Strymon co-founder Pete Celi is responsible for the sound design and DSP algorithm creation for all the Strymon pedals that use DSP. Recently I had the pleasure of sitting down for an interview with Pete. We discussed his approach to sound design, his background in music and technology and the path that led him to what he does at Strymon today, as well as his love of electric guitar.
Pete, can you please talk a little about yourself, where you’re from, what you’re into?
I was born and raised in Massachusetts and started playing guitar as a kid. I went to college there and studied electrical engineering, and moved down here [to Los Angeles] in ’89. There’s your short answer!
What made you decide to move to LA?
It was really the call of music that brought me out here. I’d graduated college in 1985 with an electrical engineering degree, and started working for Analog Devices, the chip maker [in Massachusetts]. I did that for three years. All the while I was playing guitar, jamming with friends, little projects here and there. But Analog Devices was a very kind of serious button-down environment that just really didn’t feel like it was for me for long-term, and I decided to move out to Los Angeles to go to Musician’s Institute in Hollywood. So Analog Devices was actually a pretty prestigious place to work in my field, with good pay, good benefits, and I quit that job to go to Los Angeles with no real plan other than to attend Musician’s Institute.
What was your experience at Musician’s Institute like?
It was actually a really great experience because I got to really see… see some reality. They split us up according to ability, and I ended up traveling around with sort of the top class of about 20 guys. But I could see that the top few guys in that class were not only light years ahead of me in terms of playing ability, but also had this spark of creativity and genius in their playing. I realized that even those guys were gonna be struggling to try to do something in music, and I knew I’d never attain their level of naturalness and musical ability. So that was valuable to me because it became clear that, ok, I’m not gonna be a professional musician. But it was a fun year—it was kind of like a sabbatical. I got to learn some things and have some fun, but then I needed to figure out what I was gonna do next because school was over and I didn’t have a job.
Pete in 1989
So how did you end up staying and surviving in Los Angeles?
I knew there were some companies in L.A. that made music products, and I had an electrical engineering degree. So I wrote some letters and signed them with my name and telephone number and mailed them to a few different places, and one of them that called me was Alesis. I ended up working for them starting in 1989, at an exciting time when everything they were doing was breaking performance and cost barriers. The original Quadraverb had just come out. A few years later the ADAT came out, and the company grew from about 30 employees when I started working there, to about 200 employees. It was a rocket ride, and it’s what got me into working with music technology, and that’s what I’ve been doing for about 26 years.
We were designing our own chips at Alesis, which was pretty cool and kind of rare. But the chips I would be working on would maybe be used in a product two or three years down the road that hadn’t even been defined yet, so I never really got to have my hands on actual products. So it was fun and a great environment, but I never really felt 100% connected to what I was doing. I felt like I was working on technology, but not really working on music products. Writing DSP (digital signal processor) code and working on sound design for music products like I do now was really a pretty dramatic change for me.
After Alesis, you worked for Line 6, right? What did you do there?
My title was Senior DSP Engineer. When I started there, Variax was really a blank paper project. They told me they wanted me to start researching how they could take what they did with [emulating] amps and do it with guitars, and two or three years later: “Hey, this actually works!” I wrote all the DSP sound generating code for the original Variax. Getting involved with that, writing DSP code for a product and being instrumental in shaping the sounds of that product really felt a lot more true to what I really wanted to be doing.
And now you are one of the co-founders of Strymon. Why do you feel like this is the ideal fit for you?
At Strymon I get to wear two hats and create the DSP code and also do the sound design, so instead of having those two things separated into two departments (sound design and engineering), they really are one and the same thing, so there’s nothing lost in translation. To know exactly what sound you’re going for as your write the code, and to know what you’re hearing when you listen to the sound and how to address it in the code is ideal. At Strymon we’ve got a small group. The guys that I’m working with here are guys that I’ve known, that I have great respect for, a creative group of guys. There’s a lot of collaboration.
So I assume you’ve done the sound design for every Strymon pedal from the beginning?
Yes, well with the exception of OB.1, where Gregg did the analog design, because OB.1 is a fully analog pedal. I provided input here and there on the sound, but Gregg did the analog design for that product. But other than OB.1, for all the other products that have come out since we started the Strymon line, 100% of all the sound processing and sound design has been done by me.
Can you describe your approach to sound design?
The way I look at what we’re doing is, we’re designing effects where fundamentally the goals and the process in a broad sense aren’t different than those of any other effect maker designing in any other medium. If you’re an analog designer, you start with an idea, you have a circuit, you build it, you have components, you have some knowledge base and history to draw from, and you make adjustments and listen for things and decide when it is to your liking or what direction it needs to go in. We’re doing things in a similar way, except that the medium is DSP, which does allow us greater flexibility and possibilities, ultimately.
How do you research and what is your process like?
We have tools to draw from and experience to guide us in certain ways, but it’s definitely not a robotic process like, “well this is how we do it, we just put these signals in here…” It’s different for each sound. For something with a physical or electrical counterpart, like a spring reverb or a power tube tremolo, there are more measurements and equations that are part of the process, but we’re not trying to capture an exact copy of a specific existing piece of equipment.
So once we have described the process mathematically, the next step is to craft the sound. For a tube tremolo, different oscillator topologies might distort the LFO waveform in a good way. Or the amplitude of the LFO can be varied while different tube biases are experimented with, just as an amp designer would do. So a fair amount of the process is artistic as well. Crafting the sound.
I think the cool part about DSP is that it’s so open, that there are so many things you can do, and it’s easy to experiment and try things that wouldn’t have been easy to do before. For many effects, like on the BigSky reverb, some of those reverbs have long abandoned the reality of any physical space. It’s about creating resonances and pitches and feedback and ambient soundscapes that don’t have any physical counterpart. That can be the most fun because it’s really just an exercise in creativity and freedom.
Can you talk about any particular challenges or successes you’ve experienced with any of the Strymon pedals?
Well, early on in the technology of the Strymon line we developed what we call the dBucket technology, which is digital bucket brigade delay technology. In an analog bucket brigade delay, they use chips that are called bucket brigade chips because basically there’s a transfer of charge through a transistor and capacitor, many many times, like old-time firemen with each man passing water from his bucket into the next guy’s bucket, so this analog signal is transferred through each transistor and capacitor stage, and that takes time, and the signal is delayed.
Each bucket stage adds time, and the rate at which signal is passed from bucket to bucket is controlled by a clock. If you speed up the clock, everything goes faster and there’s a shorter delay, or slow down the clock and you get a longer delay. It’s fundamentally different than a digital delay where traditionally you’re running at a fixed clock rate, and you sample things in and store them in memory, and if you want a longer delay time, you just use more memory, or in the bucket analogy, you just add more buckets, but it’s always the same rate, and there’s no loss from bucket to bucket in a digital delay.
A big part of the character of an analog delay is the fact that it’s running on this variable clock system where the number of buckets stay the same, but it’s just how fast you’re running through the buckets. So to get long delays, you actually have to slow down the clock enough that you start to hear artifacts, and there are two different kinds of artifacts. One is actually the clock itself, if it is slowed down into the audio frequency range you’ll actually hear the whine of the clock, but more than that the other type of artifact is if the clock speeds get slow enough, it starts to alias.
Can you explain aliasing?
Aliasing is where in a clocked system, which actually applies to an analog delay because it is a clocked system, the maximum frequency that can be represented accurately is one half of the clock frequency. If you go higher than that, what comes out of the system is actually a frequency that’s wrong. It’s kind of like when you watch an old film and you see the spokes of a wagon wheel and they look like they’re going backward, because the spokes are moving too fast for the frame rate. That’s aliasing in film.
So in an analog delay if you’ve got a slow clock speed, let’s say 10kHz, any input signal above 5kHz is going to alias, and it sounds like “bzzzzzzz,” a buzz coming down from those frequencies. So in order to combat that in analog delays usually they employ a lot of filtering to remove those high frequencies so that content that would be reported erroneously is filtered out. So that’s why analog delays are traditionally dark sounding. It’s not something to do with the chip, it has to do with the filtering they put in there to reduce those artifacts.
On top of that, to continue with the bucket analogy, you spill some of the water each time you transfer from bucket to bucket, and that’s due to the physical properties of the transistors and monolithic capacitors that are doing those trasnfers. How much you spill is dependent on how much water is in the bucket, and how fast you are transferring from bucket to bucket, so it’s a complex process in some ways. That’s something I know about because I have a background in integrated circuits and stuff.
That ‘water spilling’ creates the grungy and noisy aspect of analog delays. So we thought, let’s do a delay in DSP that’s actually running internally at a variable clock, and include a bucket brigade line where the loss between transfers actually occurs as it does in the chips. And we figured that out, and it does work, and what it allows you to do is to control that loss. You can make it a perfect transfer, or you can make it a not so good transfer from bucket to bucket. You can get a different range of experiences. It’s still a variable clock process using a fixed number buckets, but because it’s in DSP we can control the quality of the bucket brigade chip, the amount of filtering, the companding parameters, and the various levels of artifacts. So that was really a big thing for us, once we got going, we were like, this is really cool.
What do you think led you to finding that you had a real passion for sound design?
My whole life I’ve played guitar and I’ve always loved it and I’ve always been obsessively tuned in to the response between your fingers and what comes out of the speaker, and what the effect is doing, and all that kind of stuff. And never in the 38 years since I started playing guitar have I ever taken two months off, like “I’m not gonna play.” I probably play guitar three hours a day just for work. I guess it’s always been in my DNA.
When I was 13 or 14 and just learning to play, I was taking lessons from my neighbor and he had been letting me borrow this crappy acoustic, and one day I went over to his house and he had this Silvertone electric guitar and a Silvertone amp that had a reverb. He handed the guitar to me and said “why don’t you play this and see what you think.” It was the first time I’d ever held an electric guitar. And I played a note and there was some reverb on, and it was one of the few moments of my life that I still remember so clearly, because it was so electric, and I was like, DAMN! It was probably a seed for why I’m doing this now, because… That was COOL! It’s been a constant thing throughout my life since then, regardless of all the other variables.
Posted by Angela
Welcome to the first installment of Every Instrument has a Story. Over the coming months we’ll find interesting and unique stories about instruments, from customers, well-known artists, and even us. We thought it would be fitting to start with the Gibson Explorer owned by one of Strymon’s founders, Gregg Stock.
Ok, so we need to set the stage for this guitar. In the late ‘50s when the Gibson Explorer was first released it was not a hot seller and was eventually discontinued. After many years, Gibson decided to try again and re-issued the Explorer in 1976. In the late 70s, Gregg went to a small Seattle-area music store and saw this sweet, angular-looking guitar on the wall. And so began the journey of Gregg’s 1976 Explorer.
Turns out that the guitar that Gregg saw on the wall was actually formerly owned by Howard Lease of the band Heart. While Gregg was putting the Explorer on layaway, he actually got to meet and talk to Howard. They chatted about playing techniques and stories from the road. Finally after many months and Gregg’s last layaway payment, the Gibson Explorer was now his.
The Modding Era
Now, you may look at a photo of Gregg’s guitar today and wonder how he could have completely modified, reworked, and cut apart a cherished ’76 Explorer? You have to keep in mind— back in this era, modding and hot rodding your guitar was the thing to do!
During the Van Halen era, the inspiration came to have the pickup rewound by Seymour Duncan. But that wasn’t all, then the Gibson made it’s way to Mike Lull who did a refret and shaved the neck to be more like a Flying V. Now the guitar was almost there, but it needed to make one more trip over to Floyd Rose. With the Explorer in tow Gregg went north to Seattle to a tiny house to have Mr. Rose put in one of his first Floyd Rose tremolos. The tremolo didn’t stay forever and is no longer there. While Gregg was at Floyd’s place he spotted a guitar that was going to be shipped to Eddie Van Halen. Luckily Gregg was able to try out Eddie’s guitar before it was shipped off. So not only does Gregg’s Explorer have a story, but now Gregg had his own story that he actually played one of Eddie Van Halen’s guitars.
In present day, Gregg’s Gibson Explorer has been retired and is hanging on the wall of the Strymon shop as a reminder of the fun adventures they went on together.
Posted by Ethan
The reviews for BigSky are now appearing through the clouds of the interwebs, and we’re incredibly excited to hear what people have to say.
Featuring 12 unique reverb machines – from rooms, halls, and plate reverbs to “cloud”, “bloom” and “magneto” and more – “Big Sky is something of a triumph, producing hyper-realistic rooms, ambient space trips and everything in between with ease and finesse,” writes Sound on Sound magazine.
» Read the Sound on Sound review
Our friends at Premier Guitar think that BigSky is a “virtual mad-scientist lab for those who are endlessly fascinated with the possibilities of audio signals bouncing off unseen objects—and then being twisted, warped, chopped, and vaporized.” It’s so great to hear that what they call “otherworldly and authentic vintage tones” within BigSky provoked such a colorful and evocative reaction.
» Read the Premier Guitar review
Gabriel from Best Guitar Effects attended our recent Strymon Social event, and used that opportunity to spend some time talking with our co-founders Gregg, Dave, and Pete. To learn about his “intel-gathering reconnaissance mission” at the Strymon labs, click on over to his incredibly in-depth BigSky review. We’re pleased to learn that he ranks BigSky as “simply one of the best reverb processors available today.” And we’re glad to hear his praise for the wealth of “studio-grade power” in it’s performance-friendly form-factor.
» Read the Best Guitar Effects BigSky review
Guitarist magazine also has some very nice things to say about BigSky. They noted that the reverberations within BigSky “make your jaw drop and your playing soar” and that it makes them “want to write poetry to the soundtrack of its glorious ambience.” We’re so glad to hear that BigSky got their creative juices flowing!
» Read the Guitarist BigSky review
We’re incredibly honored that BigSky has received the SonicState Best Effect Pedal of 2013 award, with their team noting “everything about it is stunning.” When it comes to creative inspiration, they cited the real-time tweakability of BigSky as being “on a different level to any reverb we’ve tried before.” And they totally made us blush when they said our “products sound great, they look great, they’re perfectly engineered, and they essentially fit a whole studio rack’s worth of effects and patches into a tiny little box for using on the road.” Thanks guys, we’re so glad you’re enjoying what we’re up to here in the labs! 🙂
» See SonicState‘s Best Effect Pedals of 2013
Posted by Angela
Recently we asked on Facebook and Twitter for you to “go to your iTunes (or Spotify, last.fm, etc) and tell us your most listened to song.”
It was wonderful to see over one hundred songs mentioned and not one of them was a duplicate. There was such a great variety, and a lot of songs I personally had never heard before. I’m still going through and checking them all out.
I have highlighted our answers and some of yours. If you’d like to see them all check out the post on our Facebook page.
Posted by Ethan
Our very own DSP Engineer and co-founder Pete Celi recently had the opportunity to contribute an article for the March 2013 issue of Premier Guitar magazine.
“Unlike a tree falling in the woods, a pedal won’t make a sound unless an amplifier is there to amplify it. (If no one is there to hear the amplifier, that’s a different philosophical discussion entirely.) This may seem a bit obvious, but when we talk of a pedal’s tone we need to discuss it in the context of a particular rig. This is particularly relevant when discussing overdrive, distortion, and fuzz pedals.”
Head on over to the Premier Guitar site to read the whole article:
Posted by Ethan
Our very own DSP Engineer and co-founder Pete recently wrote an awesome article on flangers for the September issue of Premier Guitar. It illuminates some of the finer and more confusing aspects of how flangers work and how to best utilize them. Flangers can be challenging to understand … hopefully this sheds some light on the subject. Read the full article here.
Posted by Ethan
Sometimes to understand who you are, you have to go back to the beginning, back to where it all began. Before smart phones, before computers, before integrated circuits and the transistor—the only effects available to guitarists were tremolo and spring reverb. The guitar players of the day didn’t have the rainbow of colors that we have now.
But like a charcoal sketch, there is a stark beauty to the tone without the wash of effects that are now possible. Stripped down to the bare necessities, the contrast of the different tremolos becomes apparent. You feel the beating heart of the photo trem, the rolling waves of the tube trem and the hypnotic swirl of the harmonic tremolo.
Given the storied history of these circuits found within classic amplifiers of the 1960s, there was no doubt that we wanted to develop a studio-class pedal that faithfully delivers three of these iconic and unmistakable tremolo effects. We examined the sonic complexities and tonal interplay, and accounted for every last detail in our hand-crafted algorithms.
The result is the technology found in Flint Tremolo & Reverb. Pete Celi, our Lead DSP Engineer and Sound Designer illustrates the research and sound design process in the White Paper below.
Strymon Amplifier Tremolo Technology White Paper
Amplifier Tremolo Overview
Still incorrectly labeled as ‘vibrato’ in many cases, the tremolo effect is a cyclical amplitude (volume) modulation of the input signal. Although there are many cool tremolo effects that can be had by using a simple VCA (voltage-controlled amplifier) circuit and applying geometric waveforms (like sine, triangle, square, ramp) to modulate the amplitude, our interest is in exploring the unique, soothing, pulsing, hypnotic effect that comes out of vintage amplifier tremolo circuits.
There were three main variations that came about in the late ’50s and ’60s. The three types can be referred to as Harmonic Tremolo, Power Tube Tremolo, and Photocell Tremolo. These variations have unique characteristics that result from the very different ways that the effect is achieved
One thing that these vintage trem types share in common is the LFO (low frequency oscillator) circuitry, which is generated by a classic positive feedback ‘phase-shift’ oscillator. A network of resistors and capacitors determine the rate of oscillation, and the resultant LFO signal is a mildly distorted sinusoidal signal.
FIG. 1 PHASE-SHIFT OSCILLATOR
As the LFO circuitry is common to all three trem types under investigation, we can see that LFO waveshape is not responsible for the very different sounds that result from the three implementations. Let’s look closer at the three types.
The Harmonic Trem is actually not a pure tremolo effect. It is really a dual-band filtering effect that alternately emphasizes low and high frequencies. The end-result is a soothing pulse that has shades of a mild phaser effect combined with tremolo due to the nature of the frequency bands that are alternated. This circuit required two tubes to create a two-phase differential LFO that controls the gain of the two frequency bands, and then another tube to sum the two bands together. This implementation had a rather short period of availability perhaps due to the somewhat ‘expensive’ implementation. The basic idea is shown below:
FIG. 2 HARMONIC TREMOLO BLOCK DIAGRAM
One phase of the LFO signal is added directly with the low-band input signal, while the other phase gets added directly to the high-band signal. Essentially, the filtered signal ‘rides’ on top of the LFO signal on its way into the tube summing amplifier. This effectively changes the small-signal operating point of the filtered signal along the tube gain curve. When the LFO signal is at low voltages, the filtered signal will have more gain as the tube operates in its steepest gain region. Conversely, when the LFO is at higher voltages, the tube gain-curve flattens out, and the input signal experiences reduced gain. Since the two bands have opposite phase LFO signals, when one band is experiencing high gain, the other is experiencing low gain. When the two are combined, the opposite phase LFO signals cancel each other out, and the two alternating amplitude-modulated filtered signals comprise the output. This produces the tremolo effect of hearing a loud (bright) signal alternating with a soft (dark) signal.
Also, as a consequence of riding up and down the tube’s gain curve, the filtered signals experience slight changes in harmonic content due to the changing nonlinearities of the gain curve around the signal. This adds further complexity to the trem’s sound.
Power Tube Tremolo
Next in line was a more cost effective circuit that eliminated two tubes from the Harmonic Trem implementation. It used the LFO signal (no longer a two-phase LFO) to directly influence the power tube bias of the push-pull output stage.
FIG. 3 POWER TUBE TREMOLO BLOCK DIAGRAM
In a push-pull power amplifier, two tubes are employed and biased so that they idle at substantially less than full power. This keeps power dissipation to a minimum when no signal is going through the amp, allowing them to provide power to the speaker more efficiently while increasing tube life. The guitar signal is split into opposite phases so that one tube conducts when the signal is positive, and the other tube conducts when the signal is negative. The two outputs are added together through the output transformer.
By applying the LFO to the bias, the power tubes are being biased into lower and higher idle currents. At low idle currents, the tubes are shutting off and signal gain (volume) is reduced. At higher currents, the tubes are running hot and higher gain results. This alternating gain produces the tremolo effect.
But there is more going on than just a change in volume. Secondary effects coming into play are crossover distortion as the tremolo volume heads towards zero and the tubes are shutting off. At the other end, increased power tube harmonic distortion occurs as the tremolo nears its maximum volume. The effects of power-supply sag also contributes to some of the dynamic response when playing through this kind of tremolo circuit, as it influences the relative bias point of the power tubes. All these things add up to contribute to the ‘magic’ of this trem circuit.
The Photocell tremolo uses a light-dependent resistor (LDR) to attenuate the input signal. The LDR is coupled with a miniature light bulb that is connected to the LFO. As the LFO oscillates, the bulb gets brighter and dimmer which in turn varies the resistance of the LDR. The varying resistance works with other circuit impedances to change the signal level.
FIG. 4 PHOTOCELL TREMOLO BLOCK DIAGRAM
The light element used in the classic photo-trem circuits in the 60s was a neon bulb which has a very fast response time, meaning it turns on and off very quickly and spends very little time in between. This produces a characteristic ‘hard’ sounding tremolo that is moving between two levels, almost like a square wave. The duty cycle (symmetry) of the tremolo depends on the characteristics of the bulb relative to the LFO voltages, but the classic Photo-trem circuits were tuned to spend most of their time at the higher output level (duty cycle >>50%, bulb is ‘off’), switching to the lower level only briefly during the cycle. At maximum intensity, a choppy trem results.
Also, as the photocell trem circuit is not buffered, the tremolo creates a varying load resistance in the signal path as the bulb changes the resistance of the LDR. This in turn has secondary effects on the signal’s frequency response that contribute subtle characteristics as well.
Capturing the Magic
We can see from the discussions above that the end result of these vintage tremolo circuits is much more than a simple cyclical volume fluctuation. The depth, warmth and overall vibe of each one of these tremolo types can only be created by giving consideration to the entire circuitry used in the process. For the harmonic tremolo, the interaction of the LFO with the input signal in relation to the preamp tube’s operating characteristics must be accounted for. The Power-tube tremolo must recreate the vintage push-pull power tube section including the phase-splitter, tube characteristics, and power supply considerations. The photocell trem must involve the proper bulb-LDR characteristics in relation to the LFO signal, along with secondary consideration of variable loading in the signal path. When these things are all properly accounted for, the difference from a simple VCA tremolo is apparent. The complex and subtle nuances come to life, producing the mojo of their vintage amp brethren.
Posted by Ethan
The magical combination of tremolo and reverb is the earliest example of a perfect guitar effects marriage. Our new Flint Tremolo & Reverb pedal delivers three classic tremolo circuits, along with three completely unique and complimentary reverb types.
You get the classic ’60s Spring Tank Reverb, the inventive ’70s Electronic Plate Reverb, and the nostalgic ’80s Hall Rack Reverb. Pete Celi, our Lead DSP Engineer and Sound Designer illustrates the research and sound design process that went into creating our reverbs in Flint.
Flint Reverb Summary Paper – Three Classic Reverb Types
The ’60s Combo Amp Spring Tank
The full-size 2-spring tank was commonly used in vintage amps, and it continues its popularity today for its classic tones. The 2-spring tank uses spring segments of differing delay times (a function of the mass and tension of the spring), which adds to the complexity of the sound and smooths out the time and frequency response of the reverb. Contributing greatly to the sound are the input (driving) and output (recovery) tube circuits. These circuits are designed to reduce low-end boominess and to minimize coupling of the low- frequency cabinet resonance into the tank. The high frequencies roll off naturally due to the limits of the spring’s ability to transmit the shorter wavelengths of the higher frequencies.
FIG. 1 SPRING TANK REVERB
The signal from the driving circuit drives a coil which in turn produces a fluctuating magnetic field that moves a magnet attached to the spring. This results in a twisting wave that travels down the spring. The time it takes for the wave to travel down the spring is a function of frequency, with lower frequency waves traveling down the spring more quickly than higher frequencies. This accounts for the ‘drippy’ or ‘boingy’ sound that the reverb produces when given a percussive attack. At the other end of the spring, the signal is recovered by the inverse process which includes coils, magnets, and a recovery circuit. In addition to being recovered, the wave will continue to reflect back and forth along the spring, creating a wash of reverberation that evolves in time due to the frequency-dependent delay times of the spring. The length of time that the reverb lasts when given an impulsive input is known as the ‘decay time’, which is controlled by physical dampers that absorb energy from the spring.
At low mix levels, the 2-spring tank adds a depth and dimension to the sound. Generally speaking, the 2-spring combo-amp reverbs tend to sound a bit less splashy and trashy than their 3-spring stand-alone counterparts at the extremes, but add a full, integrated explosion of sound when cranked up.
The ’70s Electronic Reverb
During the 1970s, digital electronic systems advanced to the point where high-quality real-time electronic reverberation was possible. A single memory chip was capable of storing 1024 bits, and the possibilities seemed endless. The most famous early electronic reverb was a $20,000 plate-style reverb that used eighty(!) of these memory chips. The amazing hardware-based algorithm used multiple delay- lines configured in parallel, with each delay featuring multiple output taps and filtered feedback paths.
FIG. 2 SIMPLIFIED ELECTRONIC PLATE REVERB STRUCTURE
The lengths of the delay lines and individual taps were derived mathematically to produce the most natural reverberation. The reverb algorithm also employed modulation by mixing various taps under internal control to create changes in reflection phases to further reduce undesirable resonances and add depth. The result is a rich, smooth reverb with a quick build-up in density due to the summation of the many parallel output taps.
The ’80s Hall Studio Rack Reverb
By the late ’80s, continued advances in digital ICs and microprocessors lead to (relatively) low-cost digital reverbs that could run many different reverb algorithms and allowed for preset storage and deep parameter editing. Cost sensitivity and the limited available processing power of the day led to the necessary invention of efficient algorithms with minimized computational and memory requirements. To create a Hall-style reverb, a well-practiced technique was to create an early reflections section that fed into a late reverb generator.
FIG. 3 SIMPLIFIED ’80s HALL REVERB
A simple multi-tapped delay line was sufficient to create early reflections. The late reverberation was accomplished by a regenerating ‘series-loop’ of delays, all-pass filters, and low-pass filters. Inputs could be injected into the loop in more than one place, and the outputs might consist of the summation of several points from the loop. Delay-line modulation was employed to reduce artifacts and achieve a smoother, more pleasing decay. These hall reverbs have a signature sound of distinctive early reflections followed by the slowly-building density of the late reverberation. The modulation adds an increased sense of warmth and depth.
Enter the World of Flint
The three reverb types in Flint pay homage to these three classic reverb sounds. While not focusing on any specific recreation, these classics served as philosophical and sonic guides in the creation of our ’60s, ’70s and ’80s reverb types.
Posted by Ethan
Our very own Analog Guru and co-founder Gregg Stock recently had the opportunity to contribute an article for the March 2012 issue of Premier Guitar magazine.
In 1830, Michael Faraday submitted his most famous piece of scientific legislation‚ and this bit of genius described the physics that allows guitar pickups to exist. In this article, Gregg goes over the inner workings of guitar pickups, and how pickup loading can affect your true bypass pedals.
Posted by Pete
Soon we will be releasing our El Capistan dTape™ Echo. To capture the full experience and complexity of a tape echo machine, every last tape system attribute was relentlessly studied and faithfully recreated. The result is Strymon dTape™ Technology. Pete Celi, our Lead DSP Engineer and Sound Designer illustrates the research and sound design process in the White Paper below.
Strymon dTape™ Technology
Tape Echo Overview
Tape Echo Machines work similarly to traditional tape recorders, using electromagnetic heads to record, play back, and erase audio on a magnetic tape. A tape echo machine mixes the input with the playback signal, and adds the ability to send the playback signal back to the record head to get re-recorded (and played back) repeatedly, creating an echo effect. The time between these echoes is a function of the tape speed and the distance between the record and playback heads.
As can be seen from the above equation, the echo time can be changed by changing the tape speed or by changing the distance between the record and playback heads. As a result, there are two types of tape echo machines – variable tape speed with fixed heads (most notably the Roland Space Echo*), and fixed tape speed with moveable heads (most notably the Maestro Echoplex*). A variable tape speed machine may have multiple fixed heads that can be selected, and a moveable head machine may allow two or more discrete tape speeds. We’ll look at the sonic implications, similarities and differences of these two different categories of tape echo machines in a bit.
Part of the allure of the vintage tape echo units is the way the characteristic imperfections of the system affect the sound of the echoes. Tape echo units are mechanical systems that involve motors and moving parts that are subject to alignment issues, friction, aging, warping, etc. Additionally, the recording medium itself – magnetic tape – is prone to degradation and wear in a variety of manners, even in the presence of a (non-existent) perfectly tuned mechanical system.
The tape makes its way from the record head to the playback head by the effect of a motor-driven capstan and a pinch roller, which effectively ‘squeeze’ and pull the tape. Tape speed variations can be introduced by fluctuations in motor voltage, deformity of the pinch roller, bent capstan, maladjusted tensioner, worn bearings, and the list goes on. Since the delay time is directly related to the tape speed, any inconsistencies in the tape speed will result in inconsistency of the delay time. A small variation in delay time wouldn’t itself be noticeable, but variations in tape speed also results in variations in playback pitch, which is quite noticeable, even at small levels. The presence of some level of tape speed variation contributes greatly to the ‘three dimensional’ quality attributed to tape echo machines.
The quality and age of the tape itself will have a noticeable impact on the sound of the echoes. Old tape that has surface contaminants can cause deviations in the tape speed as the tape can ‘catch’ briefly due to friction effects. This effect occurs at higher frequencies than that of motor and pinch-roller deviations, and can result in a ‘garbled’ sonic quality. Tape splices and sections of tape that have been creased also will also impact the echo sound.
With old tape, or tape formulations of inferior quality, the ability of the tape to be magnetized at high frequencies is impaired, resulting in overall ‘dark’ repeats, and sections of ‘dropout’ where the bandwidth is severely reduced for brief periods. The effect is progressive when high levels of feedback are used to generate many repeats.
Magnetic tape is a highly non-linear recording medium. There is a threshold below which signals cannot be recorded, causing objectionable crossover-distortion. To overcome this problem, a ‘bias’ signal is introduced. The bias signal is a high (inaudible) frequency tone that is added to the desired record signal. This allows the desired signal to be recorded in the linear region of the tape and avoid the crossover-distortion region, even when the desired record signal is small, provided the amplitude of the bias signal is large enough. The bias tone is removed with a low pass filter during playback.
When recording large signals, the tape will saturate and cause soft-clipping distortion as the tape is unable to become further magnetized. From this standpoint, it is desirable to limit the size of the signal being recorded to the tape. This, however, is in conflict with the bias signal whose purpose is to write large signals to the tape to avoid crossover distortion. Thus the bias level plays an important part of tape recording quality. Some prefer to intentionally set the bias higher or lower than the recommended specification to achieve a particular sound that suits their preferences or particular equipment setup. Higher bias levels result in reduced echo volume and limited headroom, as more of the linear region is ‘used up’ for the bias signal. This causes the echoes to quickly degenerate into a wash of saturated harmonics with higher feedback settings. Lower bias settings (just above the crossover distortion point) result in the cleanest echoes with the most headroom, suitable when high fidelity of the echoes is desired.
The speed of the tape also impacts the fidelity of the echoes. The playback head reads the magnetized signal on the tape contained within a narrow window, or aperture, and the resultant playback signal is the average magnetic signal contained within the aperture. With high-frequency signals, both positive and negative parts of the waveform will be contained within the aperture and the two halves will start to cancel each other out. The playback signal will have a high-frequency roll-off as a result. As the tape speed is reduced, this effect will start to happen at lower frequencies. This results in darker echo signals as the tape speed is reduced.
Tape speed also plays a role in how the mechanical and tape quality imperfections impact the record and playback sound. Pinch-roller deformities and capstan problems result in quasi-periodic speed fluctuations that are directly proportional to tape speed. Tape friction and maladjusted tensioners may cause more problems at lower tape speeds as various friction components are relatively more difficult to overcome at lower speeds.
Tape Echo Machine Types in More Detail
The variable tape speed and moving head machines react differently to some of the factors discussed. Next we’ll look at each type of unit in more detail.
As the tape speed is slowed to increase the delay time, the repeats will get darker due to the playback magnetic aperture effect. The low-frequency fluctuations of the mechanical system will also slow down proportionally with the tape speed, so the wow and flutter effects might be noticed more as ‘pitch warbling’ as opposed to ‘dimensional depth’. The higher frequency deviations contributed by the tape itself will also slow down and become more noticeable as a ‘garbled’ sound at lower tape speeds.
This degrading of sound at slower tape speeds led to the inclusion of multiple heads to extend the overall range of delay times. A playback head placed close to the write head provides short delays for slap, rockabilly, and spacey sci-fi sounds. A playback head placed further away from the write head provides longer delays. Additional combinations of the playback heads can be activated simultaneously for rhythmic echo sounds.
When the delay time is adjusted with repeating echoes present, the varying tape speed will result in pitch deviations proportional to the change in tape speed. So doubling the tape speed (cutting the delay time in half) will result in the repeating echoes being pitched up by one octave as they continue to get played back and re-recorded to the tape. Bringing the delay time back to its original point will bring the repeated echoes back to their original pitch, and any echoes recorded at the higher speed would be pitched down one octave. Since the heads are fixed, the signal gets recorded to the tape at the same speed that it is being read from the playback head, even when the delay time is changing. We’ll see how this differs from the variable head machines next.
Moving-head machines don’t exhibit any change in repeat quality or variations in wow and flutter characteristics as the delay time is changed. These characteristics are determined solely by the tape speed, which stays constant. However, the tradeoff of maximum delay time vs. overall repeat quality is still present. Practical constraints on physical size determine the maximum distance between the read and write heads, so the maximum delay time depends on the tape speed that is chosen for the system. Long delay times require a slower speed which results in reduced echo quality for all delay times. Higher tape speeds produce higher quality echoes but limit maximum delay times. As a result, some moving head designs allow for a selection of multiple tape speeds to extend versatility.
The most well-known movable-head machines have a fixed playback head and a movable record head. In this type of system, several interesting things happen when the delay time is changed (by moving the record head) with repeating echoes already playing back. When the record head is moved, the relative speed of the tape at the record head is the difference between the tape speed and the speed at which the record head is being moved. The echoes are re-recorded onto the tape at a different speed relative to the stationary playback head. This results in a pitch artifact that is dependent on how fast the record head is moved and independent of change in overall delay time. Moving the record head back to the original delay time will not restore the original signal, even if the record head is moved back in exactly the reverse manner in which it was originally moved. This is because the actual length of tape between the heads changes with delay time, unlike the variable-tape-speed machines. Also, since it’s the record head that is moving, we won’t hear the effect of the delay change until that section of tape reaches the playback head.
Another ‘feature’ is that with longer delays and some dead space between repeats, you can move the write head towards the read head during the playback of the dead space (quickly, chasing the repeat signal that just got written) so that the subsequent shorter repeats happen without any noticeable pitch effects at all.
Strymon dTape™ Technology
To fully capture the experience of a tape echo machine, all of the attributes of the systems listed above must be accounted for. The result is Strymon dTape Technology.
Representation of Mechanical System
The mechanical systems of tape echo machines were thoroughly studied to accurately capture the nature of their imperfections. Multiple levels of quasi-periodic fluctuations from motors and mechanical components occur with pseudo-random fluctuations that span several decades of frequency disturbances. Some of these disturbances occur at both heads simultaneously, while others occur at the heads independently.
To allow for any type of tape machine to be re-created, dTape Technology allows for independent control of tape speed for each head in the system, along with independent control of head position. In this manner, heads can be fixed with varying tape speed, or tape speed can be fixed with variable head position. This powerful ‘super system’ also allows for realistic implementation of the mechanical imperfections that occur. For example, the record and playback heads can experience different instantaneous tape speeds if the tape binds momentarily. A motor fluctuation will result in a tape speed disturbance that is experienced by both heads simultaneously. A tape defect will travel from the record head to the playback head. These effects happen naturally in the dTape system, which allows for user control from ‘perfectly tuned’ to ‘in need of service’.
Additionally, the dTape system allows for independent control of the motor/mechanical disturbances and the tape friction disturbances. As the tape becomes smoother, its influence on tape speed diminishes. A tape with no contaminants, splices, or creases can be achieved by reducing the tape/friction component to zero. This ‘perfectly smooth tape’ is representative of a magnetic drum type of recorder, which operates on the same electromagnetic principles of a tape system, but uses a revolving magnetized drum as its recording medium.
The effect of tape speed, including aperture-related bandwidth effects and frequency scaling of mechanical wow and flutter is fully accounted for in the dTape system. In variable tape-speed modes, repeats degrade and wow and flutter tracks as tape speed is modified. In movable-head modes, repeat quality and wow and flutter remains constant as the delay time is varied. Selection of a different tape speed results in an accompanying change in bandwidth and wow and flutter characteristics.
Bias level adjustment is a significant contributor to the character of the repeats. Strymon dTape faithfully recreates the bias adjustment from under biased to over-biased. As the bias level is increased, the headroom and level of the repeats is reduced. The bias can be adjusted to allow for easy transition into saturated oscillation without an increase in the volume level of the saturated repeats. Lower bias settings work well with hot input signals, while lower level inputs can tolerate higher bias settings.
In addition to playback-head aperture effect, the frequency characteristics of the repeats are determined by the actual tape bandwidth, system filtering and head-tape alignment. Older tapes will have a lower bandwidth with a warmer top end. Misalignment of the heads can also greatly reduce the high end. The low frequency characteristics result from a combination of the electro-magnetic properties of the record/playback process, and the system filters. The overall result is a reduction of low frequency content.
The dTape algorithm allows for independent tailoring of the high- and low-frequency content of the repeat signal, allowing for repeats characteristics that range from full-bandwidth, to dark and warm, to the extremely high-passed nature of some magnetic drum recorders.
The various complexities of tape machines all contribute to the overall experience. When they are all accounted for accurately, the experience can be re-created with quality and reliability. Strymon dTape Technology allows for various tape machine types to be re-created authentically, including the wow and flutter, tape friction, bias adjustment, oscillation, saturation, and delay-time-adjustment artifacts. Furthermore, user adjustment of the parameters allows for a range of tape experiences unattainable with a single traditional tape machine.
*All product names used in this article are trademarks of their respective owners, which are in no way associated or affiliated with Strymon or Damage Control, LLC.