Simple Surround Sound for Digital Organs
Posted: 10 January 2013
Last revised: 10 January 2013
Copyright © C E Pykett 2013
Abstract. This article discusses the Hafler-Gerzon loudspeaker arrangement in the context of digital organs or virtual pipe organs. The Hafler-Gerzon system was invented in the 1970's and it derives an additional loudspeaker channel from an ordinary stereo pair in a simple manner without requiring additional power amplifiers. The additional channel is the difference between the two original signals, and by applying the difference signal to a separate loudspeaker or loudspeakers placed elsewhere in the room, it can provide significant extra ambience by endowing the listening experience with an enhanced spatial feel. Although the system was originally aimed only at hi-fi, this article suggests ways in which it could be applied to electronic organs. Because of its simplicity, the system can be assessed against a particular application at minimal cost and with little effort.
Modifications to electrical equipment are discussed in this article. No responsibility is accepted for the consequences. You bear the risk yourself.
(click on the subject headings to access the desired section)
Surround sound, in which more than the usual pair of stereo speakers are used, has been with us for at least half a century. Today's digital surround systems are quite complicated and they require special source material in which all the signals for the various channels are encoded. But if we go back in time we can find some long-forgotten methods which had two big advantages: they were very simple, and they did not need any special encoding. Therefore they would work with a pair of ordinary stereo signals. In the early 1970's probably the best known and last of these systems was the Hafler-Gerzon loudspeaker arrangement. It is also extremely simple, requiring no additional amplifiers. However it is a mistake in this case to confuse simplicity with naivety or amateurishness because both Hafler and Gerzon were well known and highly qualified professional audio engineers of their day , and in their case it is more appropriate to regard their invention as possessing an elegance deriving from their experience and intellect.
But first let us see what relevance this has to digital organs as opposed to hi-fi. After all, a digital organ can be provided with as many loudspeaker channels as desired in principle, each with its own amplifier and each channel handling a separate signal - maybe a particular division such as the great or swell organ, a specific stop or group of stops, or even just a single note. So what benefits can surround sound confer in this context of such potential richness? In fact there are several, even disregarding the fact that most digital organs do not have lots of audio channels. Indeed, when there are only a few channels, the Hafler-Gerzon system can be a most cost-effective and worthwhile addition to the instrument. This is because it can extract the difference between two signals and route it to an additional loudspeaker or loudspeakers. So if there are only two audio channels available in a small organ, the difference signal between them can be routed to a third loudspeaker which could be placed behind the player to give an extra spatial feel to the overall sound. This could also be applied to a virtual pipe organ (VPO) which uses 'wet' stereo sound samples, the difference signal then representing largely the ambience of the auditorium in which the samples were recorded. Or in an instrument which uses artificial reverberation generated by an effects processor, the difference signal between a pair of audio channels would again contain mainly spatial information (albeit derived synthetically) which could be separately radiated. Yet another circumstance relates to the use of monaural samples which are panned note by note across the keyboard between a pair of audio channels, perhaps to simulate the C and C# sides of a pipe organ. This technique is used to imitate the dynamic spread of sound across a rank of pipes such as an en chamade reed. Extracting the difference signal from the two channels and applying it to an additional loudspeaker in another part of the listening room can considerably enhance the subjective effect of the instrument.
Before describing the system further it might be helpful to recall that from an ordinary stereo pair one can derive two additional channels which are seldom used, the sum and (as just mentioned) the difference signals. The sum signal is what you hear when you switch your amplifier to monaural mode, in which the two channels are simply added together at the preamplifier outputs. The difference signal is exactly the opposite but just as simple – instead of adding the signals together you subtract them using a differential amplifier. But before you reach for your soldering iron to knock one up, there is a much easier way to hear the difference signal without doing any electronics at all.
Figure 1. Deriving the difference channel from a stereo amplifier
Disconnect the loudspeakers from your amplifier and then reconnect one (or both) as shown in Figure 1. The speaker(s) are connected across the two 'live' or 'hot' output terminals, ignoring the two ground (earth) connections completely. If both speakers are used, they should preferably be connected in series with the positive terminal of one connected to the negative terminal of the other. They could be connected in parallel, but there is a chance this might overload the amplifiers because the load impedance which they see will be halved. If you now listen to a stereo source you will hear only the difference signal between the two channels. Incidentally, this will only work if your amplifiers have an internal common ground connection. If they do not, you will have to connect the two grounds together. However before doing this you must ensure that the amplifiers will tolerate it. Some might exhibit parasitic oscillations in this circumstance, which can be highly destructive to the output transistors or valves. Or the power supplies in others might be damaged. Therefore it is worth checking with the manufacturer of your amplifiers first. If you are unsure about the grounding arrangements of your amplifier outputs, switch it off, disconnect the speakers, and connect an ohmmeter on its lowest resistance range across the two 'negative' speaker terminals. If a dead short is registered, then they are obviously connected internally to a common ground line. If even a few ohms is registered, then this is probably not so and you need to proceed with caution as just described. But note also that the common connection, if present, must be capable of handling the significant currents demanded by loudspeakers. This might not be so if the connection only exists via the low level signal input circuitry, for example.
Listening to audio difference signals is highly instructive and it will be a novel experience for many people, particularly if you use source material you are familiar with in ordinary stereo. Therefore it is suggested you begin by listening solely to the difference signal from your hi-fi system rather than a digital organ. This can come later when you have more experience of what to listen for. Switching the amplifier to mono should produce an almost inaudible signal because the two near-identical output signals are cancelled, and any residual sound is due to things like imperfect matching between the tone control components in the preamplifiers. Listening with the amplifier in stereo mode, one finds that sound sources at centre stage appear much fainter than usual whereas those to one side or the other are less affected. Bass frequencies can also be much reduced because in many recordings it appears in roughly equal amounts in both channels. There are also considerable differences between most classical recordings and pop, because the former tend to use relatively undoctored signals from just a few microphones whereas pop music relies on multiple microphones, complicated mix-downs and much pan-potting. But even with classical music there can be pronounced differences between different types of material. A lone singer at centre stage might mysteriously seem to come and go as s/he moves her/his head. The echoes from a solo piano or harpsichord in some auditoria can sound almost like pistol shots, whereas the reverberation from a cathedral organ more often rolls smoothly away. In fact, it is the ambience of the recording room which is strongly represented in the difference channel, and this is where much of its value lies for surround sound purposes.
Having done this experiment you are ready for the next. Reconnect your front speakers in the normal way, and now add a third connected across the two 'live' terminals as before. Place this speaker at the opposite end of the listening room and behind you, desirably so that it does not fire straight at you from the rear. For example, it can be placed on the floor facing the ceiling. Now listen to your favourite recordings. Unless you are exceptionally unlucky, virtually all of them will now sound significantly more spacious, especially those recorded in a reverberant environment. Whether you like the effect or not is a matter of taste, but you can always comfort yourself with the thought that you have done nothing to distort the original stereo signals in any way. You are merely distributing the acoustic energy differently within the volume of your listening room. As simple variants on this experiment, you can also add another rear speaker connected in series or parallel with the existing one. The parallel connection will give the greater rear volume, though again you should be cautious about not overloading your amplifiers in this case. The two rear speakers can then be tried in different positions.
Having done all this you have arrived at the simplest form of the Hafler-Gerzon system. However the complete system is considerably more complicated, having the circuit shown in Figure 2. Note in particular that two rear loudspeakers are incorporated, and depending on the settings of the various controls these do not necessarily radiate identical signals. This is because the circuit includes separate volume controls for the two rear speakers and there is also an additional, rather important, control which enables the relative proportions of stereo and difference signals fed to the rear speakers to be varied. At one extreme position of this control (called 'spread'), ordinary stereo signals identical to those fed to the front speakers are applied to the rear ones. At the other extreme, the rear speakers experience mainly the difference signal as derived above. At intermediate positions of the spread control you get a mix of both types of signal. An exhaustive description of the system is not given here, mainly because it was fully presented in an article by Michael Gerzon in Hi-Fi News in 1971 which was also available on the Internet at the time of writing . Note that the component values shown here were designed for 8 ohm loudspeakers, though some are slightly different to those in Gerzon's original article because I found they worked better.
Figure 2. The Hafler-Gerzon rear loudspeaker circuit
As you can see, the circuit is simple, although the low value wire wound potentiometers can be difficult to obtain. At the time of writing (2012) Conrad Electronics supplied them in 4 watt versions (order code 444774) . This power rating should be sufficient to enable you to gain an impression of how the system performs provided the volume is not excessive. I used the system for some while with a pair of 75 watt amplifiers without any problems. The resistance values are written using the European convention which does not appear to be understood in some countries. Therefore to remove any doubts, 4R7 means 4.7 ohms and 25R means 25 ohms.
Some remarks about speaker phasing are necessary at this point. As shown in Figure 2, the rear speakers are connected to the amplifier common ground connection via the 'spread' control using their negative terminals. This must also be the case for the front speakers, i.e. their negative terminals must also be connected (directly) to the amplifier common ground . However if the front and rear speakers are of different makes it is worth checking that the polarity markings on the units have the same meaning. Connect a dry cell briefly to each speaker and check that the cones move in the same direction for the same applied polarity. If they do not, the polarity markings in Figure 2 should be reversed.
When building the system remember you are dealing with potentially quite high peak currents, so the wiring and terminal arrangements should not be too flimsy. A topological wiring layout is shown at Figure 3, and my prototype control unit is shown in Figures 4 and 5. Hopefully the diagrams are so self-explanatory that little further explanation should be called for.
Figure 3. Topological connection diagram for Hafler-Gerzon rear speakers (rear view)
Figure 4. Rear view of the prototype Hafler-Gerzon unit
Figure 5. Front view of the prototype Hafler-Gerzon unit
Several areas of application to digital organs were suggested in the introduction to this article, and more than one system can be incorporated if you have more than one pair of suitable audio channels from which difference signals can be derived. A property of the system which is simultaneously fascinating but exasperating is that you can spend most of your waking hours adjusting one or more of the controls. The settings also depend on where you sit in relation to the loudspeakers. However I have found that the rear volume settings need adjustment less frequently than does the 'spread' control. Probably the most objectionable aspect of the system is its tendency to reinforce occasional notes or frequency peaks in the music in an unnatural manner, though this is much reduced by ensuring the rear speakers do not fire straight towards your back as mentioned earlier. In my installation the two rear speakers are hidden behind convenient pieces of furniture in the corners of the room, firing from behind them up to the ceiling.
It is difficult to describe the subjective experience of a Hafler-Gerzon set-up, and because it is so simple the best thing is just to try it. Although the results are variable, the sound from the organ really can come alive in a way that is quite surprising given the simplicity of the idea. In such cases one comes to realise the gross artificiality of ordinary twin channel audio with its closely-spaced speakers which just fire at you from in front, regardless of how expensive the speakers and amplifiers might be. Note that augmented implementations of the Hafler-Gerzon arrangement also include a front centre speaker fed with a sum (monaural) signal, controlled with yet another volume control. Yet others include additional side speakers as well. I have not tried a front centre channel, mainly because it is only really relevant when you want to separate the left and right speakers by some considerable distance to increase the width of the sound stage. In this case the central speaker can be adjusted to fill in the acoustic hole which develops in the middle.
A technical shortcoming of the system is that it inserts resistance in series with the rear loudspeakers. This is an anathema in high quality audio engineering because it reduces the motional damping of the cones of low frequency speakers (woofers) by the very low output impedance of their power amplifiers, thereby increasing the Q-factors of resonant peaks and creating a less well defined bass response in other respects. In other words the cones flap around under their own inertia rather than their motion being tightly related to the input signal. Another problem created by parasitic source resistance concerns crossover networks, whose characteristics might be affected across the whole frequency range handled by the loudspeaker, not just the bass region. Therefore, on the face of it, some of the benefits of an expensive and well-designed speaker which is carefully matched to its enclosure will be obscured, implying that loudspeakers of somewhat lesser quality might be more appropriate here. Gerzon  discussed this issue to some extent. However the effects can be less important in this application than in hi-fi more generally because if there is comparable in-phase bass power in both channels, then it will be reduced by the subtraction necessary to form the difference signal. Nevertheless, if you perceive this to be a real problem, it can be solved simply by connecting the rear loudspeaker(s) directly to the 'live' outputs of the power amplifiers as described earlier in this article, although no control of their outputs will then be possible. Of course, a complete solution can be implemented by deriving the difference signal at the power amplifier inputs and then routing it to an additional power amplifier which drives the rear speakers only. Note that the problem does not exist for higher frequencies, particularly those which are handled by tweeters, because cone excursions at these frequencies are much smaller and so they do not call for the damping provided by the amplifier.
Finally, either you will like this system or you won't, but in any event it will not have cost an arm and a leg even if you decide not to use it. However I suggest that you do not dismiss it out of hand before trying it because it is so easy to hook up a spare loudspeaker or two for a quick trial.
1. David Hafler (1919-2003) was a well known American audio engineer who co-founded several companies including Dynaco and Acrosound. Michael Gerzon (1945-1996) was equally well known as the British pioneer of Ambisonics and the Soundfield microphone, having moved into the audio field after studying mathematics at Oxford.
2. "A Year of Surround Sound", M. Gerzon, Hi-Fi News, August 1971.
Also available at: www.audiosignal.co.uk/A%20year%20of%20surround%20sound.html and www.audiosignal.co.uk/Resources/A_year_of_surround_sound_A4.pdf
(both accessed 2 November 2012).
3. Conrad Electronics GmbH, PO Box 19, Bedford, MK40 2WU, England. Tel: 0870 732 3223. www.conrad-uk.com/ce/en (accessed 2 November 2012).