The Meyer Sound internship and my final-year project

When I finished my third year of university, I got on a plane to California. Pete Soper picked me up in his Outback station wagon, drove me into Berkeley and took me up the Berkeley Hills to a friend's place. That night I met, for the first time, a bunch of people I would end up working with for a while.

Jamie Anderson and his wife Karen Anderson were there, both working at Meyer Sound at the time. So was Dave Dennison, known as DB Dave, a very famous recording engineer who is now the archivist for Jerry Garcia's recording legacy. I jumped straight in. It was amazing meeting them.

I had found a small apartment in the flats of Berkeley, a couple of blocks from Martin Luther King Park and a couple of streets from Telegraph Avenue. Meyer Sound's headquarters was at 2832 San Pablo Avenue, near Orchard Supply Hardware, in an area that used to be the Heinz ketchup warehouse. Super cool area.

I expected to work closely with Jamie and Pete, and later I did a lot more with them. But what I actually did was join the engineering group, and that is where I met Paul Kohut. Paul had been at Meyer Sound, left to create a company called CyberLogic, which built the highest-power, craziest large-scale power amplifiers, including 10,000 watt amplifiers, and then been coaxed back by the chance to build self-powered loudspeaker systems.

I landed in the middle of it. They were just starting to scale the technology up across the whole range of loudspeakers. Two systems already existed, the MTS-4 and the MSL-4. MSL stood for Meyer Sound Laboratories, and the MSL-4 was becoming the flagship of the new technology. People were blown away by it. It had horn-loaded high frequency drivers, low frequency drivers and massive subwoofers on the floor. The MTS-4 was a larger floor-mount style of speaker used in jazz clubs. All of it was being debuted and tested at the Montreux Jazz Festival. It was the highest end of system and sound performance you could imagine. I started working with Paul and became his understudy, learning loudspeakers and manufacturing.

Paul seemed to know everything. He was a one-man manufacturing pro audio guy. Analogue electronics, active filters, power amplifiers, power supply designs, amplifier output stages: whatever you named, he had designed it. I learned an unbelievable amount from him.

Meyer Sound was a perfect sandbox for learning how a product actually gets made. Within a two-block radius of 2832 San Pablo Avenue we had a full woodworking shop, with a very talented man who could build anything from a loudspeaker enclosure, and the engineering facilities, and the manufacturing facilities. Manufacturing had its own wave soldering station, so we built boards and prototypes. I learned about changes to bills of materials, engineering changes and inspecting what came off the line. I wrote instructions for the manufacturing engineers to verify that tolerances were met and that the electrical response checked out, in both large and small signal analysis. The linear systems did the filtering: high pass, low pass, crossover, all pass and band pass filters, all of them active.

Then there was the anechoic chamber. A couple of years before I arrived, they had built it in the parking lot of the main building, designed for loudspeaker measurement. It had a positioning arm that came from a telescope positioning system. It could hold 500 pounds and rotate in azimuth and elevation. That opened up a whole world of detailed measurement of the polar radiation of loudspeakers. You could rotate a loudspeaker across its horizontal axis, its vertical axis and anywhere in between, to 0.1 degrees of accuracy.

There is a pinnacle paper on polar resolution by Felicity Seidel and John Meyer, published by the Audio Engineering Society. Henrik Stafeldt was involved too. They were working on how much polar resolution was enough, and there were standardisation meetings going on at the time. I got into the chamber work and became involved in building some of the systems and taking measurements. Back at a desk I did desktop prediction in MATLAB, where we had all the filters modelled. We built scripts that took the measurements from the chamber, simulated what would happen in electrical systems, and then we built those systems and measured them to confirm they worked. All of that came together in the few months I was there. It was a phenomenal time.

Meyer Sound also made some of the best studio monitors, used by top studio engineers around the world. They were called the HD-1. I found a few scuffed enclosures that could not be used for production and some old amplifiers nobody was using, collected the boards and amplifiers, and built my own pair of HD-1s. I also got some loudspeakers that were just outside manufacturing tolerance but fully functional. I optimised the filter circuitry so they would perform the same as an HD-1. What I needed was just outside what could be adjusted on the board, so I changed some of the filter centres to my own alignment. That gave me my own pair of HD-1s. I took all of that new knowledge back with me to my final year.

For my final year I had an engineering project, with a famous signal processing professor, Dr Claude Lindquist. He encouraged me to push the boundaries of what was necessary, and I was already well beyond the technical bounds of these projects. I was going to do it basically as a master's thesis, and he knew that and backed it. I did not end up with a master's degree. It was my final year project, but he pushed me to pursue it.

My plan was to take my HD-1s and apply digital signal processing to them. So I went back to Meyer Sound and asked if I could borrow a SIM system. SIM stands for source independent measurement, and Meyer Sound had revolutionised loudspeaker measurement with it, using an array of seven or eight HP FFT analysers, each looking at a specific band of the audio frequency range. SIM System II was the first implementation fully built into a standalone product: a 386 with a master processor and multiple DSP boards attached. The boards used AT&T floating point DSPs, the AT&T 32, a revolution at the time. There were three boards, each processing one channel of source measurement analysis. It was a very expensive system, worth about 80,000 US dollars. There were a few demo systems around, and I asked if I could borrow one for a few months to finish my project. Luckily I got the demo loan, and they shipped me the system.

Next I went to the people who ran the music engineering programme, to Ken Pohlmann, the founder of the whole programme, and explained what I wanted to do. They gave me full access to the tape archive, just below the recording studio and Gusman Concert Hall. They gave me the space so I could set up a bench and the entire experiment, because moving the system around would have been impossible. That was amazing too.

I took the loudspeaker system and pulled out all the circuitry between the crossovers and filters. In its place I put two Motorola 56K DSPs. The faster chip ran a huge FIR filter, a finite impulse response filter. The other, with slightly lower performance, handled the higher frequencies. The aim was a razor-flat system frequency response from the speaker. As far as I am aware it was the first time anyone had shown literature on a decimation-based multi-DSP solution for this. I wrote my whole master's-level final thesis on it.

That system caught people's attention. Ken had people he worked with from Japan, and some people from Sony came and saw it. They were amazed by what I had achieved and asked whether it was something we might be able to license or work on. I went back to John Meyer and asked if he would be interested in talking to somebody about licensing the technology. That was maybe overstepping, as I knew little about it at the time. John said no. I think that was actually good, because when I graduated I was hired by Meyer immediately. They made me an offer, and I left Miami and moved straight to Berkeley.

That is when I started at Meyer Sound and developed the speaker systems for the Ultra Series, including the self-powered UPA-1P. It ended up being the subject of my first patent, which I shared with John Meyer and Paul Kohut for the systems we built.