While pipe organs may date back well over 2,000 years, they now use the latest networking technologies, including 802.11 wireless, to connect their complex mechanisms. A pipe organ in this discussion does not mean one that uses digital or other electronic technologies to create its sound, but instruments that use real pipes blown by compressed air. Organs have always been the most complicated musical instruments, so today’s organ builders utilize advanced technologies to make the job much simpler.

Historically pipe organs had very elaborate mechanisms, and in pre-industrial Europe, they represented some of the most complex mechanical devices built up to that time. Each key on the keyboards and pedalboard needed its own series of cranks and levers to open a valve inside a chest to make the individual pipes speak. These mechanisms had practical limitations that weren’t overcome until the 19th century when pneumatic and electric devices reduced the need for direct mechanical links.

By the 20th century, electric and electro-pneumatic systems largely supplanted earlier designs, although mechanical instruments (often called “tracker organs”) are still built today and regarded by some purists as superior, despite their mechanical complexity and high cost.

Just in case you’re wondering if anybody still builds pipe organs, the Wall Street Journal estimates the industry at $80 million in the U.S. alone. While not huge, organ building is very much alive and well in the 21st century.

An organist sitting at the console selects stops, which are ranks of pipes designed to make a specific musical sound. These are used individually or in combinations. Normally a rank has 61 pipes to match the number of keys on the keyboard. If the organist pulls five stops, when he or she presses one note, five pipes corresponding to that note all play at the same time. A pipe speaks because a valve inside the chest opens and allows the compressed air, or wind, to pass into the pipe. So an organist playing actuates a hugely complex series of valves. An organ suited to a concert hall or large church can easily have 2,000 to 5,000 pipes, so the demands placed on a switching system can be considerable.

Early electrical control systems were hard wired, comprising racks of relays and thousands of individual connections. An instrument of any size had huge bundles of small wires and usually an entire room dedicated to delicate switchgear. With the coming of diodes and transistors, relays were supplanted by solid state mechanisms, and printed circuit boards replaced many hand wired connections, but there were still too many wires. In the 1980’s, more sophisticated multiplexing systems reduced the cabling necessary even further.

Recently Ethernet-based networks have brought a new level of user functionality while simplifying internal wiring. Unfortunately, these networks brought with them the same problems that affect industrial motion control platforms: latency and jitter.

A pipe organ puts particular demands on networking. When the organist presses a key, the system has to recognize that the contact is closed for a specific note; it has to check which stops are drawn; and then it routes signals to the appropriate pipes to make them speak for as long as a finger or foot remains on that note. However, organists don’t normally play with only one finger. They use all 10 fingers, plus both feet to keep key switches opening and closing with amazing speed and agility. Moreover, combinations of stops can change constantly as well.

Organists are used to the idea of being separated from the sound they make, since pipes are usually some distance from the console. Therefore they understand that the music will always lag behind their fingers. But can an organist tell the difference between the speed of sound and a sloppy, non-deterministic switching network?

According to Duncan Crundwell, president of 1602 Group which owns Solid State Organ Systems, the answer is “yes.” “We learned from an initial installation that latency is an issue,” he recalls. “With an organist sitting 50 feet from the pipe chambers, hearing the pipes speak is delayed at least 50 ms because of the speed of sound in air, so we assumed that adding another 30 ms would simply move the instrument further back, but always a constant amount. This was not so and early tests revealed that latency was important, as was constant latency. We discovered that the onset of perception of latency is around 20 ms if it is constant, but variable latency is less tolerable.”

Scott Peterson, president of Peterson Electro-Musical Products, Inc., (EMP) reports similar observations. “Some top organists seem to have a remarkable ability to detect a difference between the way they press the keys with their fingers and what they hear in response, especially variations in latency, which they perceive as rhythmic irregularity in the music,” he says. “The parameters that are considered acceptable to most organists are quite a bit less stringent than those considered acceptable by a few virtuosos. Our experience indicates that total latency greater than about 30 ms is a problem for many organists. A variation range of around 15 ms (for example, nominal-minus 10 ms to nominal-plus 5 ms) is considered acceptable by most everyone, whereas a variation range approaching 25 ms or 30 ms may be objectionable. Still, we have found that jitter is a bigger concern than latency, although both are important considerations.”

Ethernet has helped provide sophisticated capabilities for organists while creating flexibility for the organ system designer. “With Ethernet in our ICS-4000 control system, it has been possible to keep system performance within an acceptable range, but I wouldn’t say it has been easy to do that on larger jobs,” says Peterson. “The ICS architecture uses a distributed processing system where the main microprocessor in the console communicates via Ethernet with satellite microcontrollers in remote locations. Much care has been taken to optimize the distribution of all required tasks between local and main processors with emphasis on performance. Smaller systems are easy, but very large pipe organs have more inputs and outputs to handle in each frame or scan cycle. As we’ve sold increasingly large and complex systems, we’ve [had to improve] the tightness of the code to meet the challenge. I think the actual Ethernet communication structure is plenty fast, but the data management we have to provide, before and after data is staged for transmission between processors, is what we have needed to keep very crisp.”

At the same time, Crundwell says that Ethernet brings its own set of issues. “TCP/IP in itself does not address most of our requirements for latency and frame integrity,” he notes. “Packets in a pipe organ control system are very small and frequent. A dropped frame in a voice-over-IP (VOIP) system is not a big problem, in a pipe organ it can be catastrophic unless other measures are used to retain integrity and so a user datagram protocol (UDP)-type system does not resolve many of the issues.”

The mechanical aspects of switch-contact closing are also an issue. Organists get used to the feel of a keyboard, and expect the contacts to close at the same point every time. Keys usually depress about .315 in. However, the switch contacts close sooner than that, often with as little as .100 in, which helps the organist play fast passages. A keyboard has to be adjusted very carefully to make sure the closure or “speech point” is very consistent from note to note. “A poorly designed Ethernet system will introduce variable latency that the organist will perceive as the contact position moving,” says Crundwell. “They find this very disturbing. Jitter and lack of accurate scanning and reproduction can skim off the finest nuances that an organist has strived for many years to achieve.”

“Automation systems have tracked memory technology from magnetic drum through SRAM to flash and now removable drives,” says Crundwell. “Networking has allowed an organ console with a control surface with close to 1,000 inputs to operate a distributed I/O system controlling up to 15,000 actuators in real time.”