Analog electronics in the 2000's. Why we still use them, will they disappear?

[Thecomedian]

I was recently reading a book called The Art and Science of Analog Design, and I discovered some passages that claim analog circuit production and growth and application is growing at a slower rate than digital, and it leaves me questioning why we still use analog vs digital.

For DIYers and stomp boxes, of course there's a lot of analog parts and designs flying around, but is there a function for analog in today's high tech world? What's the cost or trade off for analog vs digital circuits in applications from stuff like medical to science? How come we don't convert wholly to digital? Apart from the appreciation for analog forms of information from CRT to film grain and music, isn't digital simply better in most cases where the brain's preferences don't matter?

[greaser_au]

I reckon it's what you're used to.
People are used to hearing their cymbals swirling in compressed audio - thus iturds has an enormous market - personally i'd rather a few crackles and a little 'surface noise'. I can stand a couple of percent of 2nd harmonic distortion!

People are now used to seeing what were perfectly graduated colour tones from black to 100%  appear as a set of 8 bands of colour, so jpeg/DVD/Bluray/cable/HDTV are now completely acceptable (give me film grain any day).

The multieffect pedal or amp that gives every beginner a fantastic (but totally identical) sound sell in thousands...  bust out the dsPIC and fine tune your germanium transistor algorithms!  

david

[GibsonGM]

Well said, David!

BTW - as long as there is still a demand for analog stuff, as well as the electronic stuff WE use (TO-92 transistors, opamps etc), they will continue to make them, just like vacuum tubes.

[PRR]

> isn't digital simply better

No.

Analog is generally cheaper, because each device covers a full dynamic range. Digital uses similar devices, overdriven to 0 or 1, and must use many more devices to have any dynamic range.

However devices have become obscenely cheap. A million for a buck? (In a micro-computer such as a PIC.)

And at some level of complexity you can do things that would hard or improbable in analog. You can re-touch facial wrinkles with a pencil on the film negative, but in digital you can run an algorithm which seeks specific shapes in a digital image and modifies them. A user-set 100dB/Oct audio high-pass is a royal pain in analog, fairly trivial on a Pentium or better.

> How come we don't convert wholly to digital?

It's an analog world and there will always be analog around the edges. Microphones (though mike capsules now emit digi-streams). Loudspeakers are quite hard to do digital, so we use analog speakers driven by an analog voltage (but now generated by "digital" amplifiers {which are often analog in time rather than voltage}).

Otherwise digital will take over all jobs. The classic very-analog problem of radio reception is already nearly full digital. All that regenerative and superhetrodyne stuff is gone: the antenna can feed a ADC nearly directly, and then software picks-out the specific signal and demodulates it.

[jubal81]

McDonald's has a ridiculously high calorie-to-cost ratio, not to mention it's far more convenient. In the future, won't everyone eat only food from McDonald's?

[Mark Hammer]

Fundamentally, the rendering of some audio transformation by digital means requires that one be able to describe that transformation in terms of an algorithm.  If you can't describe the change, you can't produce or reproduce it.

To my mind, this is why digital FX have always done a top-notch job when it comes to things that are easy to translate into algorithms (take the sampled input, wait XXmsec, and then output it, in the order they samples came in), and why some sounds came later (e.g., digital phase shifters and flangers), while others (distortion of many kinds) still seem to await perfecting.  The challenge has laid partly in the limits of the technology itself (i.e., speed), but mostly in our ability to describe what we like to hear in mathematical/computational terms.

Certainly, where digital excels is in packing ever-increasing numbers of feature-sets in a small space for a low price.  As Paul accurately notes, a lot of what digital-FX can do can also be done in analog form...but it would require a lot more parts, space, and cost, and likely be temperamental or twitchy.  

I would be remiss if I did not also note that digital does a better job at keeping proprietary things proprietary.  Many of us here can readily reverse-engineer an analog circuit, but faced with a PIC or ATMEL, and a couple of VLSI logic chips, on a multi-layered board, would simply not even know where to begin, let alone complete the task.

I suspect we will still continue to use a number of analog devices for a number of years to come; partly out of mojo and nostalgia, but probably because there is a certain degree of indeterminacy we like about how they behave in semi-predictable ways.  Wecertainly want effects that do most of what we want, but we also crave at least a teeny bit of surprise and pushback from the effect, and analog will do that, where digital not so much.

[R.G.]

...  in most cases where the brain's preferences don't matter?

What cases are there where the brain's preferences don't matter?

[Quackzed]

The challenge has laid partly in the limits of the technology itself (i.e., speed), but mostly in our ability to describe what we like to hear in mathematical/computational terms.

That is an extremely insightfull truth imho, and though we CAN digitally manipulate any number of things via lines of code; knowing how to say it to a computer or even fully understand the full scope of what needs to happen to which and when is not always simple or obvious in terms of lines of code. 'there is many a slip, twixt the cup and the lip!'

[Transmogrifox]

With respect to musical effects reproduction I don't think analog will ever disappear.  I think Ham radio operators are a good example of long-standing analog enthusiasts.  They have a place for DSP equipment, but when it comes to simple circuitry that any amateur can troubleshoot with nothing more than fundamental knowledge of electronics, then simple non-optimal analog circuits show an advantage.  

Even if in the future the algorithms used to produce analog sounds are perfected, then there will still always be analog enthusiasts -- even if for no other reason than historical nostalgia.  

Take for example the small historical organizations who experiment with black gunpowder cannons.  Modern technology provides cannons that outperform the old blackpower cannons in every practical aspect of war, but an artist seeks to make something that is not a better cannon, but something that is an improved blackpowder cannon; even if the best blackpowder cannon falls short (literally) of something designed using modern technology.

 Another example would be the art of fencing -- what kind of idiot would intentionally select sabres for his soldiers to face a regiment of gunmen safely concealed in foxholes at 50 yards?  However, many people devote their lives to mastering the skill of sword fighting.  This is not to save their life in the off-chance they face an adversary armed with only a sword, but because there is artistic expression in sparring with a brother in the same art.

That is to say, we may develop digital signal processing to the point of eliminating analog design as a practical signal processing medium, but we won't eliminate artists who love analog design simply because it is beautiful.

Artistic expression aside, there remain some practical aspects of analog design that cannot entirely disappear.  These relate to the analog aspects of supporting a practical digital circuit:  Power supplies, PCB layout, transmission line termination and D/A,A/D interfaces.  Little of this represents audio-frequency analog signal processing.  

It will be interesting to see what happens to TO-92 packages and through-hole 1/4W resistors as time moves forward.  Maybe this falls into the same vein as artistic expression, although I'm a big fan of SMT even for analog prototype building...but that's another topic altogether.

[StephenGiles]

http://shadowmusic.bdme.co.uk/viewtopic.php?f=11&t=11466

This is an interesting discussion, with a video of an ex Shadow's band playing recently - may be relevant here.

[anotherjim]

What about SPICE? Analogue components each digitally modeled and then the whole virtual circuits affect on a signal is rendered. There is no need to work out a specific DSP algorithm.

From the publicity for Rolands Aria series (dance music tools based on original Roland analogue drum machines & synths), analogue circuit modelling in real time would appear to be the way they did most of it.

Now, if you could buy an enclosure containing the DSP and analogue I/O needed for an effects pedal, have software that renders a schematic into SPICE objects and download to the DSP, would that kill off the traditional build?

[StephenGiles]

What about SPICE? Analogue components each digitally modeled and then the whole virtual circuits affect on a signal is rendered. There is no need to work out a specific DSP algorithm.

From the publicity for Rolands Aria series (dance music tools based on original Roland analogue drum machines & synths), analogue circuit modelling in real time would appear to be the way they did most of it.

Now, if you could buy an enclosure containing the DSP and analogue I/O needed for an effects pedal, have software that renders a schematic into SPICE objects and download to the DSP, would that kill off the traditional build?

Not in my house!!

[karbomusic]

but mostly in our ability to describe what we like to hear in mathematical/computational terms.

Fundamentally, the rendering of some audio transformation by digital means requires that one be able to describe that transformation in terms of an algorithm.  If you can't describe the change, you can't produce or reproduce it.

True but that description is often the same math we'd have used in describing the analog circuit not how we think it sounds. Recreate the conditions in digital based on the math that describes the physics, hope it sounds like it's analog counterpart without using more CPU cycles (or memory) than real time can handle. When jumping into electronics and building pedals after a couple decades being fully immersed in digital and programming etc. I learned one very valuable lesson....

There is no copy/paste or undo/redo in analog! A copy/paste operation may be hours of "making another one". Mistakes are often a good bit of manual labor where in digital it's a simple matter of CTRL+Z. I was much more freewheeling in digital because reversing mistakes was usually very easy; I also learned more quickly because the mistakes didn't hurt so much however what I learn may not last as long since it didn't hurt so much.  I'd give anything to see red squiggles when I place the wrong-valued component on a bread board or perf board instead of a hour or two of analog debugging although spice type simulators try to bridge this gap some.


There is no such thing as truly digital circuits, only analog circuits that have been tortured into behaving like digital circuits.

[amptramp]

One thing that has not been mentioned yet - anything with digital signals or clocks above 9 KHz requires FCC certification whereas straight analog circuitry does not.  You can sell an analog effect without governmental interference.  Not so with digital.

Anything digital requires an ADC input and DAC output before anything else gets done.  A distortion could be done by mapping a PROM so the input is the address and the output is the remapped value.  So you have three digital components operated by a CPU to perform the function of a two-transistor fuzz.  If you are paying the price for mojo transistors, you might actually exceed the price of a digital unit, but mojo is more hype than reality.

Digital effects require at least 16 bits but for certain purposes and more than that may be needed for effects where the granularity of the digitization is important.  Where this happens, you may need to go with external RAM and ROM.  Cases like this occur where you subtract a signal value from another and the value of the remainder is important.  Then it can become a case like low light levels in digital television where the image looks like a contour map whereas an analog signal would not have that problem.

[Transmogrifox]

One thing that has not been mentioned yet - anything with digital signals or clocks above 9 KHz requires FCC certification whereas straight analog circuitry does not. 

This is a good point.  Be careful though, analog effects are not 100% off the hook.  If it conducts out on the cables anything 150 KHz and higher, you risk getting into trouble with the FCC (for example, PWM variable resistors).

Probably most musical effects are relatively safe because they have a low probability of creating enough of an interference with some important RF telecom signal to trigger a complaint to the FCC.  There may be many illegal analog FX out there that only appear to be legal because nobody has ever complained about it causing interference.

[R.G.]

On a deeply technical note, it is common to think that the world is fundamentally analog. That's not strictly true.

The underlying lesson of quantum physics is that the universe is fundamentally divided into a few basic units, and we think it is analog (i.e. fundamentally smoothly varying, not stepped) because there are so many "bits"; we cannot sense the tiny steps. On a very personal level, *WE* are digital. Nerves are digital. They either fire or not. We think the world is analog because there are so very many nerve cells that they present themselves the illusion that all is smoothly varying.

I'm sure this seems like pure sophistry, but it gets up close and personal in electronics at low currents. At low currents, the graininess of charge being quantized into electron-sized lumps does show up as noise. The old-time illustration of this is in low-current vacuum tube gain stages, where the individual electrons slamming into the plate can be heard as "shot noise" when there are few enough of them per unit time to be significant to the current flow. See http://en.wikipedia.org/wiki/Shot_noise

Back at the OP's original question:
We use almost entirely analog stuff in effects primarily as a legacy of the origin of effects and the fact that analog is so very cheap to get started with.

The legacy effect is easy enough to understand - practically every man jack effect builder started off wanting to copy some older effect they thought sounded magic in some way, and that's almost certain to have been analog. Other wise the planet would not have been scoured for two decades now for leftover germanium transistors.

The "analog is cheap" is more subtle. Analog parts are cheap today, although the economy of millions of transistors per currency unit is catching up quickly. Analog is cheap in the fundamental sense of "low barrier to entry".  You can have a working effect with one transistor and a few other parts. The "hump" to get over is small enough that a raw beginner can get started easily without understanding much about electronics at all. Advancement to more complex analog effects is simple and rapid. On top of that, many effects are based on the side effects of some analog circuit's imperfections. That is the sina qua non of some effects - as witness the scouring of the planet for germanium devices again. It's all about the side effects.

While there is some hope that digital stuff is getting more comprehensible, digital signal processing is still a very abstract field. Even using a chip with pre-canned digital effects requires spoon-feeding the chip a lot of very specialized currents and voltages on a plethora of pins. So there is a much bigger skills "hump" to get over to get a working circuit, and an even bigger mental hump of understanding to get over.

The digital stuff keeps getting better in the sense of more reflective of the many-bits smoothness illusion by being both more bits and faster changes in bits, as well as being dramatically cheaper per bit. Digital is doing this by pushing the boundaries of the mechanics to process a bit down into the range where each digital transistor is running into quantum-mechanical limits. Right now, for instance, the issues in making digital circuits involve limitations in the "graininess" of charge and quantum effects. Transistor features are so small that leakage by quantum mechanical tunneling is rising as a limitation on how you could even theoretically make transistors smaller.

But of course, digital doesn't have to get to infinite resolutions to seem analog - it only has to get to resolutions finer than the sensible "steps" our nervous systems can distinguish. That was the idea behind 16 bit encoding for CDs. It's arguable that even if 16 bits isn't good enough, that something fairly close - 24bit? 32 bit? - will be. It doesn't have to be infinitely many bits to be as "analog" as humans can tell. This is in fact the idea behind Apple's "retina" displays, the notion that the density of pixels is at or above the density of rods and cones in the human eye for comparable picture areas.

Why do we use analog? It's cheap, easy, and ... vintage!
Will digital displace analog? Yes, some day. There are more things you can do.
Will all analog disappear? No. There are some jobs so simple that there is no need to go digital. But analog may be very different from what we think of today, as the analog devices that exist may be quite different, and eventually, the old stocks of no-longer-manufactured analog widgets decline. Someday, the last "vintage" germanium devices (for example) will be ferreted out or cannibalized from older junked parts, and no one will make new ones, except perhaps in their 3d electron-force assembler that puts atoms where they're supposed to be one at a time.

When that day comes, the discussions will be on what algorithm to use to make the side effects in the newly-assembled germanium devices be like the last real, old vintage one in the museums.

[Mark Hammer]

If one reads the journals and newspapers of the time, people were awestruck by Thomas Edison's wax cylinder recordings that one could make for a handsome fee at county fairs.  You'd see descriptions that remarked how the recordings were indistinguishable from "the real thing" (sort of the historical precursor to "Is it live or is it Maxell?  What do you think, Ella?").

Of course, to hear such technology now, we would think that there was, in fact, a world of difference in the realism of how the human voice actually sounds, and what could be reproduced by using mechanical means to scratch into hard wax.

Over time, I imagine, someone would have said "Yeah, it's pretty good, but there is just something a little....different about it.  Can't quite put my finger on it, though.".  Others would have said "Yeah, I know what you mean", and together they would have tried very hard to describe, in more specifiable terms,what was "different".  Once they could describe in detail what was different, they would set about figuring out how to reduce that difference as much as they could, with the technology of the time.

And it would keep them happy...for a while...until someone else said "Yeah, it's pretty good, but there is just something a little....different about it.  Can't quite put my finger on it, though.".  And they would go through the same cycle, yet again, using the technology of that era.

Little tiny steps that brought us from wax recordings to the present day state of high-fidelity, in all its 120db S/N, 24-bit, 96khz, flat from here to infinity, .0001% distortion, glory.  Each of those steps relied on engineers responding to what people were now (finally) able to notice, that was...different.  They were probably unable to notice it earlier, likely because they were knocked out by what had seemed like tremendous advances at the time; same way people who had simply never realized that voice could be recorded were stupified by the manner in which wax cylinders were able to reproduce "exactly" all the nuances of sound and voice.

ALL quantification, prediction, and control begins with the human capacity to notice and describe.

[electrip]

Analog means battery voltage is gradually decreasing over time.

Digital means battery is full one moment later empty. 

Digital sucks current.

electrip

[J0K3RX]

Ok Quantum Physics and Stompboxes? Hmmmm? Gettin a little deep up in here lately!

First is banana jacks and now digital vs. analog, entangled with Quantum Physics? 

I say all bets are off when we start teleporting.............................................................

Scientists Announce Successful Teleportation
http://www.gizmag.com/light-into-matter-quantum-teleportation/33906/

Ok, so they have a ways to go before we can enjoy a "lunch break quicky" if ya know what I mean... So far they have only been able to make it work with Apple and Cherry Jolly Ranchers for a distance of 15 miles.. I have to admit, this is a bit disappointing cause I like Grape damn-it!

Imagine what this would do for the "fast food" industry alone!!
Papa Johns Portal Pan Pizza - Delivered Piping Hot to your Fat Face in 30 Seconds or less or it's FREE!!

You may ask yourself, what does this have to do with building your own stompboxes.... Well, nothing of course... Just keepin it real!

[Philippe]

stompboxes aside...how many here prefer the sound of vinyl lps over cd's & mp3's? warmth immediately comes to mind especially when played on a quality turntable + a vintage tube stereo system. personally, I don't have a problem with an occasional record scratch. digital fxs can sound kind of 'fake' in small venues...in a large arena, few listeners can differentiate.