# SW04

## Stereo width control.

### Introduction.

The most common use of a stereo width control is to reduce the stereo width of some early stereo recordings.
Many early recordings was made with excessive stereo width ( aka ping-pong stereo ) so it sounded like you were sitting in the middle of the orchestra ( in one case - like you were sitting in the middle of the piano ).
Some other uses of a stereo width control are:

• To expand the stereo width in case the speakers are placed too close together ( ie. in a sound-bar ).
• Introduce ( optionally variable ) cross-feed for headphone listening.
• Reduce acoustical feedback for gramophone reproduction.
• To create the rear channels for a Hafler matrix circuit.

The circuit shown here does this by splitting the AB ( left and right ) signal into a mono component ( the M-signal ) and the stereo component ( the S-signal ), processing the S-signal and then converting the signal back to AB format.
This circuit ( AB-MS-AB matrix ) is the most versatile for stereo-width manipulation, but for several applications a simpler circuit will work equally well.

The article here only shows the basic circuits. Details like supplies, HF feedback around the OP-AMPs and DC-couplings/bias components are omitted for clarity.

### AB-MS-AB matrix.

Fig.1: Basic AB-MS and MS-AB circuit.

The amplifiers U1 and U2 are shown to emphasize that the circuit requires a low drive impedance.
Likewise, the drive impedance for the Min and Sin inputs must be low.
Using 1% resistors from the same tape for R and R/2 you will normally have a channel separation of 40 dB or better.

Input ( AB-MS ) matrix:

Mout is the mono output. The minus signs in the equation shows that the circuit is inverting.
Sout is the stereo ( or side ) signal. This is the difference between the A and B channels.
The input matrix has a gain of 0.5 to avoid overload in case of 2 full-scale input signals.

Output ( MS-AB ) matrix:

For Min=Mout and Sin=Sout:

This shows that the stereo signal is passed straight through.

For Min=Mout and Sin=0 ( the Sin signal is grounded ):

This is a mono signal.

For Min=Mout and Sin=-Sout ( Sin is Sout, inverted ):

This is a stereo signal with the channels swapped.

These are the most basic settings. It is possible and useful to make the S-signal adjustable and/or filter it.
The results are useful, but the formulas are not intuitive to read.

### Stereo width control.

Fig.2: Stereo width control.

Fig.2(a) can regulate the stereo width from the width of the incoming signal to mono. Although not intuitive, the buffer U7 is required to maintain stereo balance.
Fig.2(b) can reduce or expand the stereo image. The amount of expansion depends on the gain around U8. 2..4 times should be more than enough ( too much ) for most applications. Capacitor C1 is a high-pass filter that set a lower frequency limit for expansion to keep the speaker cones in their linear excursion range ( this is mainly an issue with gramophone playback ). A -3 dB frequency around 100 Hz to 200 Hz normally works. Note that the gain in the AB-MS stage must be reduced to avoid overload in this stage.
It can be tempting to use a higher order filter for the S-channel, but IMO it sounds awful.

Fig.3: Stereo width control for AB signals.

Fig.3(a) can regulate the stereo width from the width of the incoming signal to mono. This circuit must be fed from a reasonable low impedance, resistive source.
Fig.3(b) can expand the stereo image. The amount of expansion depends on the component values. It is not really useful as a variable stereo width expander as the LF roll-off frequency depends on the potmeter setting. It is very useful for fixed stereo width expansion if the speakers are placed too close together like in a sound-bar.

You can find a number of different stereo width control circuits on the net, but most of these are variants of old circuits designed when amplifiers were expensive and ( dual-gang ) potmeters ( relatively ) inexpensive. Today you will get more than 10 dual OP-AMPs for the price of a single-gang quality potmeter.

Using cross-feed for a headphone is an attempt to emulate speakers in a well damped room using headphones.
A circuit was - as far as I know - first published by Siegfried Linkwitz ( see [1] ) in 1971.
The basic idea is to remove the stereo information below approximately 700 Hz.
You can find several circuits on the net that does this in different ways and some of these are shown here.

Fig.4: Headphone crossfeed schematics for AB signals.

Fig.4(a) is Siegfried Linkwitzs [1] design. It is designed to be inserted between a power amplifier and a 600 Ω headphone.
It has an insertion loss of around 12 dB.
Fig.4(b) is from Diy Audio Heaven [2]. It must have a low source-impedance and a high load-impedance.
Fig.4(c) is from Meier Audio [3]. It must have a low source-impedance and a high load-impedance.

Fig5(a) is a fixed frequency cross-feed circuit. It simply removes the S-signal at low frequencies ( at <700 Hz as shown ).
Fig.5(b) is an adjustable cross-feed circuit.
P5 adjust the stereo width at high frequencies, P6 the width at low frequencies and P7 the cross-over frequency.
In my initial design, P6 was connected directly to the Sout signal. At some settings this can best be described as a silly sound-effect.

### Acoustical feedback reduction for gramophones.

Acoustical feedback occurs when a gramophone picks up its own signal from the loudspeakers.
The audible results spans from a muddy bass reproduction through a slight low frequency rumble and all the way up to where the signal from the record is severely distorted.
This procedure is a simple way to identify low levels of acoustical feedback in the system:
Record a piece of music containing low frequency signals on a high quality tape recorder ( sorry - I wrote this in 1990 ). During the recording, the loudspeaker volume is switched between off and normal listening level.
Play the tape at a normal listening level. If there are audible differences between the parts where the loudspeakers were on and off, this is a clear indication of acoustical feedback in the system.
Another way to identify acoustical feedback is to compare the music from the gramophone when it is played in mono and in stereo.
A more precise bass reproduction when playing in mono indicates a problem with acoustical feedback as most records are recorded in mono at low frequencies.
Some remedies against acoustical feedback are:
• Place the gramophone on a stable surface.
• Move the gramophone and loudspeakers away from each other.
• Adjust the gramophone's tracking pressure and anti-scating properly.
• Insert a rumble filter in the signal path ( this may also remove some of the low frequency content of the signal ).
• Use a preamplifier with built in feedback reduction filter ( "Feedback Canceling Filter" in marketing terms ).
The feedback reduction filter shown here reduce the acoustical feedback from the loudspeakers to the gramophone and enables up to 10 dB higher SPL before feedback becomes audible.
The filter simply adds the two channels to a mono signal at low frequencies.
The idea of summing low frequencies into a mono signal to avoid feedback is not new, but all earlier implementations uses some sort of cross-over network to extract low frequency information, adds the two channels and then add this signal back into the stereo high-frequency signals.
This topology gives an undesirable phase response and will in some cases make the feedback problem worse.

Fig.6: Acoustical feedback reduction for gramophones.

Fig.6(a) is my original feedback reduction filter. It was designed for discotheque RIAA amplifiers around 1990. The cross-over from mono to stereo is around 100 Hz. It requires a low drive-impedance and a high load-impedance.
Fig.6(b) does exactly the same ( so it is basically a waste of components ). By replacing R28 with a potmeter it will also work as a width control. For most Hi-Fi applications, the cross-over frequency can be reduced by 25..50%.

### Hafler matrix circuit.

Fig.7: Passive Hafler matrix circuit.

Fig.7 shows a typical passive Hafler matrix circuit.
The 2 rear channel speakers reproduce the left/right difference signal giving a sense of ambiance.
The circuit was built into some amplifiers from the early 1970s and were also available as separate units.
With stereo program material, the power dissipated in the attenuator and rear speakers is very limited and an attenuator designed for 5 W is sufficient.
When the balance control is turned fully to one side, you have the full power from one channel and will probably have a attenuator-repair on you hands.
This problem can be solved by using an active matrix circuit, but you will need an additional stereo amplifier.

Fig.8: Active Hafler matrix circuit.

This is the circuit of fig.1 with the following modifications:
R9, R12, R13 removed.
R11, R14 replaced by jumper.
Depending on your rear speakers, you may want to roll the rear channel of above 6 kHz to 8 kHz.
This is done by placing a capacitor across R6 and one across R7 ( C = 1 / ( 2 * π * f * R ) ).
If you want to include a delay for the rear speakers this can be inserted in the Sout signal between U4 and P8.
The Mout signal can be used for a center channel or for a sub-woofer output. Otherwise, R1, R2, R3 and U3 can be left out.

### References.

 [1] Siegfried Linkwitz: Improved Headphone Listening ( Build a stereo-crossfeed circuit ). [2] DIY-AUDIO-HEAVEN: Crossfeed. [3] Meier Audio: Crossfeed. This page includes some sound examples of the effect of crossfeed.

This web-page, including but not limited to all text, drawings and photos, is the intellectual property of Poul Petersen, and is Copyright ©.
Reproduction or re-publication by any means whatsoever is strictly prohibited under International Copyright laws.
The author grants the reader the right to use this information for personal use only.
Any commercial use is prohibited without express written authorization from Poul Petersen.

The information is provided on an "as-is" basis and is believed to be correct, however any use of the information is your own responsibility.

This web-site may contain links to web-sites outside Poul Petersen domain ( www.poulpetersen.dk ).
Poul Petersen has no control over and assumes no responsibility for the content of any web-site outside Poul Petersen own domain.

Poul Petersen does not use cookies to "enhance your experience" on this website.

Poul Petersen, C/Faya 14, 35120 Arguineguín, Las Palmas, Spain.
E-mail: pp@poulpetersen.dk