Surround Sound Decoder (Mk. II)

Like the decoder shown in Project 18, this surround-sound decoder is based on the 'Hafler' principle, first discovered by David Hafler sometime in the early 1970s.  The original idea was to connect a pair of speakers Between the main speaker output terminals, with the pair connected out of phase.  While this works, it lacks the necessary delay to obtain a 'spacious' sound field, and it also lacks the ability to use the surround outputs from a home theatre system when that's in use.

This 'new' design isn't at all new in reality, but is a combination of projects that provides a 'pseudo surround' signal when the 'real thing' is unavailable.  It also lets you switch over so that the amp is used in the normal way, either so that it's fed from a 'true' surround output, or you can use the amp to provide additional coverage (or for any other purpose) with the flick of a switch.  The amplifier will typically be a P19 (2 × LM3886) or P127 (2 × TDA7293), but of course any other amplifier can be used as well.

The project itself uses a Project 87B balanced input stage to derive the 'left-right' signal and to invert one channel, and the Project 26A digital delay to provide the necessary delay to obtain the sensation of surround sound.  Because PCBs are available for all parts of this project, it supersedes that shown in the original Project 18, but it does not include a centre channel or a subwoofer output.  This is unlikely to cause anyone any grief, since a sub should ideally have its own crossover (such as Project 09) and a centre channel can be created with two resistors that simply sum the Left and Right outputs.  This is shown as optional in Figure 4.

Note that if the input signal is mono, then the signal in both channels will be more or less identical, and there will be no output from the rear speakers at all.  There is no 'work-around' for this, and it's unrealistic to expect a mono signal to have a surround effect anyway.

This circuit works by subtracting the signals in the Left and Right channels, in exactly the same way as FM broadcasts separate the two signals into L+R and L-R.  The latter (subtracted) signal is transmitted on a sub-carrier and decoded back into separate Left and Right channels by the receiver.  By allowing the rear speakers to reproduce only the difference signal between the left and right outputs (with a delay of around 30-50ms), it creates the illusion of a full surround signal.  While the drawing below shows Left and Right signals, they are interchangeable, because the output is the difference between the two.

While it's common for the centre channel to be filtered so that it only reproduces from (around) 100Hz to 7kHz, this hasn't been incorporated into this design.  There are two reasons for this decision, with the main factor being that rear/ surround speakers are generally of lesser quality than the main speakers, and will likely already be compromised in performance and/ or by their placement in the room.  Noise isn't an issue, so apart from the filtering applied by the PT3299 circuit (which is fairly mild) no additional filtering was considered necessary or desirable.

Figure 1 - The P87B Matrix And Inverter

The subtraction is done by one channel of the Project 87B Balanced Line Receiver, with left and right channels applied to each input.  Because each is earth (ground) referenced, the net result is that only the difference is passed through, namely Left minus Right.  The second channel is re-configured as an amplifier and inverter, which requires minimal changes to the P87B circuit board.  As shown, some resistors are either left out (DNI - Do Not Install) or replaced by a link.  The first channel supplies the signal to the delay circuit, and the second channel amplifies the delayed signal, and provides a 'normal' and 'inverted' output for connection to the power amplifiers.

While I have included a trimpot to adjust the level, it may (or may not) be necessary in practice.  The absolute maximum input level for the PT2399 delay IC is 1.2V RMS (sinewave), or 1.7V peak, and if the output of the subtraction circuit is greater than that the IC will clip.  For a mono signal applied to either input, the circuit has a gain of two, but it will be less for a stereo input.  However, some material may contain a fairly large difference signal, and it's better to be able to control the level than not.

If the level isn't great enough, replace R203 (10k) with a lower value, which will provide more gain.  You can add a 10k trimpot in the R203 position with not too much difficulty, and that will let you set the gain to whatever is necessary for your setup.  Most of the time I expect the standard gain will be sufficient, and with a more-or-less 'typical' input level of 500mV to the delay circuit, each output from the second channel of the P87B will be 1V.  That's more than sufficient to drive the power amps to full power.

Figure 2 - Modified P26A Delay Circuit

The delay circuit is as shown in the Project 26A schematic, but is modified to suit this application.  While there are two different 'ground' connections (analogue and digital), it's apparent that they must join, and this is provided on the PCB.  The connection between pins 13 and 14 is optional but recommended.  It's considered bad practice to leave the input of an opamp floating, and this connection ensures that the internal opamp can't cause any 'mischief'.

There is no requirement for the 'Repeat' function (which creates an echo effect), although it can be included if you imagine that it will help in some way.  This isn't recommended though, as the result tends to be aurally messy, and the main dialogue (in particular) can be affected badly if it's off-centre (i.e. panned Left or Right).  The idea is that the surround sound should be subtle, and not 'in your face'.

The delay circuit is configured in its 'minimalist' form, which works very well with a fairly high clock frequency.  With the values shown, the clock frequency is around 12.5MHz, giving a delay of roughly 55ms.  This is somewhat greater than is usually suggested, but I've tried it and it should sound about right for most rooms, but the delay can be increased by increasing the value of R10.  See the P26A page for a complete list of delay frequencies vs. R10 value.  The PCB has provision for a pot so the delay can be changed, and I leave it to the constructor to decide if this is necessary or not.

If the pot isn't used, a link needs to be added to the PCB, but this is shown in the P26A build instructions.  I used a 1k resistor for R10 and a 10k trimpot, and it was simple to adjust the delay for any delay time within range.  Although there is no necessity for an exact setting, some constructors may wish to experiment with shorter delays.  Be aware that if R10 (plus the resistance of the 'Dly' pot) is less than 2k, the oscillator may refuse to start.  The Project 26A page shows a simple transistor circuit that can be used to ensure that the clock oscillator starts reliably with a low value of timing resistor.  The shortest practical delay is about 22ms with this IC, and to get that you must use the start-up circuit.

While the delay circuit's input cap is shows as 10µF, you can reduce the bass response by using a lower value if needs be.  A value of 220nF will give a -3dB frequency of 72Hz, or you can calculate a value to suit your requirements.  The input impedance is 10k, and I've replaced C1 with 220nF (this is not shown in the photo of the boards seen in Figure 3).  Few surround speakers will have much response below 80Hz or so, and rolling off the bass helps to reduce the amp power needed for a given SPL.

Because this project uses the boards in unusual ways (especially P87B), photos are shown here so you can see just what is done.  It will make more sense when you look at the construction details on the secure site (available with a PCB purchase), but the pix do show that there's nothing difficult.  No tracks are cut, and there are no links under the PCB.  As mentioned above, I have since reduced the input cap (C1) to 220nF to reduce bass output.

Figure 3 - PCB Photos

The P26A board is on the left, and uses the minimum parts count version as shown in Figure 2.  The delay time trimpot is the blue part next to where it says 'Dly' (Delay).  The repeat pot is not used.  The only major changes to the P87B board are to the lower section, where two jumpers are used in place of resistors, two resistors are omitted, and one (R212) is added just below U2.  The right end of the resistor lead requires an insulating sleeve so it doesn't short out to the top of R210.  The upper section omits two resistors from the board.  These two boards have been tested to verify that they perform as expected (they do), and the output of the delay IC is (perhaps surprisingly) remarkably quiet.

You can use any power amplifier you like, whether a commercial unit or one built specifically for this project.  Both P19 and P127 are suitable, and you usually don't need a great deal of power.  Something in the order of 20W/ channel (8Ω) will almost certainly be enough, and that only requires a supply voltage of around ±20V.  You can use either a 15-0-15V transformer (up to 24W/ channel) or 18-0-18V (up to 35W/ channel) depending on the transformer's regulation.

To make the unit as flexible as possible, and especially if you build the complete unit into a single chassis, it's worthwhile adding a switch so that the amp can be used for 'surround' or as a normal stereo power amplifier.  This makes it easy to set up if you have a full Dolby decoder but also want to use the surround speakers with older material that has no surround soundtrack.  Ideally, the amp will be fitted with two sets of inputs, with one set dedicated to the 'normal' use of the amp, and the other specifically for the surround feature.  Make sure that the four RCA connectors' shield connections are securely wired together, with C1 to the chassis to prevent RF noise from entering the case.

If the centre channel output is included, ideally that would also feed a P09 or similar to obtain the subwoofer output.  In general, this would be done within the main preamp, but the option is there.  Note that the centre channel output has a comparatively high impedance (1.5k close enough), but this shouldn't cause any difficulties.  The level is reduced by 6dB unless a gain stage is added.

Figure 4 - Inputs & Switching To Enable/ Disable Decoder & Delay

There's nothing even remotely difficult about the switching.  VR1 (A&B) controls the level when the surround effect is used, and the 'normal' amplifier signal goes straight through.  You can add a second volume control if desired for the normal channel, but mostly the level will be set up in the surround decoder if one is used.  The power amps themselves aren't shown here because there is a range of projects to choose from, but as already noted I suggest either P19 or P127.  Either will be more than sufficient.

The switch (Sw1) is a DPDT (double-pole, double-throw) type, and simply selects the desired input.  Depending on how you use the system, it can be on the front or rear panel.  The same applies for the volume control (VR1), but make sure that if it's mounted on the front panel that you use shielded wiring for the inputs and outputs, or you may pick up noise and it's even possible for the power amps to become unstable if there is any feedback from output to input(s).

The power supply for the power amps should have a DC voltage of between 22 and 27V, derived from a simple transformer, bridge rectifier and a pair of 4,700µF capacitors.  That capacitance is the minimum I recommend, although the circuits will all work happily with more or less capacitance.  I don't recommend anything less than 2,200µF under any circumstances for the main filter caps or ripple will be excessive.  If everything is in the same chassis, a toroidal transformer should be used to minimise stray magnetic fields which may cause hum.  The transformer needs to be rated for at least 80VA, but anything over 150VA is serious overkill and is not necessary.

The voltage is low enough to connect directly to a P05-Mini or similar.  The delay has its own on-board 5V regulator, and can be supplied with +12V from the P05-Mini.  The P05-Mini supply is not shown in detail, since it is fully described in the project page.

Figure 5 - Power Supply Wiring Example

The above shows the general idea for the power supply.  Provided the voltage is less than 30V (18-0-18V transformer), the P05-Mini can be connected directly to the main power amp supplies.  Make sure that the filter caps installed on the P05-Mini have a voltage rating suitable for the incoming voltage.  Using the 10 ohm series resistors as shown in the project article will reduce the ripple voltage that the P05-Mini receives.  I recommend that the supply voltages should be ±12V rather than ±15V as would normally be used, because a 15V transformer will be loaded more heavily than normal.  This may reduce the instantaneous voltage sufficiently (with high output levels) that 15V regulators may 'leak' ripple to the preamp supplies.  12V regulators have enough reserve to ensure this doesn't happen.  You can use 15V regulators if you use an 18-0-18V transformer, but there's nothing to be gained as the system has more than sufficient dynamic range with ±12V supplies.

The P05-Mini can be run from either the transformer's AC output, or from the DC.  If the input diodes are included the DC polarity is not important, as the diodes will 'steer' the DC to the right place, and this also lets you run the board directly from the transformer AC if preferred.  There's no gain though, as the transformer regulation is determined by the main load - the power amplifiers.

Printed circuit boards are available for the P05-Mini, P26A and P87A, as well as P19/ P127 power amplifiers.  This makes construction very straightforward.  Since I do not provide PCBs for power amp supplies (they are so simple that none is needed), that will typically be hard-wired.

The unit can be housed in any suitable metal case, provided it has sufficient heatsinking for the power amps.  Surround sound in a domestic environment rarely demands much power, so the heatsink can be the base of the cabinet, provided it's made from aluminium not less than 2mm in thickness.  A fully metal case is preferred to prevent any noise (especially hum) pickup from mains cables, etc.

The chassis must be earthed, and the RCA connectors must not connect directly to the chassis because that will defeat the loop-breaker (if fitted).  Including C1 (Figure 4) ensures that RF interference is shorted to the chassis.  Since crosstalk is not likely to be a problem with this unit, the wiring is not critical, provided that inputs and power amp outputs are well separated to prevent oscillation.  You must pay close attention to earthing - the power supply centre-tap (0 Volt line) must be connected securely to the case to prevent noise pickup.  Optionally, you can use an 'earth loop breaker' circuit as shown in the supply diagram above.  This circuit inserts a resistance into the earth (ground) circuit to prevent loops from causing hum.  Please be aware that it may not be lawful to include this circuit in some countries, so it is up to you to decide whether to include it or not.

Main Index

Projects Index