In previous issues we have talked about sound reinforcement systems, line arrays, subwoofer arrays and new technologies in the fascinating world of live sound. A fundamental element for the proper operation of sound systems is the loudspeaker management processor. These digital processors replace the old analog crossovers.
They add functions that until now were practically impossible to imagine. All of this is possible thanks to digital audio signal processing technology: DSPs (digital signal processors), calculations and algorithms running in real time.
It goes without saying that these units are all the rage today, and from domestic manufacturers, through budget brands like Behringer and Alto, to high-end brands like BSS, XTA and DBX, everyone now has one or more models of these processors.
The leading speaker brands have processors for their own gear, which also include libraries of parameters specific to each cabinet and system model, as with Electro-Voice, Meyer Sound and Nexo, among others. These units sit between the mixing console and the power amps and are indispensable in traditional PA systems or line arrays.
In the photos below we see a BSS Omnidrive Compact Plus digital processor and a Meyer Sound LD3 analog loudspeaker processor.
LOUDSPEAKER PROCESSOR BLOCK DIAGRAM
The different elements that make up a processor are:
- Routing (of inputs and outputs)
- Parametric EQs
- Delays
- Filters
- Gain stages
- Limiters
PURPOSE AND OPERATION OF THE PROCESSORS
When we have a sound system with multiple ways (highs, mids and lows), we need to separate these frequencies so they can be reproduced properly and within the working range of each speaker.
For example, if we had a basic system made up of mid-high cabinets and subwoofers, we would need to configure our processor as a two-way.
Routing
The processor's inputs would come from the console's master output. So if our processor has 2 inputs, labeled L and R, and 4 outputs, labeled 1, 2, 3 and 4, those outputs could be assigned as follows:
- Processor L > Console Output L
- Processor R > Console Output R
- Processor 1 > Sub L amp input
- Processor 2 > Mid-high L amp input
- Processor 3 > Sub R amp input
- Processor 4 > Mid-high R amp input
Gain
The levels of each way must be identical in order to get a proper crossover point.
The crossover point is the zone where the ways will be at the same level, and it doesn't always coincide with the electrical crossover frequency, which can be wrong. In the example above, this zone is usually between 60Hz and 160Hz.
Filters
The upper and lower limits of each way's frequency range are determined by the type of filter we use. Filters are classified by order, which is related to dB per octave, as follows:
- A 6dB/octave filter is a first-order filter
- A 12dB/octave filter is a second-order filter
- An 18dB/octave filter is a third-order filter
- A 24dB/octave filter is a fourth-order filter
Most processors offer Linkwitz-Riley, Butterworth and Bessel filters of different orders, although what never changes is the relationship between order and slope per octave. All filters introduce delay, due to their very nature, and it depends on the filter's order.
Delays
The next controls we find are the time controls: delay. These controls let you add delay at the inputs or the outputs. They are expressed in milliseconds, although you can switch the scale to meters. With the time adjustment in the processors we can correct for the spatial offset between high-frequency drivers and low-frequency speakers. In practice, we should remember that 3ms is 1m. This will help us a lot when working with our processor and also when understanding some of the acoustic phenomena that occur in live sound.
That 3ms comes from rounding 2.94ms, that is:
T = 1/340 where 1 = 1 meter and 340 = 340 m/s
340 m/s is the speed of sound at 15º C
In most cases there are timing problems in the speaker cabinets on the market, due to flawed designs. The timing problem ranges from poorly calculated delay in the design of a cabinet's filters to the offset and misalignment of the drivers.
The main consequence is a phase shift in the cabinet, so some frequencies are reproduced with a certain delay relative to others. As incredible as it may seem, this happens with practically every cabinet on the market. The seriousness of this phenomenon has to be taken into account when the minimum values are exceeded: a phase shift of up to 90º at some frequencies is tolerable, but a larger one would have more serious consequences.
Some brands pay close attention to the phase of all the cabinet models they build so the models complement each other, keeping consistent values and complementary phase. And since we're talking about delays, we can't ignore the latency in the processors, which is due to the analog-to-digital input converters and digital-to-analog output converters. Incidentally, in the most modern processors resolution reaches 24 bits and the sampling rate 96kHz.
In some BSS models, after taking the relevant measurements, we get latencies ranging from under 1 ms to over 2 ms; the same is true of other models, such as XTA, and of digital audio gear in general. Is this serious? Well, if we take this latency into account when making our acoustic measurement to tune the processor, no problem, but if we don't, our adjustments will be wrong. Imagine we ignore this latency and it's 2ms: we'd be tuning the system with an error of 66cm!
There are also many loudspeaker processors with an AES/EBU digital option that let us go from a digital console straight into the processor with no latency at its input, although with minimal latency at the processor's output.
Parametric EQs
To touch up the EQ and frequency adjustments of the signal at either the input or the output, depending on the model, we have several parametric EQs, that is, EQs where we can vary the frequency, the gain and the ratio of center frequency to bandwidth, the “Q” factor or selectivity.
Obviously, we should touch these parametric EQs as little as possible if we don't want to alter the phase even further.
Polarity and Phase
Processors have an option to flip the polarity at the inputs and outputs, which gives us the choice of 180º or 0º.
There is a lot of confusion between polarity and phase in the audio world. Polarity is the positive or negative value of the signal, while phase is the time difference between frequencies. So a polarity change is 0º or 180º, while a phase change runs from 0º to 360º, for as many cycles as there are. Some processors also have a fine phase adjustment that lets us vary the phase from 0º to 360º, which is quite useful when finishing the system alignment so that the filter slopes match.
So if, after applying delay or filters between ways, we still have a small phase difference, it can be left at 0º, so the ways sum at the crossover point. For each phase value between audio signals there is a result:
- A 0º difference results in a 6dB sum
- A 38º difference results in a 5.5dB sum
- A 90º difference results in a 3dB sum
- A 120º difference results in a 0dB sum
- A 150º difference results in a 6dB subtraction
- A 180º difference results in maximum cancellation
- A 210º difference results in a 6dB subtraction
- A 240º difference results in a 0dB sum
- A 270º difference results in a 3dB sum
- A 322º difference results in a 5.5dB sum
- A 360º difference results in a 6dB sum
And this relationship stays identical for every cycle of difference between signals. The phase and magnitude response would be seen on an analyzer. The upper graph represents magnitude, that is, the resulting frequency response, while the lower graph represents phase; on its vertical scale it starts at -180º, the midpoint is at 0º and the top of the graph reaches 180º.
This way of plotting from -180º to +180º is very useful for representing phase, since the center is the 0º line. It is essential to have magnitude and phase analyzers to see what we're doing in real time as we move delays, filters, etc.
The human ear could never perceive this with such resolution, even though we can notice pitch differences of 12 semitones, which is equivalent to dividing an octave into 12 different signals. It is easier to hear differences in sound when we nudge the parametric EQs slightly than if we did the same thing with time or phase.
Phase analyzers exist as software, such as Smaart Live or SpectraLab, for PC or Mac, and FOH, SpectraFoo, etc., for Macintosh, and they are indispensable for tuning sound systems. Other options include specific hardware, such as Meyer Sound's SIM3, with very high resolution.
Limiters/Compressors
These protect each element of the PA system through independently connected outputs.
We set the level we don't want to exceed with a threshold, the attack and release times in milliseconds, and finally the output/input attenuation ratio, which will be very high if we want to limit.
Extra Features and Curiosities of Some Processors
The measurement mic is an important element when taking measurements, and some processor models have a dedicated input for connecting one, as well as pink noise generators and even an RTA (Real-Time Analyzer).
They may also have Auto-EQ, a function that generates pink noise at the outputs and, based on the signal returned to the mic, the parametric EQs correct that signal to flatten it.
Antifeedback is another extra function found in some processors. It consists of very narrow filters called “notch” filters that can be as narrow as 1/100 octave, able to discriminate frequencies in 1Hz steps, which let you eliminate feedback. We raise the volume until feedback occurs, the filter eliminates it, and as we keep raising the level, further narrow filters keep appearing and eliminate each feedback frequency in the order it shows up.
Some include a PC Card slot for PCMCIA cards, which lets us save the data stored on those cards. Another function high-end processors offer is the ability to adjust for the existing temperature and humidity, either directly on the unit or using a special probe connected to it.
There is a specific analog processor for line arrays, mentioned earlier, called the LD3, in which you can select the number of cabinets used, model, temperature, humidity, height, gain and distances for each send, filters, etc.
Connections
Input connectors range from 2 to 4 XLRs while outputs range from 2 to 8 XLRs. We also find MIDI inputs and outputs, as well as RS232, which allow these units to be controlled from a computer.
The latter connection is also possible over Ethernet, and even Wi-Fi.
Software control depends on the manufacturer: BSS uses SOUNDWEB, XTA uses AUDIOCORE, DBX uses DRIVEWARE and Meyer Sound uses COMPASS. Another very interesting possibility is to control the parameters of your processor directly from the SMAART LIVE tuning program and see the results of the adjustments there. On top of that, it is enormously convenient to be able to drive the processors without touching their displays and buttons, and so avoid bothering the mix engineer.
Practical Use of Processors in a Line Array System
Imagine we have a sound system like the one in figure 2, laid out as follows:
· Upper left line array
· Lower left line array
· Upper right line array
· Lower right line array
· Upper center cluster
· Middle center cluster
· Lower center cluster
· Front-fill
· Left subwoofers
· Right subwoofers
· Left monitor cabinet for PA control
· Right monitor cabinet for PA control
The PA console's output connects to the processors' inputs, while the outputs are split into a total of 12 sends in this case, so we have full control of all the parameters described so far for each send.
LATEST TECHNOLOGIES
The latest thing brands like XTA and DBX are doing is increasing their number of inputs and outputs to 4 and 8 respectively, for example the XTA 428 or DBX 4800, in addition to offering ever more possibilities for wireless remote control via PC, tablet, PDA, etc.
One of the most recent processors to arrive (when this article was written) was Meyer Sound's GALILEO, which gives you 6 analog or AES/EBU digital inputs and 16 outputs, so, for example, with this processor alone we could have tuned the previous 12-send system.
It includes temperature, humidity and altitude adjustment, etc. on each output, and can also control line-array-specific settings such as the length of the array. Through its Ethernet port we can connect it to a laptop and even to a network by configuring its IP, wired or over Wi-Fi.
