Hephaestus Audio

Let’s say you are asked to evaluate two computers in terms of user friendliness.  If one is an Apple and the other is a PC, then it’s obvious you need to do something to hide this information from the user.  There’s too much of a preconceived notion that the Apple is more user-friendly to allow for a fair comparison.  No problem, we’ll make it a blind test so the user doesn’t know which is which.  In fact we’ll go one step further and make it a double-blind test, so that the individuals administering the test don’t even know which is which.  So far so good.

Now for the next part we ask the user to shift back and forth from one computer to the other every few seconds.  While they are doing this, the users are asked to make note of any difference in user-friendliness between the two computers.  After the trials are conducted, the users sum up their computer user-friendliness feedback.  After careful statistical analysis the results are: There is no significant difference in user-friendliness between the two computers.  Hooray for the scientific method!

The two computers are then given to somebody to use as he pleases over the next couple of weeks.  He scratches his head as he quickly realizes that one of the machines is more user-friendly than the other.  What’s going on here?  Well, first off everybody knows the way this test was conducted is silly.  How are you supposed to notice subtle differences when you’re jumping from one machine to the other every few seconds?  I should clarify that statement: Everybody outside of the audio industry realizes that this test is silly.  However, in the audio industry it is considered the most somber science to conduct a double-blind (ABX) test that flips back and forth between two units under test every few seconds.

Physicist Richard Feynman had a term for this, he called it “Cargo Cult Science“.  It is the sort of science you do when you appear to go though all the right motions, but are missing the big picture.  You are doing everything you are “supposed to”, but you’re not getting useful results.  For example, I read that Swedish Radio conducted a “double-blind, triple-stimulus, hidden-reference” test over two years with 60 experts to test the transparency of low bit-rate audio codecs.  The result?  No statistically significant difference.  However, another listener was able to quickly identify a 1.5 kHz tone in the processed sample under normal listening conditions (i.e. not double-blind).

What can be done to remedy this situation?  Standard audio ABX tests are of little value except for detecting gross differences.  Conversely, going by “whatever sounds the best” is too much like reading tea leaves.  As I’ve mentioned in a previous post, you really need to be able to live with a piece of audio equipment for a while to detect subtle differences, but that sort of test is very difficult to control.  Ah, control…that makes me think of a control group.  What if there is a way to conduct these long term listening trials while maintaining a control group that is given a “placebo”?  Follow me for a moment and I’ll show how this might work.

The latest, whiz-bang, perfect-sound-forever amplifier is finally out!  The objective measurements are phenomenal and the designer is a Ph.D. who works on quantum chaos in his spare time.  The previous generation of this amplifier is very, very good, but this new one makes it seem like something an orangutan designed, the marketing department claims.  Demo samples are distributed, and true to the manufacturer’s word, the new amplifier is so good that every single person loves it and they all burn their old amplifiers so as not to be tempted to ever again listen to something so vile.

That’s the typical scenario (well, sort-of), but what if instead the manufacturer of the new amplifier did the following: Twenty demo amplifiers are sent to reviewers and early adopters, all eager to hear and comment on the new sound.  They are allowed to spend several weeks with the new amplifiers, enough to detect very subtle differences, and then they are asked their opinion relative to competitor’s amplifiers and even to the manufacturer’s previous generation amplifier.  Good so far, but now for the coup de grace: 50% of the demo amplifiers were in fact the previous generation amplifier inside of the new chassis!  That’s right folks, half of you are part of the control group and have been administered the placebo.

How many manufacturers of audiophile equipment are willing to do this?  Can you imagine the change in tone of audio reviews if this practice became common?  If an amplifier manufacturer spends time, spends money, and adds complexity in order to create a next generation amplifier, then shouldn’t it be able to pass this test?  If improvements in THD+N, or any other objective test you can think of, do not pass this long-term double-blind AB test with placebo, then how can you justify manufacturing it and selling it to your customers?

I’m an engineer, so of course I love the technical challenge of optimizing a given parameter as much as any other engineer.  However, an important difference between research and product development is that in product development any change you make should result in a measurable improvement for the customer (cost, reliability, usability, sound quality, etc.).  Unfortunately, it seems like the only people benefiting from efforts to improve THD+N are those in the marketing department.  The research simply isn’t there to support the practice.

Earl Geddes sums it up very nicely with the following quote from his Dagogo interview:

“My position is that if some manufacturer claims an improvement in some sonic property, subtle or not, then it is their obligation to measure this (even if they have to figure out how to do that) and show in a statistically significant way that it makes an audible difference.”

Recently I tried my hand at perfumery.  As with most disciplines there is so, so much more to it than you would ever think.  Interestingly enough, it reminds me very much of audio amplifier design (funny how many things tend to do that for me…).  Here is a brief summary of my first attempt:

• Took a look at the original Eu de Cologne recipe as a starting point (or at least the closest thing I could find to it on the internet).
• Ordered up the necessary essential oils: lemon, orange, tangerine, bergamot, lime, grapefruit, neroli, lavender, rosemary, thyme, petitgrain, jasmine.
• Added amounts that seemed “about right” to a base of Grey Goose vodka.  Let it rest for a while to let all the scents come together nicely.

The result?  Nice, but not wonderful.  Why?  What did I do that was so different than anybody else?  What are the variables I have control over?  Thinking this over for a bit I came to the following realizations:

• Simplicity is best!  Whether creating a new culinary experience, a cologne, or an amplifier you should only add the essential.
  • More is not always better, even if it’s more good stuff.  You must have a specific purpose in mind to add even one more thing.
• Quality is crucial!  All essential oils are not created equal.  I later purchased some very good (and expensive) petitgrain and jasmine.
  • There is no substitute for quality when trying to create the best.  It is a necessary, but not sufficient, requirement.
• Iteration is necessary!  You can keep it simple and use good quality ingredients, but you still need to try and try again.
  • This is how we learn and how true genius comes about (1% inspiration and 99% perspiration…)

Already you probably see how this applies to the design of audio electronics – specifically audio amplifiers in my case.  Lessons to take from this for amplifier design:

• “Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.” – Antoine de Saint Exupery.
  • Do not add more and more stuff, even if high quality, instead take away everything you can while maintaining the fundamental functionality.
• Good components are crucial.  Do not waste your efforts by pinching pennies in this area – you will ultimately pay even more in service costs and reputation.
  • For example, a “thick film” resistor in your signal chain can introduce a surprising amount of distortion, yet only saves you less than a penny!
• Question: “How do I get to Carnegie Hall?”  Answer: “Practice, practice, practice.”
  • Prototype, prototype, prototype.  That amazing zone where theory meets reality.  Do lots of it – it’s fun and results in beautiful audio amplifiers!

As a parting thought, here is another area of perfumery with surprising parallels to audio electronics.  There are three distinct categories of perfume scents, based on the time frame in which they assert themselves: The top notes, the middle or “heart” notes and finally the base notes.

Top Notes

These are the scents that form the initial impression of the perfume.  They dissipate quickly, but are important for the same reason all initial impressions are important.  Audio equipment also has “top notes”.  This is what you first noticed when you first heard the equipment, but which faded from your conscious notice after the break in period.  Now only your friends notice these notes when hearing your system for the first time.  Although if you haven’t listened to the equipment for a while you may be reminded.  These are the traits that it is easy for your brain to adapt to.

Heart Notes

These are the primary scents showcased by the perfume.  Once the ephemeral top notes have dissipated this is mostly what you smell.  With an audio system this is what you are left with after the break-in period.  This is the true nature of the gear – it is for these notes that you love it or hate it.  These are the traits of your audio equipment that are not as readily “filtered out” by your brain.  Maybe that’s a good thing and maybe it isn’t – it all depends on how much you enjoy that particular trait!  It is for these notes that a brand of audio equipment becomes famous or infamous.

Base Notes

The base notes are there to anchor the scent, and they do so in a well designed perfume, however they can also be the unpleasant scents left behind once the top and heart notes have dissipated.  Your audio equipment, which one is it?  Does your listening session start off great, only to lead to disinterest after a while?  “Listener fatigue” perhaps?  Or is your audio equipment the kind that invites you in and doesn’t let go until you really must be moving on to other things…  After everything else has dissipated, this is why you either keep the equipment, or put it up for sale.

A friend of mine was visiting last week and asked to see my VK-1 Dipole Loudspeaker.  He has worked in the audio industry and has done live sound on many occasions, but he doesn’t tend to sit down and listen much to high quality two-channel audio.

I put on a track of some big band music from the BBC’s Big Band Orchestra.  After a few moments of listening quietly my friend motioned toward the loudspeakers and electronics and remarked, “I forgot that these things can sound real!”

Yes folks, they can sound very real.  I think many people that have never heard really good two-channel playback, and even those that have heard it but have simply forgotten over the years, are amazed at just how realistic it can sound.

Audio recording and its subsequent playback has come a long, long way since Edison’s cylinder recordings.  But even more exciting is the fact that it still has a long, long way to go.  As much as I love the sound of my VK-1 system, I know it will sound to those of the next century much as Edison’s cylinders sound to us today!

• Is Chateau Margaux better than a 95 point California wine?
• Is Prosciutto di Parma better than the ordinary ham I get at the deli?
• Is a Stradivarius really worth it compared to a reasonably priced violin?
• Is an audiophile amplifier different than an ordinary audio amplifier?

This type of question circulates endlessly in both audiophile and non-audiophile circles.  Many folks contend that there is no difference between competently designed audio products, especially adherents to the all-powerful ABX test.  Here’s what I think it comes down to:

• There are folks that can’t really hear a difference and don’t care
• There are folks that can’t really hear a difference, but like new toys
• There are folks that can’t really hear a difference and say you can’t either
• There are folks that can really hear a difference, whether you like it or not

It’s the last category that annoys people the most.  Nobody likes to think that they have inferior perception compared to another.  “I can’t perceive a difference, therefore no difference exists.”  “He says it sounds better, therefore he’s full of it.”  Yes, there are plenty of times that somebody says something sounds better, but are just fooling themselves.  However, there are also times when it’s true.

A physics professor once told me that most of the significant discoveries in physics have already been made.  I replied that I thought we were standing on the verge of a precipice with an unbelievable new world of science before us.  Which one of us do you think history will show as right?  Along these lines, I came across something a highly regarded amplifier designer wrote:

• “The time of making great strides in amplifier sound is over.”

If you are an audiophile/audio designer like me, then you probably find this statement a little disturbing, especially coming from an authority on the subject.  However, worry is unnecessary because such claims are made all the time – some examples:

• “With over fifteen types of foreign cars already on sale here, the Japanese auto industry isn’t likely to carve out a big share of the market for itself.” – Business Week, August 2, 1968
• “Everything that can be invented has been invented.” – Charles H. Duell, US Patent Office 1899.
• “It will be dead by June.” – Variety Magazine regarding the ‘Rock and Roll’ fad in 1955.

What are some ways future amplifiers may be significantly better?  As Yogi Berra said, “Prediction is very hard, especially about the future.”, but for a bit of fun speculation, here are a couple of ideas I have: Adaptive Operation and True Accuracy.

• Adaptive Operation: The amplifier changes its mode of operation to make errors that are less offensive to a given listener.  All audio products have errors and always will, but to a given individual some errors are more tolerable than others.  Do you prefer good imaging?  High resolution?  Warmth?  Musicality?  The amplifier can operate according to YOU.  Is it a quiet evening, or a high-energy party?  The amplifier can adapt its operation accordingly.
• True Accuracy: For audio products “accuracy” does not mean what the Audio Precision measures, it means what YOUR hearing measures.  Accuracy means presenting the original material as the artist intended it to be perceived by the audience.  With recorded material YOU are the audience.  Recorded audio breaks down spatial and temporal barriers to connect you with the artist.  This has absolutely nothing to do with metrics such as THD+N.

As a parting thought, I like what Alan Kay said regarding the future: “The best way to predict the future is to invent it.”  I’m not interested in debating if there can be better sound or not.  I’m interested in ensuring that there will be better sound.

Don’t let the title scare you off: “equivalent input noise” is just another way of measuring how noisy an amplifier is.  The nice feature in this case is that it takes the gain out of the equation, so it doesn’t matter if the gain of the amplifier is 20dB, 26dB, 32dB, or 38dB – you can still compare apples to apples.

Perusing the datasheet of a class-D amplifier from one of the most highly regarded class-D manufacturers in the world (no, not me…yet), I came across the following noise specification:

5nV/sqrt(Hz) equivalent input noise

Well, is this quiet, noisy or what?  As it turns out this is extremely quiet, as quiet as a 1.5kΩ resistor just sitting there on its own in fact!  What is even more amazing is that this amplifier features an input impedance of 100kΩ!  In other words right at the input of the amplifier is something that should be generating about 41nV/sqrt(Hz) of noise, yet the amplifier only features 5nV/sqrt(Hz) of noise!?

How is this possible you may ask?  Well, it simply isn’t.  To the best of my knowledge people have not yet determined how to cancel out random noise.  Reading the datasheet a bit more it seems that the amplifier features a “minimal path voltage mode” input with an impedance of 660O.  That must be it.  It must be this “minimal path voltage mode” input that is used for purposes of the noise measurement, but is this the input typically used by customers?

This does strike me as a bit of obfuscation.  The other parameters of the amplifier are very good, so why give your customers this useless noise data?  How about simply providing the dynamic range or signal to noise ratio for the 100kΩ input – the one the customer will most likely use, or at the very least provide understandable noise data for all of the inputs?

For those interested, here’s how you calculate the noise of a resistor:

$$v_{n}=(4kTR)^{1/2}$$

k = Boltzmann’s constant
T = temperature in Kelvins
R = resistance of the resistor

Audio components are often touted by manufacturers as “one size fits all”, but most audiophiles know this simply isn’t true.  What works for you in your system may not work for me in my system.  This begs the question: What exactly makes a piece of gear work for you?

What it all comes down to is exactly what you are looking for in your musical reproduction.  Extreme accuracy?  Recreating the ambiance?  The spirit of the original performance?  Relaxation?  This list could go on indefinitely and may vary day-to-day for many listeners.  However, chances are though that there are one or two things that are absolutely crucial for you as a listener to appreciate a given audio system.

It helps to start with a concrete example, so here are a few things that may be crucial to you as a listener and the type of loudspeaker that may fit the bill.  This exercise can be done with any audio gear, but loudspeakers seem like the most intuitively easy to grasp for purposes of illustration.

“Accurate and stable imaging is most critical to me.”

Full range drivers or possibly coaxial/coincident drivers may be the best choice.  This allows for the best integration of the source signal, often at the expense of frequency response and dynamic range.

“Maximizing the dynamic range is most critical to me.”

A large format multi-way system or possibly a horn loaded system may be the best choice.  This allows for the most dynamic range over the audio bandwidth, often at the expense of poor imaging and coloration.

“Recreating the performance space is most critical to me.”

Minimizing early reflections/stored energy is essential to avoid masking spatial cues.  An open baffle system may be the best choice, although the price is poor low frequency efficiency and the need for large woofers.

As for myself, I like my music to sound, well, musical!  For me this means using drivers without nasty breakup modes – even if the modes fall outside of the bandwidth I intend to use the driver in.  Unfortunately, this puts most drivers with with rigid cones out of the running: Kevlar, carbon fiber, aerogel, etc.  Some metals might be okay, such as the highly regarded Jordan JX92S or an aluminum ribbon tweeter.

Frequency response aside, I think drivers should sound good on their own without any correction networks, etc.  Here’s a little secret about driver breakup modes: They represent a lack of control on the part of the driver, therefore it takes very little to excite these modes – even if notch filters and steep crossover filters are used!  You can’t control the uncontrollable, you can only try to minimize excitation of these modes.

A driver is simply not going to be more “accurate” if your brain has to sporadically allocate resources to “hear around” nasty twin breakup peaks at 3kHz and 5kHz.  My opinion is to do yourself and your brain a favor and use musical drivers!

Signal to Noise Ratio (SNR), or Dynamic Range, is a measure of the maximum resolution of a piece of audio gear.  The larger the number, the greater the possible range from very, very quiet sounds to very, very loud sounds.  It is easy to calculate, for example:

Noise level with no input signal = 70uVrms
Signal level at full output = 1000Wrms into 4Ω → 63.25Vrms
SNR = 20 log (63.25Vrms / 70uVrms) = 119dB

Pretty simple, huh?

Now sometimes the Signal to Noise Ratio may mention that it is “A-Weighted“.  This is an attempt to adjust the value based (very roughly) on the way we hear – i.e. the “Fletcher-Munson” curves.  The end result is a higher Signal to Noise Ratio than if no weighting were used.  Is this simply cheating to improve the on-paper specifications?  Probably, but at least it has some basis in psychoacoustics.

Speaking of cheating, I came across an amplifier with the following specifications:

Noise level with no input signal = 90uVrms (A-Weighted)
Signal level at full output = 1000Wrms into 4Ω → 63.25Vrms
SNR (according to specifications) = 120dB (A-Weighted)
SNR (according to calculations) = 20 log (63.25Vrms / 90uVrms) = 117dB (A-Weighted)

What’s going on here?  There’s a 3dB discrepancy between the spec sheet SNR and the calculated SNR!  The trick in this case is that the manufacturer is using the peak output power (2000Wpeak) of the amplifier instead of the RMS output power (1000Wrms).  It is then compared against the RMS noise level – i.e. it’s not comparing apples to apples!  Here’s the calculation:

Noise level with no input signal = 90uVrms (A-Weighted)
Signal level at full output = 1000Wrms into 4Ω → 2000Wpeak into 4Ω → 89.44Vpeak
SNR = 20 log (89.44Vpeak / 70uVrms) = 120dB (A-Weighted)

There it is now!  So if you are simply “creative” with you calculations, you too can add 3dB to your amplifier’s SNR.  Is this cheating?  Yes, I believe it is.  However, you are now armed to check if the SNR numbers add up for a prospective piece of audio gear, to help weed out the cheaters.  Good luck!

In order to evaluate two components in an AB test, it is essential that they are carefully level matched.  If the gain differs by only one dB, the louder unit will almost always be perceived as sounding better.  Unfortunately, this phenomena has been exploited by unscrupulous hi-fi dealers to promote overpriced components, as I have personally witnessed:

My friends and I once went to a local hi-fi shop to check out the goodies.  There were some very lovely components we had the opportunity to see and hear, but there was also a one meter length of heavy power cord retailing for about $2,000 (this was over 10 years ago).  The salesman conducting the demos (err…Jim) saved this gem for last – here’s the process he used (I watched very carefully):

• Carefully set volume knob to specific level
• Play track from reference CD to completion
• Turn down volume knob to zero
• Insert special power cord between power strip (!) and amplifier
• Carefully set volume knob to specific level
• Play track from reference CD to completion
• Turn down volume knob to zero

Sure enough the second time the track was played it sounded significantly better!  It would have been truly amazing if I had not been paying really close attention: When Salesman Jim carefully adjusted the volume knob the second time – the time with the special power cord in circuit – he turned it up just a little bit higher!  It wasn’t enough to notice it was louder, but it was enough for all of us to think it sounded much better.

My friends had unfortunately not noticed this scoundrel’s trick.  I’m sure I only did because I knew to look for it.  I quizzed Salesman Jim on exactly what made this power cord work.  At first he tried to dodge the question, but then it went something like this:

Salesman Jim: “Okay, okay it’s a ‘PFM Circuit’.”
Me: “A ‘PFM Circuit’?  I’ve never heard of that – what’s it mean?”
Salesman Jim: [laughing] “Pure Fricking Magic.” [note: slightly edited]
Me: “Okay, ha ha, seriously what is it?”
Salesman Jim: [growing annoyance with this punk] “Hey Bob, what’s the technology of this [brand] power cord?”
Salesman Bob: “Uh…geometry and metallurgy.”
Me: “Geometry and metallurgy?”
Salesman Bob: “Yup, that’s it.”
Me: [sigh] “Okay, well thank you guys for your time.”

So there you have it folks – the secrets to any mysterious audio gizmo: PFM Circuitry and Geometry/Metallurgy – what will they think of next!

Please forgive the bitter tone that follows in this post, but recently I came across an “audiophile” device that I feel may damage the already precarious position of the audiophile industry.  First however a little necessary physics background…

A “Blackbody” is an object that does not reflect electromagnetic energy incident upon it, or allow any to pass through it.  Any emission from it is entirely thermal in nature with no characteristic emission/absorption lines from any element.  This is referred to as “Blackbody Radiation”.  Classical theory describes this spectrum with the Rayleigh-Jeans Law:

$$I(\lambda ,T)=\frac{2\pi ckT}{\lambda ^{4}}$$

$$I(\lambda ,T)$$ is the power per unit area
$$c$$ is the speed of light
$$k$$ is Boltzmann’s constant
$$T$$ is the temperature in Kelvin
$$\lambda$$ is the wavelength

This is in agreement with observation at lower frequencies, however the problem with this result is that at higher frequencies (i.e. as $$\lambda \to 0$$) the power approaches infinity!  This is certainly not in agreement with observation and has been referred to as the “ultraviolet catastrophe”.

A physicist by the name of Max Plank made an amazing assumption that was to prove instrumental not only for resolving the paradox of the Rayleigh-Jeans law, but for all of quantum mechanics.  The assumption is that radiation can only only assume discrete energy values:

$$E_{n}=nhf$$

$$n$$ is the energy level
$$h$$ is Planck’s constant
$$f$$ is the frequency.

Therefore the energy between two adjacent energy states is given by:

$$E=hf$$

The power per unit area spectrum that results from this important assumption agrees with observation and is given by the following expression:

$$I(\lambda ,T)=\frac{2\pi hc^{2}}{\lambda ^{5}(exp(\frac{hc}{\lambda kT})-1)}$$

Now I return to the bitter part of my post.  As I make clear in my Engineering Perspective article, I am not a big fan of overpriced audio interconnects.  I am not saying there isn’t a subjective improvement, but that it may not provide the greatest benefit to cost ratio for the audiophile.

However, at least these overpriced interconnects are functional.  At worst, they get the signal from point A to point B.  Even something as esoteric as a special wooden volume knob that improves your sound at least allows you to turn up the volume!  But a product that claims to dramatically improve your system by acting as an electromagnetic blackbody for your equipment is a complete waste.

There are people in this world with incredible perception that can detect “impossible” subtleties.  The purpose of the audiophile industry is to cater to these individuals – those for whom minute differences matter.  Please try to give these gifted individuals the audio experience they deserve and not just another piece of junk backed up with horrible pseudoscience.

There is much debate about optimum grounding strategies in audio electronics.  This article explores a “Star Ground” versus a “Ground Plane”.

It may come as a surprise that, despite the common use of the star ground approach in audio electronics, the ground plane is superior.

Also included in the article is a list of helpful advice for a successful layout, particularly for high-power switching audio designs.

Enjoy!