Wendel Questions (answered by Roger himself circa 2007)

Background

During the making of Steely Dan records before “Gaucho”, we sometimes used little rhythm boxes as a guide instead of just a click. For example, on the song “FM”, a Rhythm Ace was used as the guide. The entire record was completed without drums. The very last overdub was Jeff Porcaro replacing the rhythm machine. Due to track shortage (24-track analog), we had to erase the rhythm machine to make room for Jeff’s drum tracks. Since everything on the record was recorded to the rhythm machine, we hoped it would work out.

Donald and Walter wanted to discover a new method of producing steady drum tracks that sounded like real drums but with the feel of a real drummer. In 1976, I bought a home computer and started exploring digital audio. I had my 1.8 MHz 8bit computer modified to add four extra S-100 card slots, expanded my memory to 56 Kilobytes, and added a custom-built 12bit, 100k sample-rate A-D and D/A digital converter board. I then wrote the assembly language program to handle the sampling, editing, and triggering of the audio.

I made a critical decision about fidelity early on. The choice was between being able to play all the drums simultaneously at lower fidelity or use all the computer’s bandwidth to play one drum at a time at full fidelity. I chose the latter.

We hired drummers like Jeff Porcaro, Rick Marotta, Steve Gadd, Bernard Purdie, and others to sample their drums. After using the Wendel to build a track, Donald and Walter played over the drums to ensure everything felt right, then other musicians performed their over-dubs.

When the session contracts went to the Musician’s Union, no drummer was listed on the sessions. The Union called us after hearing rumors that we were replacing musicians with machines. They warned that if it was true, they would fine all musicians who played on those songs. We assured the Musician’s Union it was a clerical error, and we would add the drummers. If we used Jeff Porcaro’s drum samples, then Jeff was listed on the session contract and album credits as the drummer. The Union then left us alone. This is when we coined the name Wendel for the machine, so everyone would think it was a person. The credits read “Rhythm Augmentation—Wendel”.

Building the drum tracks on “Hey Nineteen”

By December ‘78, I had put together a new computer with a 3Mhz 8bit processor and assembled everything into a portable case. “Hey Nineteen” was the first song ready to record for the album. Donald recorded a piano-vocal demo of the song in the studio, and then I went home to start programming the drum part into the computer.

The computer was only a 3MHz machine, 1000 times slower than today’s 3GHz computers. There was no such thing as an IBM PC or a Mac. Wendel could only play one sampled drum sound at a time at the hi-fi sample rate that we used.

There was also no such thing as MIDI. I put an eighth-note click on one track of the analog tape with a UREI Metronome. The audio from the click track was fed to Wendel, where it was converted to digital for triggering.

The drum pattern for the song was based on the eighth-note click. The computer would count the clicks and determine where each of the sampled drum should be played. If a drum needed to play on a sixteenth-note boundary, Wendel knew the distance between clicks and would trigger at the appropriate place.

Because Wendel could only do one sample at a time, I would print the kick drum, change tracks, print the snare drum, then change tracks again and print the hi hat. We then listened to the track and decided if everything felt right. If the kick drum sounded early, we could have Wendel play the kick live while listening to the snare and hi hat that were already printed, and adjust the kick timing until we liked it. We would then do the same thing with the snare and hi hat and go around in this circular adjustment mode until the feel was right.

Is the final mix 100% your Wendel sequencing?

The drum fills on Gaucho were overdubs by real drummers. Remember that this was the first attempt at this sort of drum replacement, and the computers were not fast enough to do more than one thing at a time at this sample rate.

How long did it take you to program it?

It took about three months to create the actual program that was Wendel, but the experimenting and building the boards had been ongoing for almost a year before that.

In addition to perfect triggering of the samples, the drums had to feel steady, yet human. When I sampled each drummer’s drums, I also figured out how they played. They way they moved around the beat, and they way they transitioned from verse to chorus and back. These parameters were used to make the finished track sound fluid and steady at the same time.

The actual final printing of the drums on Hey Nineteen was January 19, 1979, but I was taking home tapes with UREI clicks, and staying in the studio late after Donald and Walter left to test things. I remember that Donald kept asking “When is your machine going to be ready?” Finally I said “We can do ‘Hey Nineteen’ this Friday.” Donald, thinking he was going to scare me said, “OK, then we will schedule musicians to come in and play to it on Monday.”

Fortunately, everything worked perfectly.

Did any of Rick Marotta’s live playing make it onto the song?

Rick came back in to do the fills, but the live track we tried wasn’t kept. The Wendel track was started from scratch.

Reflections on this landmark recording

Hey Nineteen was the first stepping stone for future Wendel developments.

In 1981, we started working on Donald Fagen’s “Nightfly” album. This one was digitally recorded on the 3M 32-track. We used Wendel to build some of the drum tracks as guides, but it was already decided that I was going to build a new 16bit machine with digital interfaces to the 3M machine. The drum samples were recorded on the 32-track, then transferred digitally into Wendel II. All triggering was done digitally, and the samples were transferred over the interface back to the 32-track. Triggering was fast. It took less than one sample for the triggered drum to go back no tape. The sample was 50Khz 16bit.

At this point, the computer was a whopping 6MHz 8086 processor, but because of the new 16bit samples and only one digital interface, the process was still one drum at a time. However, now we could move things around by samples, not milliseconds.

Wendel II was used for more than drums. It played Piano samples, congas, did drum fills, sequenced synthesizers, and 10-second long ride cymbals.

In 1985, I built Wendeljr. I built 700 of them and sold all but the three I still have in my rack. Clair Bros. still uses them on every concert tour to replace or add to the kick and snare on live shows. I went to the Simon & Garfunkle tour last year with Jim Keltner drumming. They were using Wendeljr for the kick and snare augmentation.

Wendeljr credits were growing so fast that I quit keeping track after the first six months. NARAS thought Wendel was a person and was nominated for a Grammy. Wendel was removed from the nomination list when they found out it was a machine.

MIDI was the new rage. Everybody wanted Wendeljr to be MIDI triggered. I said “Wendeljr is 1000 times faster than MIDI. Wendeljr will never have MIDI.” The pressure for MIDI grew greater, so I quit building them. Once in a while, I see one on eBay for triple what they sold for new in 1985.

Wendel Chronology: (from rogernichols.com)

1976: First computer purchase, a COMPAL-80, and 8080 assembly
language classes.

1978: Developed Wendel as a sampling drum machine and audio
sampler. 125kHz/12bit

January 1979: Wendel used on Steely Dan “Gaucho” album
for drums and percussion.

1981: Wendel-II 16bit 8086 version with digital i/o to 3M digital
32 track recorder. Used for drums and percussion and audio sampling on Donald
Fagen’s “Nightfly” album. 50kHz/16bit.

1984: Wendeljr, playback only percussion playback unit. 50kHz/16bit
Below is a photo of part of Wendel’s converter, a photo of Wendel,
and a couple of screen shots of the Wendel editor that was used to record
and edit the cartridges for Wendeljr in 1984. Wendel was powered by a 3mHz
8085 processor. The programs were written in 8085 assembler. The I/O was analog,
the converters were TRW microwave converters with a Teledyne sample and hold
amplifier. The sounds were sampled at 125kHz/12bit.

George Peterson Wendeljr Review 1987

First computer. COMPAL80. 56k memory, 1.8mHz 8080. 315k Micropolis
floppy.

2.5″x5″ sample and hold for the 12 bit converter.

The original Wendel.S-100 card cage to the left of the 5 1/4″ floppy
drive, 9″ monitor to the right. 1978, the first laptop computer.

Wendel-3 editing screen 1984. Editing conga for Wendeljr Carts.

Wendel-3 screen with edit markers 1984