Making the immeasurable measurable

Simulations also play a key role in the adjacent room. Before a prototype part is even produced, Guy Layfield knows more about its properties than can realistically be measured. On several screens, programs and sub-­calculations are running, which we are asked not to mention. How does the driver behave under condition X, and how does the driver concept work? “For example, here we can see what happens inside the drivers at any point, even though this isn’t measurable. We nevertheless know it, and we know it very, very accurately — and can carry out development work much more precisely even before the prototypes.” All scripts and many software enhancements are in-house developments by the team. The simpler brute force simulations usually run overnight on very powerful workstations. More complex calculations take days, sometimes even a week. Which components do they simulate most frequently? The many moving parts and materials of the drivers — and the magnet system, of course.

Leif Böhme fetches a prototype of the KH 80 DSP housing from the cabinet one desk away. “The housing is another component that contributes significantly to overall performance,” says the developer. How exactly do the side walls have to be designed, strutted, and manufactured so that they do not resonate? Böhme points to the cover of the prototype: “In this version, it still had too much movement at 500 Hertz. We then tested and eliminated this movement in a FEM (Finite Element Method) simulation.” BEM (Boundary Element Method) calculations run on his second screen to simulate sound radiation. But development goes far beyond resistance to resonance. The waveguide, which contributes to the extraordinary directivity, was also created here. This significantly determines the neutrality outside the beam axis, thus providing for the interaction of the loudspeaker with the space. This is a very important factor that makes ­Neumann monitors sound so neutral even under the widest range of installation and room conditions. The elliptical shape ensures a wide listening range.

Good sound comes from good code: Software developer working on the signal processor.

The laser finds every nuance — even inaudible ones

Only when all simulations meet the requirements does the actual prototype construction begin. Guy Layfield uses a ­laser to measure the first drivers at 1,400 points across all frequency ranges. One measurement run takes eight hours. Do all parts and materials behave as expected? Does the beading or the diaphragm move as it should? Is acoustic energy building up? “We detect peaks in the measurement that you can’t see and sometimes can’t even hear. It is nevertheless better to exclude the possibility of them happening. You never know what else will happen with it in the overall concept.” Dozens of drivers from a myriad of iteration steps lie on the shelf on the rear wall, because the design is only confirmed when the measurement findings match the simulation results.

A natural allergy to everything that rattles

The long series of measurements in the large anechoic chamber are mentioned by the employees rather incidentally; it seems to be something taken for granted. It also seems to go without saying that the size of the room and the shape of the wedges in it are precisely matched to the lower limiting frequency that is to be absorbed here. Further back in the workshop, a wooden box jiggles and hums. A rattling test is underway. With stethoscopes and hearing tubes, the engineers spend hours on circuit boards and interior structures. From the very first design, they make sure that nothing would rattle: no components, no resistors, no capacitors. No contact surface may vibrate either — absolutely nothing.

Is something ratteling? It is standard procedure to conduct extensive tests on the vibrating board.

A stress test — only much worse than usual

The large workshop contains a high-level chamber, a monstrous box whose walls are filled with quartz sand. The heavy reinforced door can only be opened with a great deal of force. Inside it smells like hot electronics — very hot electronics. It’s the place for the ultimate stress test: The engineers test the long-term stability of each prototype under continuous load in the chamber. Such tests usually run for 100 hours, but at Neumann they take 1,000. After that, all parameters of the monitor must be identical to what they were at the start — otherwise the development engineers return to the drawing board. For this testing, music signals that represent a realistic load in terms of dynamics and requirements are used. Standard noise according to IEC 268 is of course also part of the test series. However, the standard is already 40 years old. “We have an additional version that is adapted to meet the requirements of modern amplifiers: The peaks are changed, a few dynamics are added, and the CREST factor is modified,” says Markus Wolff. In short, the team doesn’t find the industry standards to be sufficient. They have higher expectations when it comes to stress testing and reliability.

In contrast, it is suspiciously quiet in the programmers’ room. Programmers? Yes, because a digital signal processor (DSP) operates in the KH 80 DSP near-field monitor. Apart from phase fidelity, the DSP provides the team with enormous precision through filters and adjustments to the room. For example, dozens of parameters can be easily optimized with an app — something that would be unthinkable with analog variants that are adapted to their place of use via switches on the rear. How do the screens on the desktop or mixing console affect the lower midrange? How much acoustic energy is appropriate in the work area? Professionals can adjust the entire frequency range via app, and beginners are guided to better alignments through a series of questions. What runtimes result from the signal processing and calculations? What protocols does the system use? This is a crucial point: “If things are to go well, software developers need to understand all the interrelationships right from the start — and we have to understand the programmers.” For example, how do you mask possible interference artifacts from analog switch events in the digital domain? Programming is done in C++. The protocols are the result of in-house development in part. The operating system for the central processor alone needs 512 kilobytes; as much as is available for the programming. Any large ­image on a modern website needs more storage space than the programmers have left for the code that is the core of the digital monitor. They smile in a way that is quite reminiscent of their colleagues in the physical development. As if they are having fun. As if not everyone would be able to. And as if they might also know that.

The visit ends in the developers’ listening and testing room. For Markus Wolff and his colleagues, the tension in this room is palpable. “When a pre-production loudspeaker is set up here for the first time, it’s done. No one has ever heard it before, and it has taken up to three years to get to this point. That’s something special.” He hospitably extends an invitation to the guests to come to the central listening location, offers jazz, rock, blues, and classical music, and connects to Neumann’s series of monitors. Music that is regarded as standard is deliberately included, music that you feel like you’ve known through and through for a long time — until the friendly Mr. Wolff presses “Play.”

A 1,000 hour endurance test is mandatory for all new developements.