Delay - So What's Under the Hood? September 14 2014
Caution: I may get into nerd territory in this post, so go in knowing that I'll try to keep it at a "common sense" level, but may go off the rails. This is also just my understanding of the topic, so take it for what it's worth. You've been warned...
"Is this a digital or analog delay?" is one question that I often get about both the Fuzzy Memory and other delay pedals I've built. While this is an easy question to answer, the implications are a little more complicated.
I use PT2399 chips in my delay pedals, which use "digital" technology to delay the incoming signal. This means that the audio signal is transformed into a series of 1's and 0's, then that data sits in the chip's RAM until the next pulse comes from the chip's clock, at which point in time it is pulled from RAM, transformed back into audio, and then fed back out in a feedback loop to the chip's input. By feeding it's output back into the input, you get repeats that either stop immediately after one repeat (slap-back echo), trail out after a time, or increase in volume in an infinite fashion (what you will commonly hear referred to oscillation). The type of echo is determined by the amount of resistance and filtering that take place in the feedback loop. As the repeats get fed back into the chip each time, there is some loss of resolution, so what starts off with a relatively high resolution in the first repeats degrades into lower and lower bit audio resolution. In other words, digital echo chips like the PT2399 are limited by the fact that the analog to digital conversion changes the signal a little bit each time. In theory, a digital delay with good enough converters could repeat the original signal for many, many cycles with very little degradation, kind of like a rudimentary looping pedal.
On the other hand, "analog" delay pedals utilize "bucket brigade" chip technology first designed back in the late 60s. It got this name due to it's design, where it loads the analog signal as a charge to a capacitor, which pass the signal on to the next stage with a transistor being used as a switch to control the passage of the signal. This process continues through the remaining stages of the chip, each time slowing the signal down. Just like in digital chips, a clock source controls the speed of the switching between the stages of the bucket brigade (BBD), and that controls the delay time of the chip. The output is also fed back into the chip, just like with a digital chip, except the signal is kept analog throughout the signal chain. There's a lot more to the process, but that's a 30,000 foot level view.
When it comes to the differences in sound, in my mind, it really comes down to two things: the resolution of the audio and the change of the original audio over the course of several repeats. Remember audio in early Atari games? Just a series of bleeps and bloops, right? Then as video game consoles were able to increase in resolution, you started getting more complicated sprites and audio, and eventually even 3D models and CD quality audio. Digital delay chips exhibit that evolutionary process in reverse, degrading the signal's original resolution over time. There's less and less data for the chip to convert back into audio. BBD chips never convert the audio to a series of 1's and 0's, so there is no loss in resolution. Where BBD chips alter the original sound is in the degradation that occurs as the sound passes through the capacitors in the chip. There is also clock noise introduced to the signal, which has to be filtered out using low pass filters. The "cork sniffer" crowd will often refer to this analog signal degradation and (especially) low pass filtering as "analog warmth". Low pass filtering is also used in digital delay applications, however the signal degradation can introduce digital "artifacts" (which I think of as mistakes in the translation of the audio back and forth to binary information) and distortion so you have to keep that in mind when designing the circuit. The good news is that, with tasteful filtering applied, digital delay chips can produce some very natural sounding echo sounds that I think can more than hold their own against analog chips. It really comes down to understanding the limitations of both analog and digital chips and playing to their strengths.
Beyond the sound of the chips, there are also differences in terms of availability, price, and consistency between analog and digital chips. The PT2399 chip is pretty readily available, fairly inexpensive, and relatively consistent from chip to chip (at least, in comparison to analog chips). Those are being used in many pedals that you'd swear were analog without knowing better. BBD chips are now being made by companies like CoolAudio, so they are easier to get than they once were, but to really capture the sound of the originals, you have to search out new old stock (NOS) chips. Get prepared to spend some money. You're also taking the gamble that the chips will behave consistently across applications and be consistent within a batch. Good luck. A third option is DSP (digital signal processor) chips that pack an amazing amount of horsepower into some relatively compact pedals. Have you heard the current pedals coming out of makers like Strymon, Eventide, or TC Electronic (their Alter Ego x4 in particular)? They are just flat out AMAZING! Anyone that tries out one of those pedals and then starts complaining that they aren't "analog" seriously needs to have their head examined. Strymon's algorithms are setting the bar for what companies will be trying to emulate for years to come. Take a look at the relatively recent release of the Zoom MS-70CDR, and you'll see emulations of Strymon, TC Electronic, and Eventide in there.
So in the end, I will say this...there are examples in the signal processing world where analog is best. I believe that drive pedals, whether you're talking distortion, fuzz, overdrive, or a boost to add a little dirt, are still best in analog form. Some companies are getting better at certain types of digital drive sounds, but I still feel that they are a little wide of the mark, mainly because it is hard to build algorithms that can translate 0's and 1's into the behavior of an analog clipping device like a diode or transistor, or how an operational amplifier will react, and, more importantly, all of that reacting in the context of a larger signal chain. That's why the software that I use when designing a circuit will only give me a basic understanding of how a drive circuit will react, and from there I have to build it out on a breadboard to then tweak it to match the sound in my head. Fuzz pedals are especially hard to model in the digital realm due to their unpredictable nature being one of the aspects we love most about them in the first place. Delays and reverbs are one area where I feel that digital has finally eclipsed the benefits of analog. Some DSP based digital delay pedals still come across sounding sterile, but many have perfected the ability to get warm, analog sounding repeats without the negative aspects of analog, you'd never need to worry about your vintage analog delay pedal pooping out on you during a performance (or getting stolen from your gear closet). It's not taking anything away from the beautiful, organic sound you get from analog delay pedals, it's just that you are no longer limited to analog if you want that great sound. And, if something were to go wrong with your DSP based pedal, you can buy another one and not worry about it reacting differently than your last one in the context of your pedal chain. It's really a great time to be a guitar player, and we shouldn't limit ourselves to matching the gear of our heroes, pedal for pedal, when there is so much exciting stuff out there. Heck, I use some delay pedals in my rig, and some of my delay and reverb sounds come from a Lexicon rackmount unit. Getting the signals to behave between the guitar rig and the rack gear is a PITA, but that's a discussion for another day. Point is, don't let anything hold you back when it comes to finding your sound.