Time Delay for Ambience by Len Feldman (Dec. 1976)

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting

Nearly a year ago, I heard an early prototype of a consumer electronic audio time-delay unit developed by a company called Audio Pulse, Inc. This first audition was conducted in a rather small hotel room in Chicago and, by the time I had spent an hour or so listening to that room "expand" into various sized and configured concert halls, I was convinced that audio time delay would be the next major "wave" in the advancement of the audio art as it relates to home high-fidelity equipment. After several months delay, the Audio Pulse Model One digital time-delay system is in production and available to high-fidelity enthusiasts. A company called Sound Concepts, Inc. has come up with the Model SD-50 Audio Delay unit, which is also in production and available.

Both units purport to offer the same end results but, as we learned from a hands-on, in-depth study of each, both the circuitry and approach of each unit differs widely from that of its competitor. In terms of price, the Sound Concepts unit has a slight edge, with a suggested retail price of $600 as against Audio Pulse's current retail price of around $630.

Why Time Delay?

Some of the sound we hear at a live concert performed in a large space comes not from the stage where the performers are located, but reaches the audience after repeated bouncing or reflection from walls, ceiling, and floor. As sounds leave the stage, they begin to mix with reflections from almost every available surface in the hall. This process is so complete and complex that the listener sometimes hears as much or more reflected, delayed or reverberant sound than he hears direct sound from the players.

For realistic recreation of what has come to be called "hall ambience," it is necessary to both delay and reverberate the original music signals. A pictorial representation of a typical sound field is shown in Fig. 1. In addition to the time delays of the early and later reflections, the rate of decay of the entire sound field plays a part in our ability to identify a given hall.

Finally, as sound travels about the hall, bouncing off surfaces, its high frequency content is progressively absorbed and irregularly attenuated. So, to simulate this aspect of a hall's characteristics, any electronic time-delay device designed for home use must "roll off" the high frequencies reproduced by the additional channels that will be used to recreate the delayed/reverberant sound fields desired. Both the Sound Concepts and Audio Pulse units take care of these requirements in varying degrees by means of totally different circuit approaches.

Fig. 1--Representation of a sound field showing direct sound, echoes, and reverberation.

Digital Time Delay

One way to electronically delay an audio signal is to first convert it into a digital code by means of an A/D (Analog-to-digital) converter. The digital signal is then passed through a series of shift-registers to introduce required delay, after which the digital signals are reconverted, by means of a D/A converter, back to analog signals suitable for amplification and application to the extra rear or side speaker systems. This is the basis of the approach used by Audio Pulse in their Model One. An overall block diagram of one channel of the unit is shown in Fig. 2. Main Input jacks are coupled directly to the Tape Out jacks and to the Monitor switch, where either the Main or Tape input signal is passed to an input buffer stage. Its output goes to the Primary output switch and to a contour circuit which applies a boost of 6 dB/octave below 100 Hz when the back-panel Contour switch selects that option.

Fig. 2-Block diagram of the Audio Pulse Model One circuitry.

Fig. 3-Block diagram of the Audio Pulse Model One Delay/Reverberation circuitry.

Seven-step, switchable-gain amplifiers precede and follow the delay/reverberation circuits (about which more in a moment) so that the level of the incoming signal is matched to the dynamic range of the digital delay circuits. The use of input and output gain controls (which are matched) provides unity gain operation of the system. The level-adjusted signal at the input to the digital delay circuits also feeds a series of 12 front-panel LEDs which show when the user has selected that gain setting which is ideal for the digital delay circuitry. The delayed and reverberated signal is then fed to a mixer where it is mixed with the primary (undelayed).

signal and fed to the Delay position of the Primary (main channel) output switch.

The delayed and reverberated signals in both channels are also fed to an inter-channel mixing and phase shifting circuit. Audio Pulse maintains that the reverberant-field energy arriving at a listener's ears is largely incoherent and random-phase in nature. The output circuitry of the Model One therefore includes a phase shifting matrix intended to ensure that the delayed signals will have a non-localizable character. The resultant output signals are fed to the Secondary output switch, where the user may select either the direct or delayed signal to be fed to the secondary channels and speakers.

Fig. 4--Block diagram of the Sound Concepts SD-50 with one channel in the Stereo mode.

A detailed block diagram of the "Delay Circuits" block of Fig. 2 is shown in Fig. 3. Again, only one channel is shown. The input signal from the Level Match controls goes to a buffer where it is mixed with recycled opposite-channel signals. A low-pass filter attenuates frequencies above 8 kHz. The audio signals are then encoded into digital on-off pulses which are then fed into shift registers. When the pulses emerge from the shift registers (time delay is proportional to the number of shift registers used), they are decoded to recover the audio signal. Four different delays are obtained, two in each channel. The delayed pulse trains are decoded to yield audio signals. In each channel the fully and partially delayed audio signals are fed to a variable mixer. The relative proportions of partial and full delays are controlled via a front-panel bank of selectable Decay Time switches. When any of the Decay Time switches is engaged, the delayed output from each channel is fed back and mixed into the output of the opposite channel, where it is further delayed and then fed back to the input of its original channel and so on. In addition to the cross-channel recycling loops, there are also recycling loops within each channel (not shown in the block diagram). When each signal is recycled, it is mixed into the input at a reduced level, corresponding to the attenuation of a sound wave as it is reflected off the walls of a concert hall.

In addition to being fed to the decay time mixer, the fully delayed and partially delayed outputs of the digital-to-audio decoders are also fed to additional rear-panel Long and Short output jacks. These taps are within the recycling loop and contain reverberant signals, but their echo patterns are different from those at the Secondary outputs. The extra taps may be used to create a six-channel or eight channel system or, when the recycling is cut off (by disengaging all the Decay Time switches on the front panel), the taps may be used to provide discrete, non-reverberant time delays for use in sound reinforcement applications.

Analog Time Delay

The Sound Concepts SD-50 uses a completely different electronic approach to achieve discrete time delay.

A block diagram of one channel of this unit is reproduced in Fig. 4. A high input-impedance, operational amplifier stage, including the input level-set control, is followed by a low pass filter to block out any distortion causing FM subcarriers or other high frequency leakage signals. Next comes a high-frequency pre-emphasis stage and a 2-to-1 compression circuit. This circuit halves the dynamic range of material presented to the SD 50 input and helps to reduce internally generated noise to below audibility. The compressed signals are passed to a delay chain of charge coupled devices in large-scale integrated circuit (LSI) form. These so-called bucket-brigade chips operate at a rate determined by a digital clock, with the clock's frequency determined by the front-panel Delay control.

From the bucket-brigade chips, the delayed signals (which have remained in analog or true audio form through the process) pass through appropriate buffering to a 1-to-2 expander circuit

the complement of the compression circuit on the input side of the delay chain. Here, the signal is restored to its original dynamic range and de-emphasized. The combined compander action serves to effectively double the signal-to-noise ratio of the SD-50's delay chain.

At the output of the expander, the reverb signal is picked off, buffered, additionally rolled off, and its level chosen by the front-panel Reverb control. This reverberation component signal is passed on to the input of the opposite channel's compressor system. Thus, the reverb input to the right channel is the delayed left-channel signal, and vice versa. In this regard, the design philosophy of the Sound Concepts unit differs fundamentally from that of the Audio Pulse, since rear channel delayed and reverberant information remains coherent, rather than random. Following beyond the reverb pickoff point, there is a ganged potentiometer which serves as a level control operating on both' channels simultaneously.


Table I-Audio Pulse Model One specifications.

Input sensitivity for 0 dB: Seven levels, selectable via Level Match control-0.12, 0.2, 0.3, 0.5, 0.8, 1.2. 2.0 Volts rms.

Input impedance: 50 kOhms.

Output level: Primary output identical to input level (unity gain); Secondary output variable from no output up to the input level.

Output impedance: Primary output, 500 ohms; Secondary output, 1000 ohms.

Maximum indistorted input level: Direct outputs (both Primary and Secondary)

4.5 volts rms; Delayed output, frequency-dependent, 5 dB above "0 dB" at 1 kHz.

Noise level (20 Hz-20k'i-lz, A-weighted): Direct outputs, 85 dB below 1 volt; Delayed output; better than 65 dB below max. input level at 1 kHz (typically 68-70 dB). Frequency response: Direct outputs, 10 Hz-40 kHz±0.5 dB; Delayed output, 20 Hz 8 kHz ±3 dB; frequencies above 8 kHz are attenuated by 18 dB/octave filter.

Distortion: Direct outputs, 0.05% THD; Delayed output, under 1.0% THD (measured with 1 kHz 0 dB input). Initial delay: Four simultaneous delays ranging from 8 to 94 mS. Reverberation decay time: Variable from 0.2 to 1.2 seconds (measured as the time required for "reverberant" output to fall by 60 dB following a transient). Echo density: Maximum interval between successive delays is 20 milliseconds; typical interval at most settings of the Decay control is under 10 milliseconds.

(Echoes spaced more than 30 mS apart would be heard separately; closely spaced multiple delays blend into a lifelike ambience.) Coherence of delayed outputs: Frequency dependent, typically under 20%. (A low percentage is desirable, reflecting the fact that in a real acoustic space the reverberant-field energy arriving at the listener's ears is largely incoherent and random phase Li nature.)

Dimensions: 14 1/2 in. W x 10 in. D x 4" in. H. Weight, 10 lbs.

Power requirements: 110 Va.c., 60 Hz, 20 Watts.


From the description of the two circuits just given (and they are by no means complete, in the interest of brevity), it is clear that the Sound Concepts unit is somewhat simpler in overall design than the Audio Pulse unit. The front panel arrangement of each tends to confirm this (see photos). The added complexity of the Audio Pulse unit arises, in part, from the nature of the digital encoding and decoding system which requires, among other things, a careful level match (more critical than in the case of the Sound Concepts SD-50) and which utilizes discrete steps of delay and reverberation or decay as opposed to the continuously variable delay and reverb controls featured in the Sound Concepts unit. A reading of each manufacturer's published specs sheds some additional light on the operation and performance capabilities of each of the units, so these have been reproduced in their entirety in Tables I and II. Though not stated in Sound Concept's listed specifications, elsewhere in their owner's manual we are told that delay is variable from 5 to 50 milliseconds in stereo, from 10 to 100 milliseconds when used monophonically.

Laboratory Measurements

Units such as the Audio Pulse Model One and the Sound Concepts SD 50 do not lend themselves to very many meaningful test measurements.

Fig. 5-Time delay of tone burst with the Sound Concepts SD-50 Delay control set to (a) 5 mS and (b) 50 mS.

Ultimately, both of these units are best judged by studied listening tests, since their primary objective is to provide the spacious feeling of a concert hall when music is reproduced in more confined listening environments. Nevertheless, before proceeding to our protracted listening experiments, we checked out certain basic performance criteria of the two units, such as distortion (of the secondary channels only since front-channel signals, in both cases, remain unaltered), signal-to-noise ratios and frequency response. We examined the primary delay action of these two units with dual-trace scope photos and tone bursts to depict the delays.

In all of the scope photos that follow, sweep rate was set so that the time of a single trace is approximately 50 milliseconds. The upper trace represents an input signal, while the lower trace represents the output from a delayed channel. In Fig. 5(a) we see approximately 5 miliseconds of delay when the SD-50's Delay control was set to its minimum point, while in Fig. 5(b) the delayed signal has shifted over so that it occurs approximately 50 milliseconds after the input signal-the result obtained when the Delay control is moved to its maximum setting. In both cases, the Re verb control was set fully counterclockwise to the zero position. To examine the action of the Reverb control of the Sound Concepts SD-50, we rotated it to about mid-point and reset the Delay control to its minimum (5 milliseconds). The resulting photo (Fig. 6) clearly shows how a series of pulse-trains appear following the initial delay, each diminished in amplitude and spaced apart by the basic 5 milliseconds.


Table II-Sound Concepts SD-50 specifications.

Power Requirements: 115 ± 7 V a.c. at 3.5 watts.

Front-channel input impedance: 100 kOhms nominal, 60 kOhms minimum.

Front-channel input voltage for full output: Set for 1 Volt, adjustable from 0.2 to 20 Volts.

Front output: Connected directly to front input for connection convenience.

Rear-channel output impedance: 6 k-Ohms nominal, 10 k-Ohms maximum.

Rear-channel output voltage: 10 Volts rms maximum.

Rear-channel output gain: Set for 3.3 (10 dB) at maximum setting of Level control; adjustable with Adj. Peak and Adj. Bal. controls.

Rear-channel frequency response: ±1.0 dB 30 Hz to 4 kHz with Delay control at 5 milliseconds and Rolloff set at Flat.

Rear input: Connected directly to rear output in Ext. mode, otherwise no connection.

2-Ch. mix output impedance: 300 ohms nominal.

2-Ch. mix output voltage: Twice the front input voltage plus amount of signal set with Level control.

Signal-to-noise ratio: Typically over 70 dB* depending on settings of Level, Rolloff and Delay controls. Over 55 dB* at maximum settings (affected by Peak and Bal. adjustments).

Distortion: Under 1% most frequencies and listening levels.

*Early owner's manuals listed these S/N figures; the current manual specifies 90 dBA and 85 dBA, respectively, for these two figures.


Fig. 6-A 5-mS Delay plus Reverb on the Sound Concepts SD-50.

Fig. 7-Delay observed with the Audio Pulse Model One when Delay button is set in (a) Short and (b) Long positions.

In the case of the Audio Pulse unit in order to obtain equivalent time-de lay-only photos, it was necessary to disengage the Decay buttons and to measure output at the extra taps (where the time-delayed but un reverberated or mixed signals are available). Figure 7(a) shows one of the four delays obtained with the Delay button in the Short position, while in Fig. 7(b) we see the additional displacement of the output signal when the Long delay position is selected. The random nature of the decay and cross mixing of signals is depicted to some degree in the scope photos of Figs. 8(a) and 8(b) which were taken with different Decay buttons depressed, but with the basic Decay button set to the Short position.

While the rear or secondary channel frequency response of the Audio Pulse unit is essentially constant regardless of decay and delay settings, as shown in the graph of Fig. 9 (roll-off occurs, as specified, above around 8 kHz), the high-frequency response of the Sound Concepts unit is dependent upon the amount of delay introduced. The curves in Fig. 10 depict this rear-channel response with the roll-off front panel control set to Flat and with the delay control set to its two extreme positions, 5 milliseconds and 50 milliseconds and it is clear that the SD-50 rolls off at the high end much earlier than does the fixed response of the Audio Pulse unit.

Distortion measurements were also made for both units. In the case of the Sound Concepts SD-50, distortion varied both with frequency and setting of delay. At 1 kHz, THD ranged from 0.21 percent (at maximum delay) to 0.45 percent (for minimum delay) for an input of 1 Volt and gain adjusted to unity by means of the front panel level control. As received, the unit is adjusted so that a maximum of 10 dB of gain is realizable by means of this level control, but if signals tend to cause the peak-LED indicator to flash excessively, it is possible to readjust the input level match and thereby to change the overall gain-range achievable by means of the front-panel control.

In attempting to measure THD of the delayed channels of the Audio Pulse unit, we found that the lengthy discussion and cautionary advice regarding the level-match buttons and associated LED indicator bank were well worth reading. The digital circuitry of this unit is easily overdriven and can deliver reconstituted audio signals containing high levels of distortion unless care is taken to select the correct level-match button on the front panel. Of course, we were using as a source a signal generator capable of putting out a very wide range of signal amplitudes and so, perhaps, this problem was a bit mo re tricky for us than would be the case once you installed the unit with a given set of components where signals (from Tape Out or Preamp Out) are likely to be more nearly constant. In any event, once we got over that hurdle and optimized the Level Match, THD at 1 kHz varied from 0.3 to 0.65 percent, depending upon the combined settings of the Initial Delay or Decay buttons.

With 1 Volt into the SD-50, signal to noise ranged from 78 to 80 dB depending upon the setting of the Delay potentiometer. For the same signal input level, the S/N of the Audio Pulse unit ranged from 67 to 71 dB, depending upon Delay and Decay button settings. Overload of the SD-50 occurred with an input voltage of 3.0 Volts, a signal level not likely to be encountered in actual use with an integrated amplifier, receiver or separate preamp/power amplifier setup.

Fig. 8--Random nature of lower amplitude decay signals (lower traces a and b) show the echo density of reverberant signals obtained with the Audio Pulse Model One.

It should be noted that in addition to the normal high-frequency roll-off of the Sound Concepts unit, a front panel control offers addition high-frequency attenuation settings labeled

-3,-6 and-9. With the delay control set to minimum, the rear channel response tor these various roll-off settings is plotted in Fig. 11.

Use and Listening Tests

The question of speaker placement when using extra channels of time delay is dealt with somewhat differently by these two manufacturers. While they do not say so outright, we gathered that the Sound Concepts people favor the familiar two-in-front, two in-the-rear speaker arrangement common to four-channel setups. Audio Pulse, on the other hand, seems to favor side-firing arrangement of the secondary channel speakers and suggests that they be placed above listener's ear level. Needless to say, we experimented with these and several alternate speaker settings in our tests.

Our own conclusions (and they are far from the last word) were that we liked the results obtained with the Audio Pulse unit when rear speakers were positioned as recommended by that company, namely slightly behind us, side firing from the two side walls, and positioned about six feet above floor level. In the case of the Sound Concepts units, while we started with the speakers at the rear, in the usual four channel arrangement, we quickly altered that arrangement so that the rear speakers, though still against the rear wall, were angled inward and somewhat removed trom side wall surfaces (front radiating speakers were used in both cases). One thing we quickly discovered is that the illusion of a large space can be readily obtained with lower-cost, lower-quality speakers than those used for the primary stereo channels.

The units we used for the secondary channels had little useful response below 60 Hz. There of course, a tendency to over-use the rear channels before one becomes familiar with how effective the system can be without those rear channels becoming obtrusive or overly discrete in the sounds they produce. With the aid of our handy Russound/FMP QT-1 switch box (which accommodates as many as four 4-channel devices all via a single tape-out/tape-in facility on our main reference system), we were able to make direct comparisons between the two units on an A-B basis.

Of course, the number of control permutations on each delay unit is almost countless, and it took some time to adjust each unit so that the variables reduced to design approach, rather than level or basic timing differences.

The only instrumentation used to make such adjustments was our own two ears.

For relatively short decay times, we were able to obtain what we considered to be amazing hall realism with either unit. There are differences, to be sure, primarily we think because of the incoherency of the rear-channel information of the Audio Pulse units and the discrete left-right sense of rear channels on the Sound Concepts unit, but both expand the apparent listening room in a quite dramatic manner.

The Sound Concepts unit, when played with maximum delay and nearly maximum reverb or decay, tends to take on a bit of artificiality--suggesting almost--that something "electronic" rather than a natural phenomenon is taking place. In its Long delay and extreme Decay position, the Audio Pulse unit fared better in this respect, as we seemed to be transported to a cavernous cathedral.

The Sound Concepts unit is, for the most part, easier for a consumer to use properly, we feel, requiring little indoctrination and experimentation.

The Audio Pulse unit is a bit trickier, simply because so many sonic things are happening at once, but it is that very feature of the unit which provides so many subtle variations in the apparent sound field.

There was no clear winner in this back-to-back evaluation of these two interesting and innovative components. Both the Audio Pulse and Sound Concepts time-delay units seem destined to find consumer acceptance from those who recognize the limitations of a home listening environment (as compared with concert hall music listening) and are prepared to take the trouble and spend the money to augment their stereo systems with a device of this kind plus the necessary extra channels of amplification and the extra pair of speakers required. Although $600 or $630 may seem like a high price to pay for this added dimension and realism in sound, when you consider the fact that just a couple of years ago any electronic time-delay system (even those which did nothing but offer discrete time delay without reverb/decay capability) available for use in studio applications cost several thousand dollars, the accomplishments of both Audio Pulse and Sound Concepts are considerable. Audio time delay has been described by some as a viable alternative to four-channel sound. From our brief experience with these two units in our lab we would say that, at least as executed in the Audio Pulse Model One and the Sound Concepts SD-50, it offers a realization of four-channel's original goal-that of recreating the live listening experience-a goal that was regrettably passed over by four-channel's very earliest proponents in favor of spectacular effects that had little in common with real music.

Fig. 9--Response at outputs of secondary channels with the Audio Pulse Model One.

Fig. 10--Response at the rear channel outputs of the Sound Concepts SD-50 varies with setting of the delay control. The front panel Rolloff control is set to Flat.

Fig. 11-Variation of the rear channel response of the Sound Concepts SD-50, with the different Roll-off settings on the front panel control. (Delay set to 5 mS.)

(Source: Audio magazine)

Also see: Miking the PRO Way (Nov. 1977)

= = = =

Prev. | Next

Top of Page    Home

Updated: Wednesday, 2017-06-21 21:50 PST