2021, Chapter 9
There is No T in Engineering
You want capital-T truth, join a cult.
No, seriously. Engineering ain’t a place for “One True Path.”
Sure, there are tons of little truths, the basic requirements for making a thing that someone wants to use—does it work, is it reliable, is it affordable, is it safe, stuff like that. Ignore little-t truths like that, and you won’t do well in engineering.
But as far as HOW you design something, or what’s RIGHT way to do it, or the BEST implementation of a particular thing, yeah, that’s wide open. Give an engineering task to 10 engineers and you'll be lucky if there are only 10 different solutions proposed (because, hey, you might want to have a B and C choice to hedge your bets.)
I mean, let’s take the example of a toaster. Dead simple, right? Run AC through a wire, put a simple thermal timer on it, done.
But there's no reason an engineer might not design a wood-fired toaster that needs to be started up with a manual friction fire, or one powered by outside sunlight concentrated by a series of smart mirrors constantly tweaked and controlled with an app on your smartphone, or one that uses old 1000W theater lamps to burn a library of custom-cut gobos onto your bread of choice, while simultaneously using a 12 megapixel camera array to optimize exposure for varying sugar levels in the target.
Of course, these other designs might fail…
…because the wood-burning one takes forever to get lit and it sets off your smoke alarms (even if the marketing pitch is all about the great workout you’ll get while making a friction fire every day)…
…because the one powered by concentrated solar energy might crash and slice your poodle in half, or take out one of your kids…
…because the theater lamp one might be bright enough to blind you if it had any light leakage, and it might be so powerful you end up with toast that’s frozen inside and carbonite on the outside.
The point is: there’s no Holy Book that states how a toaster should be designed.
Now, you may not want the psycho wacky dystopian cyberpunk/steampunk bread-burning devices sketched above, but they are valid engineering solutions aimed at achieving the same goal: toasted bread.
So why don’t we see things like this at the store? And what does that have to do with audio? Where the hell are you going with this, Stoddard?
Aha. Great questions. So let’s talk about the little-t truths of engineering a bit…and then take a left turn to butt dynos and tube amps.
The Little Truths
The reason you don’t see crazy wood-burning or solar-concentrated toasters at Best Buy is the same reason you don’t see speaker power amps with an internal alcohol-powered Stirling-cycle engine and magneto to provide its AC voltage—because they fail one or more of the little-t truths.
(I mean, yeah, it would be a hell of a thing to see such an amp in operation, but the heat from the blazing alcohol fire and the noise of the engine and magneto make it, well, not ideal from a listening standpoint.)
Many of these little-t truths can be examined by answering the following questions:
- Does it work?
- Does it perform?
b. Better than alternatives?
- Is it safe?
- Is it reliable?
- Is it easy to use?
- Is it affordable?
b. Or worth the price?
- Does it meet a need?
b. Or answer a desire?
c. Or bring you joy?
I went ahead and put these in rough order, most important to least important. Or maybe I have it backwards. Or maybe it’s like a story anthology, where the best stuff is always first and last.
So where did the imaginary toaster examples fail? Well, assuming the engineer got them to work (not always a good assumption), 2b spikes all of them:
does it perform better than the alternatives. Er, um, no. 3 is also a problem for all three examples. 4 will definitely trip up the solar one—what if it’s cloudy? 5 is gonna take out the wood-fired one. And all may be taken out by 6.
Same with the Stirling-cycle/magneto audio power amp. 2, 3, 4, 5, and 6 are all “um, no.” It might bring you joy to see it operating, but that probably isn’t enough to save it.
Let’s dig into these little-t questions a bit deeper:
- Does it work? I’m not kidding. Lots of designs don’t work. Engineers design things all the time that don’t work. Or at least I do. Hell, the other day I had to throw away a run of prototype boards because they simply would never, ever work. Because physics. Because I wasn’t paying attention. It hurt even more that those boards were my new business card (I thought, hey, wouldn’t it be cool to do a business card you could build into a small balanced headphone amp…and ended up boning the power supply so bad that it would never actually work. This is sometimes where you end up when you’re trying to be elegantly simple.) Consider the embarrassment of passing out your smarty-pants business cards that don’t work. So, yes. “The design must work,” is one of engineering’s small-t truths.
- Does it perform? Whether we’re talking power output from an audio amplifier, 0-60 times for cars, or Time To Toast for a toaster, there are performance metrics you need to meet. A 0.0005W power amplifier is not very useful, same as a car that does 0-60 in 40 seconds or a toaster that takes 2 hours to get lit. “The design must perform at least adequately for its expected task”is another small-t truth.
- 2b. Better than alternatives? Wait a sec. There’s a corollary here. In any established market, you’re always going up against alternatives. Does your product perform better than those, in at least one metric? And, to be totally clear, there is never any single metric, no most important number. People buy Porsche Boxters over mid-engine Corvettes, even if they are slower 0-60 and cost more. Others only buy electric. Some will buy just because it looks pretty. These are all metrics, and “better” is in the eye of the walletholder. To succeed, the design should be better than the alternatives in some important metrics, or have other redeeming features.
- Is it safe? A product that could fry children and pets like the solar-concentrator toaster, or something that might burn down the house, like an alcohol-burning Stirling-cycle audio power amp, just ain’t feasible. Want to give yourself a real scare? Look up “Electric Tablecloth.” Yes, that was an actual product. For use where people will spill things. Probably salty and acidic things. Yeah. Real-world safety is usually a little less, ah, insane than that, but expecting a product to try to protect the owner from common mistakes shouldn’t be shocking. “The design must be safe,” is an important small-t truth.
- Is it reliable? When you use something, you expect it to work. Simple as that. And reliability, at least with modern electronics, is amazingly good. Not many people know the masic reliability equation—that designing something with two parts of 90% reliability gives you a total 81% reliability. The more parts, the less reliable. And now you know why keeping something simple is a good idea. Considering the hundreds of components there are in a modern electronic product, their reliability is amazingly good. And yeah, I know, I know, there’s differential equations and matrices for real probabilistic systems, but let’s not scare the kids. “The design must be reliable” is another small-t truth.
- Is it easy to use? Something that seems to be forgotten way too much in this era of screens and menus. Or even if it has only a single button…let me tell you how much time I waste trying to turn on our FLIR C5s—push the button, nothing happens, is it on? Press it again, ah crap no its just slow, I turned it off, push it again, it doesn’t know what to do. And so on. They also turn themselves on randomly and draw down the battery. All the toasters and the power amp utterly fail at easy-to-use (well, maybe not the 1000W theater light one, but that’ll depend on how lightproof they can make it, and how they manage the gobos.) Another small-t truth: “The design must be easy to use.”
- Is it affordable? All of the above means nothing if the product costs a mint. I recently saw a self-leveling pool table. As in, it has servos on all 4 legs and it dynamically repositions itself in real time to ensure a level playing surface. It’s a helluva trick. It also costs $300,000. Considering that it’s really intended for superyachts, maybe that’s not such a crazy price (see the corollary below). But, yeah. “The design must be affordable,”is another small-t truth.
- 6b. Or worth the price? There’s always a huge caveat when it comes to price: “affordable” means a lot of different things to different people, and even the same person may call something “affordable” one day and “crazy overpriced” another. You ever buy a $6 water at an amusement park or a $17 orange juice in a hotel? Yeah, there you go. Or maybe that $300K pool table. Because in a $50MM superyacht, that’s getting towards rounding error.
- Does it meet a need? You can have a working, well-performing, safe, reliable, easy to use, and affordable design that is completely worthless. I mean, yes, people have made really massively stupid things like USB pet rocks, drone umbrellas, and walking sleeping bags, but they didn’t light the world on fire, did they? “The design must meet a need,”is another small-t truth, but it has more than one corollary. Because “need” can be very, very subjective.
- 7b. Or answer a desire? You may not need a tube amp for your headphones, but you may want a tube amp for your headphones, and you don’t care there are probably better ways to amplify sound, and better ways to spend your money, and that’s that. Hell, I bought a nixie tube clock. That’s so beyond “need” that it’s silly. But I wanted it. So I got it.
- 7c. Or bring you joy? Or maybe the nixie clock was more this last corollary. I really liked the way it looked. It was cool. It brought me joy. It’s just like someone who has to have a crazy 10-tube OTL/OCL amplifier on their desk, even if it’s huge and hot and more than a little impractical.
Let me be clear: I don’t think you should be choosing products based on this list of little. I have a horrible vision of someone turning this into a rating spreadsheet in order to tick boxes next to a comparison matrix. This is just a guide to help sort out the better engineering approaches.
Hell, and the whole thing is for naught if 7c hits you really hard.
Butt Dynos versus PDRs
Here’s the thing: it’s wayyyyy too easy to over-analyze. Especially these days.
Take cars. In the old days, car guys often referred to the “butt dyno.” As in, how fast the car feels when you’re getting on it. It was usually used as an estimate of the difference in performance when doing tweaks or mods. As in:
“Yeah, butt dyno says that advancing the timing was a good thing. More punch!”
“That new headers aint impressing my butt dyno, the car seems slower off the line.”
“Holy hell, that blower musta be 150 more horses on the butt dyno…until the engine came apart!”
Today, you press a button on the dash to enter Performance Data Recorder mode, pull the paddles to get neutral, mash the throttle to the floor (because you can’t overrev with the rev limiter), and dump the paddles for a 2.7-2.9 second as-measured 0-60 time. Put a new exhaust on the car, and you know instantly if it’s better or worse (even if it feels the opposite.) Go crazier with turbos and engine work, and you get the same instant feedback, glowing on an HD screen on the dash.
And yet…what’s more
interesting?
What’s more
exciting?
What’s…
fun?
To me, it’s kinda sad we’ve distilled the old fun of speculating about what tweaks and tricks and mods made a car faster into a single cold number, a single glowing unassailable Truth.
So. Please. Don’t turn my small-t truths into a sacred writ or Holy Analysis.
No Diks Allowed
What’s nice about audio—especially in the headphone realm—is that there’s a large variety of engineering solutions. There’s no capital T. No diktats.
I mean, in transducers alone, you have dynamic, planar, isodynamic planar, electrostatic, ribbon, balanced armature, multiple-transducer, over-ear, on-ear, in-ear, powered, unpowered, open, closed, high impedance, low impedance…we aren’t quite at the woodburning toaster yet, but there’s no doubt that a 13-driver IEM is a very different engineering solution than an open-baffle ribbon!
Amps? In the headphone realm, low power gives you freedom to explore all sorts of topologies that are, at best, insanely wasteful in the speaker world. Pure Class A, single-ended MOSFET, no gain, low gain, negative gain, multistage op-amps with nested feedback, tube with transformer output, tube hybrid, OTL tube, classic Lin/Blameless voltage feedback, current feedback discrete, balanced discrete, differential, no overall feedback…it’s not a huge deal to throw away 20x the output power in heat, so you can really experiment with some wacky stuff.
Speakers and speaker amps? Sure, there’s a bit less variety, but that’s because the requirements of running speakers are a lot steeper, and speakers are, well, more visible. They’re furniture. Aesthetics get much more important. And speaker amplifiers need to run a lot of power without cooking the people in the room (if Vidar was as inefficient as the original Asgard, it would dissipate 3500W at idle!)
Again, you gotta admit that a pair of powered JBL 305 monitors is a radically different engineering approach than an MBL omnidirectional planar. And a Pass Class A design is about as far as you can get from a BAT integrated, which is on another planet when compared to a paperback-sized Class D amp that probably puts out more power than either. Plenty of variety, no capital-T truth.
Hmm. Funny, no Class D headphone amps.
Maybe that says something. Maybe not. All I know is that it would be really sad if speaker amps all became Class D. Same as it would be sad if they were all Class AB. Variety is good!
Ideally, this is the way it will continue—many different engineering approaches, many different products, many different companies that provide them. I’ve said it before, but this is really the only way we’re going to have a vibrant, interesting, and fun audio market.
Making Our Own truths
No. That’s not a typo. Small-t truths is what I’m shooting for.
Here’s the thing: we’re going to continue down our own paths, sometimes getting lost, going in circles, ending up where we were, and figuring out a new way forward.
Because there is progress. Remember what I said about reliability? There were once TV repairmen. Who came to your house. To fix an expensive and not-super-reliable device. There were once tube testers in supermarkets. We once bought and threw away 9V batteries for transistor radios. Hell, just a few years ago, the idea of a practical electric car was crazy.
Audio is changing, too. Our ongoing efforts to make our entire product line better is one thing, but we’re also experimenting with new stuff (maybe Folkvangr is our solar-powered toaster), going back to old stuff, and designing to the new realities, like the increasing unavailability of tubes.
Hence this chapter. Engineering is top of mind. I’ve been doing a lot of it…and a lot of it for things that are more exploratory, more ambitious than ever before.
Remember how I quipped that if you give an 10 engineers a design task, you’ll be lucky if you end up with only 10 solutions? Many projects I’ve been working on try 2 or 3 approaches…some have up to 6!
(Yes, that’s like 6 different prototypes for some things, all trying different ways of doing the same thing. Not because we’re trying to make the equivalent of friction-fire toasters, but because we’re exploring better ways of doing things.)
Now, it’s not always easy.
One project in particular has been super interesting, resulting in me having to go back to some very basic stuff (in the discrete world) to see how performance is affected by some extremely subtle and esoteric changes.
Which then led to challenges with compensation.
Which then led to tradeoffs relating to bias (and therefore how hot the product ran).
Which made the “better” candidate run too warm for the intended size.
Which, after 6 prototypes, ended up back almost where I started.
But…note the “almost.” The long design detour I took taught me a lot—I became more confident in the solution we’ve already been using, and I also found some subtle tweaks to make it better.
More importantly, though, is the “better” approach that wouldn’t fit in the chassis size I wanted to use (or, likely, within the budget) is maybe very interesting for the next size up. In a larger chassis, I can dissipate more heat. And I can add the stacked rails it needs to avoid the loss in power output. And I can play with some additional ideas that are out of the budget.
So this “better” idea isn’t dead yet. It just may not work in its intended application.
And that’s one of the primary reasons for this chapter. Engineering has been top of mind, because I’ve been doing a lot of it. The 6-rounds-of-proto-to-get-back-to-where-you-started project is only one of them. I’ve also been looking at:
- What gets us a leap ahead on sound quality? We can tweak the same basic ideas forever—Continuity™, Coherence™, Nexus™, etc—hell, lots of companies have only one topology, or none (just using op-amps). But it’s way more interesting to explore new things, or sometimes very old things. Trying new devices—there are whole crops of new discretes out there, some very interesting. Increasing the inherent linearity of a topology through distortion cancellation, rather than feedback or error correction. Melding the best of old and new with modern oversight and topologies that were never feasible without it.
- What to do about the ongoing tube shortage. Sure, we can sell Freya+ without tubes, and bring Freya S back, but what happens if the tube shortage gets worse? We’ve already run out of most NOS tubes. New production seems limited. How do we design for a future where even new production tubes may be scarce?
- How to keep everything we make ahead. As in, how do we continue to increase performance and reliability, without raising the price. As we look at increasing prices for steel, aluminum, and components, do we raise prices…or do we find a way to avoid it? And no, I’m not talking decontenting. I’m talking smarter design that results in a better product in a better chassis with even more features. Can we do it? We’ll see.
And that’s where we bring it all back together. All this exploration is necessary because there is no capital-T truth in engineering. If there was, we could all fall in line and build The One True and Eternal Amplifier and be done with it.
But there’s no Single Right Path or One Sacred Way or Absolute Perfect Approach...there is simply no capital-t Truth.
Now, if I wanted to be reductionist and inflammatory, I could say that audio amplifiers are 90% thermal management and 9% compensation, with the final 1% being your topological choice (including all your crazy cascode/complementary feedback tricks)--and 90% of that 1% is whether you're going voltage feedback (where the challenge is slew) or current feedback (where the challenge is low loop gain and PSRR).
But that’s belittling all the amazing things you can do in engineering. Defeatist thinking like that doesn’t get you ahead. There are literally infinite ways to design an audio product, and playing with a whole bunch of neat ideas (to find the really amazing ones) isn’t work.
It’s
fun. (At least for me.)
Back to it…
