How we voiced Spartak Stadium
At the Otkritie Arena stadium, 45 360 people are accommodated - this is the audience of a good post on Habré. Photo at the beginning of our work.
Approximately 9% of users during the match are busy screaming, making noise and making fireworks. 90% of TK carefully listen to the commentator’s speech and advertising, and the remaining 1% spit and says that they would make the sound better.
It was these sounds (advertisements and speeches) that we had to convey to each viewer so that, on the one hand, his eardrums would not burst, and on the other, he heard everything normally and clearly. That is, it was necessary to ensure uniform coverage and high speech intelligibility for everyone.
The stadium can be any. An open arena, like the former Dynamo or Luzhniki stadium (to the Luzhkov roof), or partially indoor, like most modern stadiums, where there is a roof only above the audience. Or even completely covered, like hockey arenas or such ambitious Fisht projects, albeit with a temporary roof. There is no universal recipe for good sound in all these cases. The fantasies of acoustic engineers are usually limited, on the one hand, by the budget, and, on the other hand, by the requirements of the technical specifications, to combine the system with the control system according to all Russian standards and the possibility of placing equipment. We are going to voice the Spartak Stadium.
Omit financial interactions with the customer. Just note that any sane person, before making an offer on the voice acting of the stadium, will prepare a computer model. Without it, it’s difficult to get into TK, even approximately, if there is no experience behind. Of course, there is a nuance of modeling - this is the accuracy of the model, and the characteristics of the materials used, and the models of acoustic systems, and noise accounting, and the mathematical modeling method, and much more. And many more reasons to challenge the correctness of the calculations. But this is the only thing that will give a result in a situation where your object looks like this, at best:
Of course, you can wait for construction readiness, take measurements and put real RT60 measurement results and impulse characteristics at control points into the model. And that will be right. This will be as correct as possible. But building is a funny thing, surrendering a day after they give you tension and paint over the last flight of stairs. All systems are surrendered at the same time, do not look at the contracts, do not write letters - this is a reality.
Based on the architectural drawings of the object, we were able to reproduce its shape, but the second problem remained - materials. And if plastic seats, concrete structures, a lawn could be taken from reference materials, then it was difficult with the rest (for example, it was impossible to say offhand what was the absorption and dispersion coefficient of the rubble covering the roof of the western sector). By the way, under the field itself, an incredible amount of communications has been piled up, but for us only the top layer with grass and ground is important - the rest did not affect the distribution of sound.
The following was important:
- Maximum uniformity in direct sound coverage.
- The direct sound energy of speakers should always be greater than the reverb tail (at least in the first 50-80 ms, very simplified). Those. We look at the value of the coefficient C50 and C80.
- The stronger the systems are directed, the greater the chances of getting into TK.
In general, when we talk about sound, we need to understand that sound is not light. These are 10 octaves, these are waves that behave differently - some quickly die out in the atmosphere, others go around everything and everyone. At a distance of 30-40 meters - not the limit for stadiums - you can see all sorts of things.
So, when the model was ready (it is convenient to draw in SketchUp and then export to EASE 4.4 or Bose Modeler 6.8), we wrote down the boundary conditions and drove it into the calculation software. Preliminary data were disappointing: the reverberation time ranged from 5.5 seconds at a frequency of 512 Hz. This is a lot. Imagine: you turn off the music, and the last chord sounds for another 5 seconds. But it was spelled out in the statement of work - to organize a sound system without changing the architecture or introducing additional materials.
Ok, another week of work together with the equipment manufacturer, and a more or less real scheme of the arrangement of systems and their models loomed. The expanses of the stadium are long tracks. Long cable routes made of refractory cable. Loading a long line with great power is stupid. We’ll lose more in the cable. You need a high-impedance load, either 100 V (but in the transformer sound there are claims to quality and a power limit of 400 W), or 16 or 8 Ohms. It would be nice to split the system into well-directed satellites and a low-frequency section, which can also be directed by EndFire or a cardioid so that the bass does not hit the roof. Of these conditions, there is not such a large selection of decent brands - the installation version of the expensive JBL PD series or the horn ElectroVoice, LT from Bose or ARCS from L-Acoustics. There are others. But look at sensitivity, pressure, dispersion,
A fragment of the scheme for calculating sound pressure, different schemes for voice acting of the upper and lower tiers
It became clear that for work we will use Bose LT speakers and Bose PowerMatch 8500N amplifiers. It turned out that the cost of this equipment is half the estimate, the rest is installation work, suspensions, cable, trays, software and other peripherals. So, the basis of the project is a distributed horn system created by the manufacturer specifically for stadiums and similar facilities - two rows for the upper and lower tiers of the stands, 72 subwoofers assembled in assemblies of 3 for EndFire.
It is high resistance (8-16 Ohms), it can be successfully loaded on a long line. Bose LT is only 560 watts at its peak, the pressure is 134 dB (on average, it depends on the model), that is, the cable cross section is not the biggest (which saves the budget, because the cable requirements for this object are almost like in space). The longest line from the amplifier we got 90 meters - of course, the entire cross section was calculated by the level of attenuation taking into account the weather inductance, dumping factors and other things.
Once again, we checked the entire circuit and started installation.
In late February, cable systems began to arrive. The requirements for the installation of the cable and the cable itself were some of the most stringent: the warning system and evacuation control all the same. Security is at the highest level (for example, we used the most non-combustible cable of all the diameter in the Russian market - German KLOTZ and Sommer). By the way, the whole office laughed at us - we actually got cable samples and burned it. Specifically, with this (Klotz LSE440) it turned out like this: we are burning, but it is not burning. We burn further, but it does not burn, foams, blackens, but does not burn, the wires remained cold after 10 minutes under the direct flame of the burner. So it’s suitable (how did the simple working people break off who collected pieces and pieces of cable in order to burn and hand over the color color reception).
First of all, we mounted suspensions for all systems. Systems were assembled in clusters, it was necessary to clearly position the module relative to another to accurately cover the stands. General contractor people walked around the stadium and photographed every jamb with German pedantry. Burr on the trays - photo, on the carpet, redo.
Installation took place at an altitude of 30–35 meters above the unfinished (and, of course, unseeded) field.
Then the speakers began to come - 184 pieces (these are two and a half trucks of heavy equipment). All this was put in a warehouse, which was 300 meters from the stadium. In the morning the car with the tail lift drove the speakers to the site, if the installation team did not have time to place them, then dragged them back. We did not know how to carry out the installation of such large objects at the start of the project, nor did we understand how much time is required for the unit and what are the optimal procedures. By about the 10th column, it became clear how to optimize the process, but at the 40th level everything was already done on the machine and was documented. A cluster weighing up to 350 kilograms needed to be lifted by an electric winch, but we were not given electricity. More precisely, they gave it, but only later, as a gift for the Day of Russian Sovereignty. Therefore, we bought a three-phase diesel generator for lifting, so as not to depend on anyone.
The third part was the installation of the cable - from the racks of amplifying equipment to the speakers. The racks were assembled in the office and tested (6 amplifier racks, 48 amplifiers). These racks, by the way, are weatherproof - there are sensors for temperature, humidity, fire, information about opening the racks. We assembled the first counter in the office and plugged, without thinking, into an ordinary outlet. They didn’t even turn on the main functionality, but still the group of outlets took off entirely, hello to the service).
Things weren’t limited to tests; software was needed. The controller over IP controls all the hardware, polls all devices, and they had to be programmed. They remove statuses from all equipment, for example, the amplifier can determine the line break or short circuit, but the amplifier needs to be asked about it, and the entire file received from the LOG amplifier should be translated into a form that is understandable to the user - a red and green light on the monitor, etc. As a result, the stadium controller now works with 95% load, collects all system statuses into a large file and gives it out through our API. We see open lines, ragged, short-circuited, improperly loaded and so on. Why was this needed? Because - ta dam! - In the statement of work it was prescribed that our sound system also connects to the emergency warning system and notifies the stadium bowl.
Testing the monitoring system
Typically, SOUEs, such as Bosh and Inter-M, use 100,500 KHz tone generators that analyze line integrity and are at the ends of the lines. We don’t hear them, but the data is analyzed by the return signal and counted over long distances. But these systems do not sound like ice, and they are not designed for such large capacities that are not expected in JI systems (the manufacturer had to do a separate work on certification of equipment).
In April, about seven and a half kilometers of cable was mounted with a cross-section up to 4x6 (multi-core, to eliminate the skin effect on HF, oxygen-free copper). All connectors, of course, soldered, because the power is large, everything is in the air. For a 6-square cable, a 300-watt soldering iron was used.
The project was approved in almost 5 instances (up to the loads on the roof, walkways, point and integrated tracks). Despite this misunderstanding between the subcontractors, it was still (for example, they wanted to put light directly on our mounting point, but then the lighting was mounted a few meters further).
Before starting the system, it was necessary to remove polyethylene from the cabinets, which protected from building dust. This is a separate joke - we hired a local climber - our team dropped anchor and was at another facility. Watching it hang on clusters is a controversial sight. But when it got dark, by 12 o’clock, his appearance didn’t really bother us.
The first test we conducted on a subwoofer turned on in 4 Ohm - just to clarify how it works, and it was very convincing. To hear something on the sub, I had to twist their amplifiers almost to full, up to 18 dB. The fall on the 60 m line was just huge (hi to amplifiers with a dumping factor of 100.5 million). Now there is 8 ohms per line, the losses have decreased significantly - amplifiers are +6 dB, the bass is clearly audible. Colleagues, do not save on cable!
Work had to be done both day and night, which the locals were not very happy about. In a residential complex in the neighborhood, I could constantly hear how hissing, seething, whistling and periodically playing tracks of Simply Red and the Nine group were constantly heard. We checked the polarity of each cabinet, lowered the microphone on the fishing rod to the column, then ran the test signal, looked at the results.
By this time, installers and engineers had ceased to be afraid of heights.
We also did not have a testing methodology, and this became a problem at some stage. The fact is that since the system is integrated with the evacuation control system, it was necessary to ensure operation even if several amplifiers failed (to make a safety margin). We made it so that each cluster was connected to different amplifiers, and when one block fell, the sound in the sector remained, however, it became less legible. Because of this, at first it took a very long time to search for schemes where where. Then we made special verification cards, and things went a lot easier.
And also it was necessary to understand whether measurements fell into the model. We started everything, checked it on an empty stadium (the calculated one was with a 70 percent load by the audience). The RT60 in the empty stadium really turned out to be a little over 5 seconds across a larger frequency range. We drove new data into the model, emptied the model from the audience, updated the materials. It turned out that they hit very accurately, only minimal adjustments were needed. Thanks to the creators of Bose Modeler and EASE, it turned out very accurately, although the mat model of the programs is slightly different.
Then they began to measure speech intelligibility. During preliminary measurements at an empty stadium, they got good readability values, while directly measuring MTF. Rejoiced. A special “man-ear” came to take work. We walked with him for about three hours, showed TK, pressure data, averaged characteristics, numbers (105 dB per place, 106–107 at the peak, 110 right in front of the columns). Peak values, by the way, were measured using rock and roll. He provided 115 at a peak in a number of places under the columns. They limited the gain percentage (despite the fact that there is a limiter in order not to set fire to the speakers). The need for limitation was felt on ourselves - to experience what we experienced while standing below with a microphone, is another adventure.
There were test launches in beta. For example, stadium workers came to watch the match Russia - Belgium - 2014 on large scoreboards. We took the sound from the broadcast and drove it into our system. Commentators and commercials sounded good, but rap was extremely inaudible, literally a mess. Scorpions were heard, on the contrary, perfectly.
After several test events, they began to move inexorably towards the official opening of the stadium. And here they encountered a problem that they had not thought of before. It turned out that even the most autonomous complex needs to be managed. You need to follow him. It is necessary during the events to work at the remote control, to prepare the content. In the next test, the beta team operating the stadium tried to build up without our help (and without a sound engineer). The sound has deteriorated sharply. They called us to find out why this happened. It turns out that the main and backup signals were mixed together. Understanding that the stadium team itself could not cope in the initial stages, we accompanied the first events ourselves:
A rental company was invited to the opening, which used not only its own, but also our stationary sound. Without our knowledge, without the knowledge of Spartak. It was oooooooo loud. To discomfort in the ears. Operating a sound system without appropriate skills is not easy. After several matches, there were unofficial complaints - the sound was illegible, quiet, loud, too loud, echo, etc. It should be noted that speech intelligibility very much depends on the SPL level. The louder the play, the more illegible the sound. This must be monitored. Optimum 85-90 dB. Well, the iconic object - they began to understand. Now with Bose. We scored twice as many devices, microphones, smart programs, 300 m wires and sat on the site for a week.
We started by measuring the impulse response. We drove the sweeps (the sound reminds “buuuuiii”, Spartak joked that it’s the birds being chased at the stadium), they counted the stadium for different loads. In the general case, if the stadium is empty and speech is legible, then the sound will be even better in a filled stadium.
Prior to the adjustment, the
week took the characteristics of each cluster, watched how the system changed, how the setting changed depending on temperature and humidity, found that several columns were simply disabled physically. What else happened during our absence is a matter of further tests. Unfortunately, to see if the column is in order or not is not easy. There are sounds from which even the limiter can not save, something like a delta function (Dirac) ...
We also did a press center. FIFA requirements govern the equipment of the press center. For example, you need to provide 4 places for sync translation. But at the same time, these requirements are, to put it mildly, general. No one says in them that the translator needs to see who he is translating, but cameras are needed for this. The first part of the work was the translation of standards into the TK language, the second was the conviction of the customer that additional systems would be required.
The first time it was deleted from the budget, and then it turned out that you can’t live without additional equipment - it was necessary to collect data from cameras and broadcast it to the stadium’s internal network. There were a lot of pitfalls: auto-guidance on microphones, filters in racks, constant reconfigurations and so on.
As a result, our comprehensive system included subsystems of sound amplification (bowls of the stadium and fan sector), audio routing, sound broadcasting, interactive control, and equipment performance monitoring. In addition, the press conference hall was equipped with a complex of synchro-translation, sound amplification, video display, switching, and technological television systems. In the racks under the roof there are 48 amplifiers with a total power of 160 kW. Mounted 184 speakers, which together create a sound pressure of more than 105 dB. In addition, a local network has been deployed to manage and monitor the operation of the complex.
Spartak Stadium is one of the objects of the 2018 World Cup. Now they pay serious attention to this stadium: stadium builders come from all over the world, look at the engineer, architecture, solutions, usability. They adopt experience - where it is bad, where it is good. Therefore, on October 26th we will again turn the knobs, exposing the best sound quality. Despite the fact that we do not have a service contract, and in the stadium service team we are a sound engineer, we want to show the maximum class in matches.
Everything. If you suddenly have a stadium lying around that needs to be voiced, or a temple , or something else complicated, I can answer questions in the mail MVasilev@croc.ru or right here in the comments.