The problem with having an amazingly transparent signal chain (although I doubt that valves are the way to achieve this but he seems to know what he's talking about,) is that it pushes all the responsibility onto you the listener, to choose program material that is well engineered and artistically resonant enough to justify all that outlay. It's like having special eye surgery to visit galleries. It doesn't quite add up.
Consider that the gear Miles Davis made all his music through sounded completely fucked by this guy's standards. For some reason, harmonic distortion produced before the 80s is art, and harmonic distortion produced after the 80s is an affront. Its like some kind of analog/digital existential nostalgia. I blame postmodernism.
The more transparent the amplifier, the better the music has to be, or you really are wasting your time. If you just want to build high performance electronic equipment, there are much more interesting areas to work in than simply applying gain to a pair of AC signals while preserving their phase relationship.
But yeah, each to his own I guess. I expect these questions are half the fun if you're into designing your own amplifiers.
The thing is, audiophilia was somewhat justified in the 60s when gear that wasn't "high-fidelity" was of objectively lower quality, and also much more variable between manufacturers and even individual units. When taking your valves to the the tube tester at Radio Shack was a viable business model.
This built an entire sociology of competitive audio listening, like a much nerdier version of the car performance tinkering of the same time. People got very emotionally invested in this kind of quality one-upsmanship.
In the 80s MOSFET amps became viable, cheap and widespread. The baseline quality was now so much better that non-experts stopped caring much - you could buy anything in the hifi shop and get reasonable results. But the culture lives on, lost in the jungle, worshipping at a shrine that almost all the population has abandoned.
TVs and computer graphics are approaching this tipping point from "more money buys you noticeably more quality" to "actually I can't see the difference from the cheap baseline and neither can the people I want to show off to". Notice how 4k hasn't taken off dramatically.
It's funny that you mention TV and computer graphics, because TV and computer sound has not been a case of "buy anything in the hifi shop and get reasonable results". When using the audio chain built into modern motherboards I consistently hear noise from some bus or another polluting the actual audio output. Poorly-designed, poorly-shielded amps still happen and unlike modern video output it's not a digital signal that has error correction built in.
Very true. PCs ship with lowest-common-denominator powered speakers and noisy headphone outs. The average customer and even the gaming customer doesn't seem to care much. People who do get USB output devices or send SPDIF to their existing surround system. Maybe Macbooks are less bad at this?
Macbooks are indeed great at this. They can output 24 bit 48khz IIRC. And I'm pretty sure there's even DSP in there to model the resonances of the case and cancel them out when using the speakers.
And BTW gaming motherboards now commonly have audio interfaces with isolated power rails and stuff.
The real benefits of a macbook is the 192 kHz (on recent models) tosLink optical out integrated into the headphone jack, which is just an awesome feature that I really wish other laptop manufacturers would pick up.
DSPs for resonances/tuning has however been picked up pretty widely (under a wide variety of brandings/quality)
In the 80's I had a Hafler DH-220 amp, one of the first to use discrete MOSFET devices at a reasonable price -- IIRC $700, or $600 if you built it yourself. With a quiet preamp and the then brand-new Sony CDP-101 CD player ($650-ish) you had some amazing sound. Hit pause on the CD and listen to the dead silence, even at high volume levels. Amazing.
These days - spend your money on anything of reasonable price and you'll get excellent sound. They're marketed more on features (i-device integration, network audio, etc) than audio quality, because decent sound quality has become a given.
That Sony CD player may have been quiet, but that was its only grace. When I first heard it I wondered why they had selected such a harsh system for the demo. It wasn't until later that I figured out it was the CDs and player itself that was generating the harshness. Thankfully digital has progressed greatly since those early days.
Before I threw it away a few years ago, I hooked it back up and listened to it. Yeah, it was bad. But good for the time. I figure a $40 Walmart player from today would sound better than it did. Wikipedia says it's because Sony only put one DAC in there to save cost, and routed the left/right channel through it alternately. So one channel was always slightly lagging behind the other.
I still have some of the CDs from the mid-80s. Pink Floyd The Wall sounds like they just dumped the album tracks on it, RIAA curve and all. Dire Straits Brothers in Arms still sounds pretty good - there's a reason it won two Grammys. Bought the MoFi remaster last year of it.
"...audiophilia was somewhat justified in the 60s..."
I agree that amps are not the best arena for innovation any more, and that audiophiles ("audio fools") are often emotionally invested in their particular choices, and covering it up with jargon and BS.
But moving away from amps -- speakers and room characteristics do still make a big difference. I do most listening on a decent system (mid-size Dynaudio speakers) but hearing their larger cousins in a shop with great source material is still a thrill.
I think there is a difference between transparent amplifiers, etc in the recording stage vs transparent amplifiers in the playback stage.
In other words, harmonic distortion is art if it was put there during the recording process and not art if it is put there during the playback process.
But presumably the average home listening gear in 1969 was truly awful too. Does that mean nobody had really appreciated In A Silent Way until decades later, when they were finally able to discern every detail of its noise floor in high resolution?
When you abandon the traditional audiophile silliness of subjective judgements in favor of scientific measurement, you just enter an even deeper mire, and it becomes completely impossible to determine why you're doing what you're doing, or what you're looking for.
The best thing about art, and especially music is that your soul/subconcious/whatever can tell you if its good or not, immediately, for free. Your brain does not need six sigma of certainty to decide if you should currently be weeping with joy.
It's not like having special eye surgery to visit galleries. It's more like creating special lighting in a gallery for viewing art. The difference being you don't have to use the amplifier all the time, for everything. What your complaint about having to select well-engineered and artistic music doesn't spell out is that this material does exist, and I imagine the setup would absolutely be worth it if you care enough. I'm not saying that all the modern audiophile stuff is true, but if you want to sit in a room and hear as close as possible recreations of your favorite symphony conducted by your favorite long-dead composer, why not try to get the signal as clear as possible?
I suppose my point is that transparency for artistic/entertainment purposes was achieved long ago. The fact that we need to involve lab equipment to decide on the superior signal chain is testimony to this.
Now, people seem to be trying to get better fidelity than that of the original recording medium, which is obviously insane. You're listening past the art to examine details of the noise floor of 20th century equipment that weren't even detectable at the time. What on earth is the point of that? I mean it's one thing to do that to a Stockhausen record, but I don't think Miles would be terribly impressed.
There's a law of diminishing returns for this stuff, and after a very small amount of time and money you're going to be better off contemplating the music and not the sound, even if the sound still aint great. You can be sure that's what the composer was doing.
I rarely read this. At which point do we shift the focus to the medium and not the message ? And even worse, when we optimize production toward the medium (3D 60fps movies).
Well, they're supposedly inseperable aren't they. I think that's part of the problem of the digital age, we haven't learned how to cope with what seems like the abstraction of message from medium.
By way of explanation, here is the preferred, "natural" signal chain for the music closest to my heart, UK rave music:
1. Source materials taken from old vinyl and put into a cheap sampler, often with only 12 bit resolution.
2. Recorded through a budget Mackie mixing desk, and bounced to DAT.
3. Pressed to a one-off acetate record (the analog version of a CDR)
4. Played through a cheap DJ mixer at a pirate radio station, and pushed through a rack of cheap compressors/limiters for broadcast.
5. Encoded as a microwave laser and beamed several miles to the actual transmitter, to obscure the location of the studio from the police.
6. Encoded again into FM and transmitted.
7. Picked up on a consumer radio and recorded to consumer cassette tape.
8. (Optional step) Said tape left in a box for 15 years under your older brother's bed.
Apart from step 8, this was largely how people listened to the music at the time. I have since obtained step 2 copies of music I had previously only heard at step 8, and all the emotion was gone. You could tell it had been shorn of context. So, as you can hopefully see, I find ideas of "fidelity" rather difficult.
When I and other young producers try and make records in this style, its a constant question as to how far we should go to imitate the sounds of all those dead media and ghost rituals. To me it feels pretty self defeating, but I still do it to an extent, because I think it sounds better. Sometimes I wonder what the modern day, internet-age equivalent is, and how to make it. I'm probably too old.
I know, I questioned the perception of records and movies since a long time. I was often disappointed when I finally get the disc I heard on the low grade radio in my mother's car, I actually liked the distorded sound better, somehow it made some harmonics pop better (at the expense of clarity).
Same for movies:
New movies : high resolution image, low resolution scenario/direction, low joy.
Old movies : crappy tired VHS image, probably better scenario/direction, high joy.
That's an interesting analogy, because the lighting in art galleries is not at all neutral. Most galleries point a few narrow spotlights at each piece on the wall. The spotlights are warm and incandescent, and their narrow spread illuminates the center of the work more than the edges. This makes the work look like it's emitting light. It has an enhancing effect compared to normal flat interior lighting. So the actual lighting in art galleries is more like the warm distortion of vinyl and naive tube amps, not high accuracy of this amp.
>For some reason, harmonic distortion produced before the 80s is art, and harmonic distortion produced after the 80s is an affront.
You're comparing apples to quantum dots.
Tape recording/mixing didn't add distortion in the same way that a bad amp adds distortiion. There certainly is such a thing as cheap transistor sound, and it's both measurable and audible.
There's also such a thing as a high-end studio equipment sound. It's why a good proportion of professional recording and mixing engineers prefer expensive outboard over plugins - even though some of the plug-ins are older than they are now.
There's also such a thing as amplifier sound. Pretty much anyone can hear the difference between a selection of affordable amps on a straight A/B comparison.
There's also such a thing as hifi wankery and outright scamming, especially in the accessories market and at the very high end.
But most people build/buy systems that help them enjoy the music more. If you can't hear the musical differences between - say - music played on a crappy cassette player from the 1980s and music played on a Nakamichi player from the same period, that's fine.
A lot of people could - including all my friends, in various store demos when I was helping them buy music hardware. Quite a few of them had the same reaction I did. Which was 'Wow, I had no idea.'
>If you just want to build high performance electronic equipment, there are much more interesting areas to work in than simply applying gain to a pair of AC signals while preserving their phase relationship.
But audio is completely fascinating because such tiny, subtle changes have such a big effect on the result, and often no one is exactly sure why.
Digital dither is such a small signal it should be completely inaudible. If you play a dither signal on its own, there's nothing to hear.
But the effects are very, very obvious.
Audio is like that all over. Tiny, supposedly trivial distortions change the quality of sound perception in unpredictable ways.
Some make the music sound more enjoyable, some make it sound worse. That's partly why people take an interest.
Competitive consumption is also a thing, but that doesn't change the fact that a lot up-to-mid-price buyers want something they can afford and enjoy. They're not seriously interested in showing it off to their friends or writing about it in hifi magazines.
Funny you should mention Nakamichi. I had one of their cassette recorders from about 1980, and a Panasonic receiver of the same vintage. I was under the impression that solid-state amplification had progressed to the point where you didn't need to worry about it anymore - until I unplugged my headphones from the receiver and plugged them into the recorder. What a revelation! From that point forward I had the hi-fi bug, although I never went overboard with it.
"Kind of Blue" is one of the most transparent recordings I have ever heard. Doesn't sound completely fucked at all.
Also, digital distortion is not harmonic distortion. That's why you preferably don't want to have it on a recording (unless you have a very specific taste or want your music to glitch or sound robotic or whatever).
Anyway, most sources sound better through a superior playback system than through an inferior one. I have no problem listenings to the Wiper's "Youth of America" through a proper HiFi. There might be some odd exceptions that sound better through a Tivoli mono speaker, but those are far and few between. So I wouldn't let that be my concern while designing the best playback system I can. I definitely wouldn't call it a problem.
What do you mean by "digital distortion"? That term makes me think of high frequency intermodulation artifacts produced by the sample rate of an ADC (aka aliasing). This is not a concern with transistor amplifiers.
This is basically the pinnacle of 1960s audio design. You can tell the author was trained in this era and knows his stuff by his reference to Bob Pease. On the other hand, he seems to regard MOSFET power supply regulation with suspicion.
People in other comments have asked how modern class D amps perform. There are people doing a similar kind of thing with aggressively purist design craftsmanship, e.g. http://www.hypex.nl/technology/ucd.html - producing a similar nice flat Bode plot. The really cheap ones may be suspect and spew EMI everywhere, but if you're not quite so concerned with shaving pennies off the price they're perfectly acceptable.
Maybe I can help translate the author's EE work into "HN speak" in that what the author did with this amp design is similar to how programmers do "turing tarpit" language challenges.
The game rules are usually picked based on how much the player wants to put up with. His spec for low freq 3dB down is weirdly low to me because I don't own drivers that run that low (LOL) although the spec for 1dB down is probably more "audiophile relevant". I suppose as a design goal if a recording has a 5 hz signal I'd want it to be properly amplified for the speakers to turn into heat instead of sound, rather than clipping and making noisy harmonics every 5 or 10 hz up the band, so there is that justification, but, why not input filter instead of just making heat in the speaker crossover?
In this story, the game rules were use as little post-1970 components as possible while maxing out various specs to beyond human hearing. A good analogy for programmers would be writing a quicksort in BF or intercal. You don't do that because you need production code, you do that on a dare or to see if its possible or for the sheer heck of it.
Personally if I needed 50 or so watts I'd use a TI TPA3116 and call it good. There's kits available on ebay and elsewhere, its a stereotypical more than a couple watts less than a hundred watts modern class D amp. Simple design one chip no need for heatsink at lower power out. Does need 24V power and a heatsink for full power out. The chip costs about a buck. But that breaks the rules of the game because those didn't ship till maybe two or three years ago, not 1965.
I think you don't understand the project at all. The point wasn't to play some game by doing something retro, the point was to create an amp that sounds as good as reasonably possible. There are many people who believe that tubes are the best way to do this. I can't say they're wrong; the most realistic sound I've ever heard came from a tube amp in a shop.
I wonder how literal the parent was being. He seems to be walking the line. I think from the parents point of view he 'gets' what the original post was going after. Its just more... busting out the oscilloscope graphs to prove transparency at sub human audible levels while at the same time picking components because you can 'hear' the difference is a contradiction. Basic fundamental point is from a pure amplification of signal without distortion standpoint there are components that can get you there for a lot less effort today.
This is absolutely fantastic, and thoroughly technical -- particularly the SPICE simulations. After becoming equally engrossed in designing a "perfect" microphone capsule, I'm tempted to build one myself.
This was written 12 years ago -- I wonder how the author feels now. Specifically I'm curious if the author is still contented with the equipment. Also, apropos of nothing save for sheer curiosity, I'm interested to know what input media (i.e., specific recordings and formats) and output equipment the author used (or uses) for both testing and pleasure.
I'm actually the author's son, but I can confirm that yes, he's still happy with it. Remarkably, it seems to have resulted in a permanent remission of what had previously seemed like terminal audiophilia (remodelling the house to properly dampen its resonance, that kind of thing). He has since channelled his obsessive tendencies into digital imaging, which is an altogether more profitable endeavour.
> I thought about commercializing it, especially when the lab where I worked announced it was shutting down, but I soon realized that marketing high-end tube amplifiers is not a reliable way to make a living. (I'm paycheck-addicted.)
It is wonderful that Kickstarter has removed much of the risk for people like this, that are obviously incredibly talented and have produced something great, but don't necessarily have the desire/risk-appetite to make it commercially without a proven market.
Speakers still add orders of magnitude more distortion than your amps, much like they did in the 1960s, unless you're using something insane like a servo-controlled Meyer X-10 (with an integrated amp, natch).
Also, more important than amp transparency (which is pretty easy these days) is how the amplifier reacts to the difficult loads many high end speakers present
For decent, modern amps, not really, as long as you're not getting something like the interaction between the output filter of a class D amp and the loudspeaker (as happened with some Crown class D amplifiers and some EAW speakers)
Here is another "make your own incredible tube amplifier" from a Dutch gentleman. I'd be interested to hear from someone who understands this stuff, how it compares.
http://www.mennovanderveen.nl/eng/index.html
As a piece of performance art, they are polar opposites.
One is a craftsmanship project hand built according to some peculiar rules (pre 1970s tech when possible, etc) and methodically designed to optimize performance and then tested to verify that performance.
The $30 amazon mass produced amp ships with a 1.5 amp power supply that was restickered to 2 amps (hope your fire insurance is paid up) and can't provide marketed power with the supply as provided anyway depending on speaker impedance connected and no one has any detailed performance stats so who knows how well it compares in all the parameters that matter. Also it runs about 1/10th the power of the tube amp. Finally it is a class D amp but the chip mfgr went all closed source on making a "special" D that got patented as "T", the "T" designator is merely marketing. Its a relatively early class D amp, I would assume modern class D amps would outperform it.
Note that both amps are antique in their own way, one is a "70s amp" and the other is a "turn of the century amp".
As a piece of art, one is roughly a highly talented semi-pro artist classic inspired painting, and the other is a walmart poster copy of a "Dale Earnhardt on black velvet" memorial mass produced kitsch, should be pretty obvious which is which.
First time I visited Koren's website and read about his "perfect amp" was about 10 years ago (around 2006 - when I was designing and building my "perfect amp"). So this article definitely brings me back to simpler times.
I always found this design waaaay too complex. Some sections seem unnecessary while others don't make sense (but that's just me - and it doesn't help that the schematic is drawn a little convoluted). But I suppose it totally depends on his amp's application (e.g. playing records, studio monitors, live guitar, bass amp, etc), so it really may be the perfect amp ...for his specific application.
My thoughts... the power supply is a little too "verbose" for a tube amp -- just unnecessarily complex, and may actually remove some of the imperfections associated with that crispy "classic tube amp" sound. But I am a strong advocate of a well-designed power supply, so I can't say anything bad about it. But there's too much going on between phase-inverter and power amp stages (output drivers I presume). I would never do that, but to each their own. However, I do like that he used 6550s. Great tubes!
I am wondering why high-end amplifiers are still so expensive. Amplification should be a solved problem by now.
Also, what I miss most in the whole sound chain is that I can't change the volume of the tracks that existed before they were mixed into a single track. For example, making the volume of that trumpet in the background a little louder, etc.
> I am wondering why high-end amplifiers are still so expensive.
I think it's a bit like why Gucci handbags are still so expensive. Because they are high end; there are people who are willing to spend that much, even if what I would call an engineer's assessment "do I get a reasonable bang for buck ratio" clearly answers NO. People by an experience, they buy a feeling. And it may be money well spent for them.
Consider that the gear Miles Davis made all his music through sounded completely fucked by this guy's standards. For some reason, harmonic distortion produced before the 80s is art, and harmonic distortion produced after the 80s is an affront. Its like some kind of analog/digital existential nostalgia. I blame postmodernism.
The more transparent the amplifier, the better the music has to be, or you really are wasting your time. If you just want to build high performance electronic equipment, there are much more interesting areas to work in than simply applying gain to a pair of AC signals while preserving their phase relationship.
But yeah, each to his own I guess. I expect these questions are half the fun if you're into designing your own amplifiers.