Emulating Dolby Surround Encoding With A Stereo Mixer

If you have a cheap Dolby Surround decoder in your home — say as part of a 'home cinema' entertainment system — and a stereo mixer, you have everything you need to start mixing in Surround sound. Hugh Robjohns explains how to enter another dimension...

It is hard to believe that the Dolby Motion Picture 4:2:4 matrix Surround sound system has been around for 20 years. The first well‑known feature film to use this format was Star Wars, back in 1977. Since then, there have been over 6000 films released in the Dolby Stereo format (as the cinema Surround system is known), and there are over 30,000 Dolby‑equipped cinemas worldwide to watch them in!

For those not familiar with the format, Dolby's 4:2:4 MP matrix combines — or encodes — four separate signals (Left, Centre, Right, and Surround, or LCRS), in such a way that they form a stereo‑compatible 2‑channel format for recording and broadcasting. The encoding process is relatively simple, with a useful side effect in that we can often take advantage of 'free encoding' with some normal stereo material (see the 'Free For All' box).

On playback, a decoder unit is used to separate the stereo‑compatible signal and re‑generate the original four LRCS elements. Unfortunately, any matrix system is inevitably compromised, and it is impossible to completely recover the original signals with perfect isolation — there will always be some crosstalk between channels. The decoder disguises this problem through a 'steering' process which emphasises the signal emanating from its appropriate loudspeaker by cancelling out a proportion of the crosstalk in adjacent channels. It is this 'active' steering process which is both the strength and weakness of the Dolby Matrix Surround system.

Note that however many speakers you decide to use to handle the rear‑channel Surround signal (normally two in a domestic setup, but many more in a larger venue — see Figure 1 to give you an idea of the layout of a typical TV‑based home system), they will all receive the one 'S' signal.

A Domestic Science?

In 1989, Dolby licensed an integrated circuit version of their Surround decoder for the domestic market, entitled the Pro Logic system. This replaced the previous, and rather inferior, passive domestic system, called simply Dolby Surround. The new active decoder opened the flood gates for Dolby Surround‑encoded material for domestic consumption: CDs and cassettes, radio and television broadcasts, video tapes, LaserDiscs, computer games — you name it. Dolby Surround‑encoded material is everywhere.

If you are in any doubt about the size of the Surround market, Dolby claim there are now over 21 million home cinema systems worldwide equipped with Dolby Pro Logic Surround decoders, with the numbers growing daily. Just have a look around your local electrical showroom at the number of TV sets, midi hi‑fi systems, hi‑fi receivers and dedicated separate Surround units that boast built‑in Pro Logic decoders. On top of that, most media software producers are now making Dolby Surround‑encoded products. You might be surprised at the number of TV programmes which are Surround‑encoded (although relatively few carry a Dolby logo) and again, the number is growing rapidly.

Come Out — You Are Surrounded

So, what does all this have to do with you? Well, if more and more people are equipped to listen in Surround sound, shouldn't you consider taking advantage of the extra creativity this format allows? Furthermore, if you are (or want to become) involved in producing background or theme music, jingles and sound effects for radio, television and computer games, Dolby Surround is increasingly the required format.

The first hurdle most people think will prevent them from producing Surround‑encoded material is getting hold of an encoder. Worry not — you don't need one, although there are a few practical advantages if you do have one. It is possible to purchase suitable encoders from a number of different manufacturers, including Dolby (about £1469 inc VAT for the professional SEU4 encoder) and RSP Technologies (£871.85 inc VAT for the Circle Surround Encoder) to name but two. The purpose of this article is to explain how the encoder works, and how you can manage without one using just a stereo mixer.

What you absolutely cannot do without if you're working in this format is a Pro Logic‑compatible decoder. There are lots of options here, starting at under £200 (inc VAT) for a basic domestic Pro Logic‑equipped decoder (check out Tandy's offerings). Denon, Marantz, Meridian, Sherwood, Sony, Yamaha, and countless other domestic hi‑fi manufacturers have an enormous variety of suitable units at an equally wide variety of prices.

The professional studio decoder from Dolby themselves is the £1469 SDU4, and the 'compatible' RSP Technologies system (the Circle Surround High Performance decoder) is £1030 inc VAT. The professional systems are often easier to integrate into a pukka studio setup because of the higher operating levels and balanced connectors, but the domestic systems often offer more flexibility (such as splitting the bass away from the centre speaker).

I cannot emphasise enough how important it is to monitor your Surround mix through a Pro Logic decoder. The key aspect about the decoder, as I mentioned above, is that it 'steers' signals to the various loudspeakers (the 'active' part of the decoder) and this can have very pronounced effects on the perceived balance and on instrument positioning (especially across the front). If you monitor through the decoder, you can tweak the mix slightly to compensate for its influence, but if you don't, your mix may suffer from some very unpleasant effects, and you will be none the wiser!

Encoding

The standard Dolby encoder is a relatively simple device — at least in overview — which allows the easy construction of a home‑made encoder. The standard encoder unit has four inputs and two outputs. The latter are referred to as Lt and Rt (it stands for 'Left Total' and 'Right 'Total'). The left and right inputs pass straight through to the outputs of the encoder with minimal signal processing (see Figure 2), and the Centre and Surround signals are combined with these in specific ways.

The centre signal is simply reduced in level slightly (by 3dB to avoid reducing headroom too much, and to ensure an even power distribution across left and right when listening in stereo), and is then added in the same polarity to both left and right channels equally. Exactly the same effect can be produced by simply panning a signal centrally between the left and right mix busses on your mixer (although in this case, the level reduction is often more like 4.5dB, but this will not significantly alter the results). Adding the centre signal to left and right in this way guarantees stereo and mono compatibility for the Surround‑encoded material.

The Dolby system processes the Surround signal in a much more elaborate way, but it is still possible to simulate the processing with little more than a simple stereo mixer. Firstly, the Surround signal is reduced in level by 3dB for the same reasons as the centre channel. It is then band‑limited to remove everything below 100Hz and above 7kHz, (a process which is repeated in the decoder). The reason for removing the frequency extremes is three‑fold: firstly, if left and right channels are not accurately level or phase‑matched in the distribution media, there will be a significant increase in the level of crosstalk to the Surround channel. By removing high frequencies from the Surround signal, it is easier to disguise this potential problem. Secondly, in real life, distant sounds from behind a listener are shielded from the ear canal by the pinnae (outer ears) which impose a roll‑off above about 7kHz. Mimicking this effect makes the Surround channel sound more natural and less obvious. Thirdly, removing the requirement for low frequencies (below 100Hz) allows small 'satellite' Surround loudspeakers to be used (which are cheaper and easier to install) — and the three front channels are quite capable of producing sufficient LF signals. Although this band‑limiting is performed in the encoder, it is also repeated in the decoder, so we can simulate the encoder processes without applying this filtering, relying on the decoder to impose the bandwidth restrictions instead.

After the level adjustment and band‑limiting, the signal passes through a modified Dolby B‑type noise reduction encoder, which applies 5dB of high‑frequency boost (a normal system would apply about 10dB). B‑type noise reduction is added because the Surround signal is delayed in the decoder (typically around 25ms in a domestic listening environment) and early systems even used analogue devices to perform this function. The 'bucket‑brigade' circuitry employed (so called because it achieves the required delay by passing the signal through many stages of short delay, somewhat like a line of firefighters passing buckets from hand to hand) are inherently pretty noisy, and the B‑type noise reduction helped to maintain the quality of the Surround signal as it passed through the decoder. Fortunately, the high‑frequency boost applied by this stage of the encoder processing is so small that it can be omitted with negligible detrimental effects in a simple home‑made encoder.

Finally, after all of the other processing stages, the Surround signal is split and each half is phase‑shifted by plus or minus 90 degrees (giving 180 degrees phase shift between them). These phase‑shifted signals are then added to the left and right channels. In simple terms, the Surround channel is effectively added in opposite polarities to the left and right channels, which is, fortunately, very easy to simulate in a home‑made encoder.

Home‑Made Dolby

It is simple to create a pseudo‑Dolby encoder from a stereo mixer, with pretty respectable results. Anything panned fully Left or Right on the main mix buss will only appear on the left or right output channel of the decoder — exactly as required. Anything panned centrally will be of equal amplitude and phase on left and right, producing exactly the same conditions as the real encoder and, once again, the decoder will route such signals to the Centre loudspeaker. The only thing to watch here is that panned centre signals may appear slightly too quiet in relation to other parts of the mix, but by listening to the programme through a Pro Logic decoder, you will be aware of any deficiencies and will compensate automatically as you mix. In this way, the normal channel pan pot can be used to pan a signal anywhere across the front three speakers, although it can be a little tricky to perform real‑time pans — a point I'll come back to shortly.

To get a signal to the rear Surround channel, you need to add it in opposite polarities to the left and right busses at the same time. A stereo channel on the mixer is often ideal for this — just arrange for a mono input to appear on both sides of the stereo channel, and switch in a phase reverse to one side. Hey Presto — the signal will be decoded to the Surround loudspeakers.

To make the whole process a bit easier and more practical to operate, you could use an auxiliary send buss (deriving the signal pre‑fade allows easier rear‑to‑front pans) to gather the Surround elements from the appropriate channels, and then plug this auxiliary output into the stereo channel to be combined with the Lt and Rt signals, as described above. There is a slight technical hitch here, however. Consider trying to make a sound 'fly' overhead from back to front by panning the signal between the Surround channel and the Centre channel. This could be done on our home‑made system by starting with the pre‑fade auxiliary send turned up and the fader closed (with the pan pot central). By turning down the auxiliary send at the same time as bringing the fader up, the sound will be panned from the Surround channel to the Centre channel. However, the centre signal is effectively (L+R) while the Surround signal is (L‑R). When the panning reaches mid‑way, the +R and ‑R signals will cancel, and the end result will be that the sound 'flies' around the left of the room rather than overhead.

To overcome this annoying effect, all we need to do is use a simple delay line to de‑correlate the Surround and front channel elements of any signal panned to multiple destinations. This delay is not needed when using a genuine encoder because of the particular way in which the Surround channel is added to the left and right signals using 90‑degree phase shifts. A good delay time setting which seems to work well with everything is 7ms — just long enough to avoid phasing effects, but not so long that it is perceivable as a delay. If you really want to do the job properly, you could use the stereo channel EQ to roll off everything above 7kHz and below 100Hz. (See Figure 3.)

Decoding

The job of the decoder is to try to extract the four signals from the 2‑channel combination. The Centre signal is detected by looking for anything which is at the same level and polarity in left and right (ie. the sum), while the Surround channel is found by looking for same‑level but opposite‑polarity signals (ie. the difference). The nature of the simple matrix encoder means that the centre signal is mixed in to both left and right, as is the Surround (albeit in opposite polarities). If you think about it, this means that there is inevitably a large amount of crosstalk between adjacent channels — Centre on Left, Left on Surround, Surround on Right, Right on Centre and so on.

To try to reduce this crosstalk, a pretty complicated sub‑system determines where the 'dominant' sound should be coming from at any moment in time (the dominant sound is any sound which is clearly louder than the background sounds). Having established which channel should be louder, moment by moment, the system tries to cancel out the crosstalk from each adjacent channel by adding in some of the dominant signal, out‑of‑phase. This creates far better image definition for the dominant signal, while re‑distributing the non‑dominant sounds over the other channels. You could argue — and many have — that this is not the ideal way to listen to Surround material (hence the recent development of various true discrete‑channel Surround systems), but it is cheap, and mightily effective.

One of the worst problems with this matrix system is that poorly‑matched Lt and Rt levels or phase will cause excessive Centre‑signal crosstalk to the Surround channel. In the cinema, this means dialogue coming from behind you — a very distracting effect indeed. To overcome this problem, Dolby included a delay in the Surround channel, so that even if there is crosstalk to the Surround channel, listeners will not hear it until after the frontal sound has been heard. This takes advantage of something called the Haas (or Proximity) effect, where the brain manages to ignore secondary sound arrivals (at least as far as directional cues are concerned), provided there is about 20ms difference between them.

Round Up

The simple DIY encoder system described in this article is a perfectly valid and practical way of experimenting with Surround mixing. I know of a number of experimental Surround‑encoded TV programmes which were produced with this very system before the company in question decided to commit to the proper Dolby encoder system (for copyright reasons, you must use the Dolby encoder to be able to use the Dolby Surround logo).

Provided you always monitor through a Pro Logic decoder (or equivalent), you cannot go wrong with mixing in Surround. If your efforts don't sound right, at least you will have spotted the problem, and armed with the knowledge of how Surround works, you should be able to adjust the mix balance to fix matters. Give it a go, and enjoy the extra dimension that Surround offers, both in the production and in the listening — I'm sure you won't regret it one bit!