to the upper end of the gain stage collector load resistor. The signal voltage at the emitter follower emitter, and so that at the top end of the gain stage collector load resistor, is likely to be greater than 99% of that presented at its base, which is the same as that at the bottom end of the gain stage collector load resistor. So to the signal appearing at the gain stage collector, the load resistor looks to have been magnified by 100/(100-99.nn), or to put it more technically, quite a lot.

In the Revox circuit, the collector of the gain transistor is AC-coupled to the base of the emitter follower. This in turn allows the emitter of the emitter follower to be DC-coupled to the top-end of the gain stage collector load resistor, which puts the emitter follower in DC-series with the gain stage. As in the “regular” case, the emitter follower has gain stage collector signal voltage at its base, so 99.nn% of that at its emitter. Thus also as in the “regular” case, the signal at the gain stage collector sees the load resistor as being magnified by the same multiplier. In both cases, the signal output is taken from the emitter of the emitter follower. Of course, the DC analyses would be different.



A similar circuit was used in the input section of the Revox A77 tape recorder:



It may be seen that Revox, like Quad, was quite early in opting for higher open loop gain in a simple inverting amplifier used for a feedback tone control, although doing it unconventionally.

It was also early in opting for higher open loop gain in a discrete non-inverting amplifier used inter alia for RIAA equalization. As far as I know, that wasn’t done with the Bailey circuit until Self did it in 1976. The Revox circuit could be considered to be an enhanced Dinsdale, done in silicon and retaining the bootstrapped input load resistor.


Cheers,

[Synchrodyne]

The Revox A50 and A78 also had switched single-position filters, one high-pass and the other low-pass, obtained from R-C networks in the input and feedback arms of a two-stage non-inverting amplifier. Following this was a passive switched “Loudness Low” control that provided an attenuation of 16 dB at 1 kHz and applied a contour. Thus Revox was doing similarly to Radford and Sugden when it came to the loudness control. The A78 added switched single-position presence control to that group, ontained by an R-C network in the feedback arm of the filter amplifier. Here is the circuit:



And here are the curves:




Cheers,

[Craig Sawyers]

In post 57 I showed a 3D rendering of the layout for a recreation of the Quad tilt control. By way of an update, the boards arrived and one has been populated. This is for my lower impedance version, with lower value resistors and higher value capacitors.

I decided to use switched resistors in the end rather than continuous variation using a dual pot. The track matching on a Bourns 91 series dual pot was not good enough to guarantee a flat frequency response other than by careful fiddling.

I've gone a bit bonkers and used 0.1% resistors. The capacitors are 18n and 5.6n. So I bought ten off each Farnell 2468984 and 2495967, nominally 5% tolerance and then used my GR bridge to find matched pairs that were good to 0.1%.

Yet to wire it up - it is actually part of a larger project - but Spice simulation putting in the maximum tolerance bands for resistors and capacitors yields a worst case flatness of +/- 0.021dB, which is good enough!

Craig

[Craig Sawyers]

This is the sensitivity plot. The overall envelope of +/- 0.021dB is for all components at the worst tolerance in the worst directions.

Craig

[Radio Wrangler]

That'll be hellish if there's ever a signalling failure!

David

[Craig Sawyers]

Ha ha!

[G6Tanuki]

I remember some 1970s/early-1980s "Audio separates" where the amp had a Loudness-switch which delivered an undocumented-but-noticeable boost to the mid-range frequencies.

In the amps I was designing back then, such a switch was entirely redundant: the bands who bought my gear [typically based around a pair of 807s in zero-bias Class-B, the drivers turned-up-to-11] always wanted maximum-boost-at-all-frequencies-all-the-time!

[Synchrodyne]

Originally Posted by Radio Wrangler

That'll be hellish if there's ever a signalling failure!

More complex even than Linsley Hood's "Clapham Junction" tone control.



Cheers,

[Edward Huggins]

I have never known of a "Loudness" control to boost the mid range - it did just the opposite.
Evenso, Philips featured what they called a "Presence" control which did have some effect on the low-end of the high frequency curve.