personalized expeditions in music from a musician who doesn't matter

STEEL MY IDEAS PLEASE: Two theoretical advancements in steel guitars I’ll probably never be smart enough to build

Posted by r on February 17, 2014

A couple of proposed DSP-driven pedal-steel-substitute instruments courtesy of my thick head. This post could probably use some helpful drawings, but I’m not going to provide any, so there’s also that.

My fascination with steel guitars of all kinds is now entering its fifth year, having hit me out of nowhere somewhere in the winter-spring of 2009. You’d think by now I’d actually be able to play one, but a few dumb things / personal hangups have held me back from practicing on a scale commensurate with my degree of theoretical obsession.

One of the things that really bothers me about these beautiful instruments: Everyone who knows how to play one is quickly gettin’ to dyin’ age. And it’s not just because this is an instrument whose time has passed it by.

The bar (no pun intended) for entry for younger players who might do something really exciting and new with the things is quite high, at the very least in the case of pedal steel guitars. We’re in an age where playable $100-and-down guitars are absolutely everywhere, thank you China (for now). But between laws of supply / (much-smaller) demand and the degree of craftsmanship required to build a playable instrument with facility for mechanically changing its tuning on the fly, it’s a challenge for even a curious young player to obtain a passable entry-level pedal steel– even used!– for much less than 15-20 times the cost of a secondhand or even brand-new Squier Strat.

Add to this that the pedal steel has evolved almost entirely around the idiomatic requirements of country music / is not easily modified to do other things, along with its ten or more strings and extremely asymmetrical, beginner-alienating “standard” tuning / copedent, and you’re left with an instrument that will scare off 99.9% of any possible musicians in the Twitter-and-Garageband generation almost as soon as they start learning about it.

The PSG, for all its seemingly-effortless shimmering beauty on record, is basically a master-level game of chess in Western wear… with an initial price tag even more shocking to a curious outsider than a decent-looking cowpoke shirt. (Every time I go into a Western-wear shop in the hopes of getting my needless fringe on, I always leave empty-handed and wondering how the hell cowboys can afford to be cowboys… but I digress somewhat.)

I earnestly dream of a related, similarly-capable instrument that could be mass-manufactured and sold much more cheaply, and/or at least built at home cheaply in a weekend without a machinist’s expertise (to say nothing of a machinist’s expensive power tools). I definitely dream of a steel-like instrument that would allow much more drastic and flexible changes of tuning – and even realtime “modulation” of that tuning – than “simple” steels can allow.

One possible way to build such an instrument, of course, is by harnessing the cheap CPU power and infinitely-copyable software available to us in 2014, replacing the expensive, inevitably-hand-crafted mechanical “drivetrains” that currently enable the pedal-steel player’s “tuning modulation maneuvers.” I’ve had many ideas for slapping such a new-fangled virtual steel guitar together, but lately I’ve had two such ideas that I think could actually be implemented fairly easily with good results… and with a drastic reduction in manufacture costs vs. a conventional PSG.

So here we go – let’s-get-real descriptions of two tech-driven theoretical steels that I theoretically wish someone more tech-savvy / driven would steal and make real. One way or the other, I’d love to see these possibilities end up in the hands of kids who could make beautiful music with them, and for the one-of-a-kind steel guitar to thusly receive a lifetime extension in our culture… albeit in technologically-enhanced, newbie-friendlier form.

Model A: BUT THAT THING AIN’T GOT NO STRINGS. The “steel” itself basically becomes a control-signal-generating device, set up on a small “table” in front of which the player may sit (or stand).

Replacing the bar, on the left hand: some kind of mechanism for registering the player’s “bar-hand” position (along with an intonation-guiding “decorative fretboard” just like classic steels). This could be a ribbon, a series of magnetic sensors, a camera-based setup requiring the player to hold some kind of camera-fixation implement, whatever.

The right hand of the player would be fixed on “pluck switches” capable of registering attack velocity, and would also allow the player to stop a given “string” easily (perhaps the buttons could be levers like a power-window switch, but also able to sense when fingers were resting on them without pressing / pushing anything). There would probably be something like 4-5 of these arranged in an ergonomic, natural pattern.

Pedals / levers / etc. would still be attached to all of this, and would likewise send “deeply variable” control signals to a computer of some kind, preferably with a small heads-up display attached to the “control table.” The computer would be responsible for not only processing all of this control-signal input but also for tone generation, with a synthesis module based on physical modeling of a PSG’s natural strings as closely as possible. The computer could also send complex OSC output from the translated control signals, and its proprietary internal “steel instrument” could also receive / sonically respond in realtime to the same information in its proprietary format returned from, say, a sequencer / DAW.

Advantages of Model A:

  1. Infinite tunings / copedents that can not only be “gently modulated” a la a PSG’s available changes, but actually drastically / totally changed from player input / presets on the fly, with the option to sustain the notes sounded just before such a virtual tuning change for sonically seamless switching.
  2. “Tuning modulation” from pedals / levers etc. can take place over a much greater intervallic range than the minor-third-absolute-max possible pre-string-breakage on a mechanical PSG.
  3. Advantage #2, in conjunction with on-the-fly variable “meanings” of pitch-pedal control input, could mean a guitarist could make use of a simpler, easier-to-master tuning / arrangement of the virtual strings themselves. Right-hand technique would be easier, perhaps, for new players to master, since the picking hand would rarely if ever have to move in the “Y” plane (instead of swapping the hand around on higher / lower strings, you’d simply adjust the register / tuning of the fewer-in-number “string buttons” with a “register shift control”).
  4. New players could have the synth module “auto-swell” attacks for them instead of using a traditional right-foot volume pedal, or could have any other number of parameters auto-modulated for particularly wild and PSG-impossible effects.
  5. The completely OSC-encoded output of the controller means that a player could carefully study and possibly adjust their recorded performance, a la fixing a single bunk note from a MIDIfied keyboard take in a sequencer.

Disadvantages of Model A:

  1. It’s probably still going to be very difficult, if not impossible, to totally emulate the sound of the PSG / responsiveness of real strings with physical modeling or any other synthesis method in our current era. While this instrument still has some strikingly new and beautiful possibilities, the sonic result will probably be more like an extremely expressive one-player polyphonic theremin than anything that sounds convincingly like a steel… and accordingly, might need to be marketed as kind of its own thing to a new market vs. any more conservative musician who might be interested in a computerized steel guitar.


The “control platform” for Model B would be essentially a (barely) glorified lap steel guitar, with real strings, real tuning machines at the headstock, and no sensors / switches anywhere… at least up top. Again, compared to a 10-14 string PSG, the number of strings would be limited – perhaps 4-6. The main thing differentiating the control platform at the “hand level” from a conventional plank lap / console steel would be the pickup; it would have individual polepieces and thusly separate audio outputs for each individual string.

On the floor / underneath the “plank,” the pedals and knee levers would again be electronic / control-signal-generating devices. Although attached to the body of the instrument, they would not be physically connected to / able to directly modulate the tuning via the “string drivetrain” of the lap steel “upstairs” in any way.

The individual string-audio outputs and the control signals from the floor would again be registered by a computer. The idea with Model B is that individual string tunings / outputs would be digitally pitch-shifted by the onboard computer prior to output to the amp.

This is very similar conceptually to the highly-controversial Peavey Autotune guitar, except that Model B is made to register realtime pitch control input via the pedals and knee levers. The tuning-changing pedal action of a PSG takes place entirely inside the CPU instead of “under the table.”

Advantages of Model B:

  1. Probably cheaper to implement / build overall than the Model A approach, or could even be offered as a retrofit kit for already-existing lap steels.
  2. More likely to be capable of generating a semi-authentic-sounding “real steel” sound vs. Model A, and probably will feel more familiar to (very common) “conversion guitarists.”
  3. Still offers a much more flexible range of possible tunings / virtual copedents and possible on-the-fly switchability of said tunings / copedents, theoretically expanding the possibilities of the mechanical pedal steel into exciting new territory.

Disadvantages of Model B:

  1. Again, much rides on the quality of the pitch-changing DSP in the box. Even in 2014, it’s not easy to make a smooth pitch-shift algorithm that won’t sound less like a guitar and more like gremlins imitating a guitar.
  2. The range of convincing-sounding pitch-shift from pedals etc. as applied to an audio signal is going to be more intervallically limited than the range possible with a purely synthesized signal. The pedal shift range available on tap before gremlinization might not be much more than that available from a mechanical PSG.
  3. The discrepancy between the acoustic / unamplified output of the instrument and the repitched / pedal-modulated outputs might be persistently disorienting to some players and/or require them to turn up their amps higher than they’d like in order to “wash out” the unplugged sound of the lap-steel platform, or to simply get good intonation when using the pitch-modulating controls.

5 Responses to “STEEL MY IDEAS PLEASE: Two theoretical advancements in steel guitars I’ll probably never be smart enough to build”

  1. Mister Voxius said

    why not little servos attached to tuning pegs and controlled by a (or several) Raspberry Pi? it wouldn’t be instant but those tuning changes would sound amazing. combine this idea with Model B.

    • r said

      Hey, A, thanks for the read (as always!!!) and comment. Sorry you got run through the automated spam-ringer somehow, hopefully not a problem for you in the future.

      I was thinking about something like what you propose a while back… wow, I guess it was 18 months ago now!… and there are some interesting cats who know both their servo-sorta stuff AND their steel guitars throwing in on that thread I just linked.

      The bottom line, most likely, is that you’d end up with an instrument just as complicated and expensive as the steel proper, if not more so, and it wouldn’t do much to overcome the steel’s inherent limitations (although you COULD implement a broader / more flexible range of tuning changes a la Model B). My biggest worry with a servo-driven changer (yes, that’s the steel-word) is, as you said, that it wouldn’t be essentially instant nor consistent / touch-learnable in response, and that would itself sort of kill off a lot of the appeal of the steel. Even beginning players quickly learn to use the pedals and levers variably / expressively, that’s a very important aspect of the instrument, and quick response of the changer / pitch-slide-y mechanics is particularly vital for many of the PSG’s most idiomatic licks.

      I’m mainly trying to think of things that would not quite *be* the steel as we know it but would allow a lot of the same expressive possibilities, and then some, while being much cheaper to build and remarket. It’s possible to build your own pedal steel if the price of an off-the-shelf instrument is a barrier, but it’s not only very difficult / intimidating to do so, it’s also kind of impossible to do it well or right unless you’ve already owned your share of said instruments made by others, which creates a catch-22 situation. If you’re building something that is largely based on software / CPU and has only very simple “hardware” involved, you’ve got something that a DIYer would stand a much better chance of building successfully with fewer tools / expense / risk. And then I guess it could be a gateway drug to the real thing if they liked what they’d built.

  2. aren’t there midi appendages that could do it? I seem to remember Paul Leary making a midi guitar album and I know Belew does similar things. Roland had a thing out where you could program different instruments and pan settings on each string. this was a decade ago at least, that must have gotten cheaper. likely designed for the six stringer though

    • r said

      Yep, MIDI guitars use individual-output pickups like the one I describe for Model B, often along with some other method of determining per-string fretted pitch to improve accuracy. It would be possible to maybe get a B-bender sort of thing going (Google it 😉 ), but an off-the-shelf MIDI guitar is still not going to be playable like even a lap steel, let alone a pedal steel. The pitch conversion in MIDI guitars is meant for tracking slightly bent fretted notes as its most “deviant” possible input. It definitely isn’t ready for things like bar runs / glissandi that might span 2+ octaves before the player is finished and take place in a fraction of a second.

      Beyond that, I suspect the old-school 8-bit resolution of MIDI messages is likely to be a problem for getting a remotely playable / steel-like result from either the input controller pedals / levers *or* the synth / audio-transposition engine itself. I think you’d quickly need more detailed, bandwidth-heavy messages going everywhere than ye olde MIDI standard can really reasonably handle, which is why I’m thinking OSC in my description of Model A instead.

    • r said

      MIDI guitars haven’t gotten all that much cheaper / better, either – law of supply / demand. There’s now a toy version called the You Rock guitar that sits around $150. I have no idea if it works well at all. It’s definitely the cheapest such option on the market by about $800ish.

Sorry, the comment form is closed at this time.