TR Forums

I generally just go with "tweak the EQ until it sounds about right". At the end of the day it's all very subjective anyway, and my gear isn't high-end, so it's mostly about compensating for the shortcomings of my speakers/headphones.

Speaking of subjective listening tests (and sorry for the tangent), I did notice some improvements in the punch of the bass and clarity of the highs when I switched from onboard Realtek to a Xonar DSX recently. Not a huge difference, mind you... but noticeable. Onboard has come a long way in the past decade and a half, but IMO when VIA faded from the scene and Realtek became the de facto standard we took a small step backwards. I still have a couple of Asus motherboards from the late '00s that have VIA onboard on them, and they sound better than the more recent Realtek ones... especially when directly driving headphones. Unfortunately they're old enough that they are now relegated to server duty, where the better onboard audio implementation goes to waste.

I wonder if any of those very oddball AOpen P4 tube mobos are still in service?

Captain Ned wrote:

I wonder if any of those very oddball AOpen P4 tube mobos are still in service?

Heh. I wonder how many they sold. And whether long-term use didn't cook the other system components from the heat of the tubes.

Even if you buy into the superiority of tube amplification for small/clean signal applications (I don't), pairing what amounted to a tube pre-amp with an early-'00s Realtek codec was an exercise in absurdity. It was the sort of thing I would've expected from a tech version of The Onion.

If you happen to prefer a touch of even-order harmonic distortion, OK fine... but don't claim it is somehow more "accurate" or "natural".

just brew it! wrote:

If you happen to prefer a touch of even-order harmonic distortion, OK fine... but don't claim it is somehow more "accurate" or "natural".

A claim I would never make. To me, tubes belong in guitar amps due to the even-order distortion. That said, the amplification ideal of "straight wire with gain" just implies sterility and reminds me of just how bad the '70s-era receivers sold during the power/distortion wars actually sounded despite their near-perfect measurements.

Oh, and the only way to get repeatable frequency response curves from headphones is to use dummy heads with instrument-grade mics, but even these won't capture the full interaction of the soundwave with the ear's pinna (the outside bit we all think of when we hear "ear"). That, and they cost a fortune. The headphone/ear interface is just too chaotic and too individualised to predict how anyone will "like" what headphone X sounds like on their head.

Captain Ned wrote:

just brew it! wrote:

If you happen to prefer a touch of even-order harmonic distortion, OK fine... but don't claim it is somehow more "accurate" or "natural".

A claim I would never make. To me, tubes belong in guitar amps due to the even-order distortion. That said, the amplification ideal of "straight wire with gain" just implies sterility and reminds me of just how bad the '70s-era receivers sold during the power/distortion wars actually sounded despite their near-perfect measurements.

...and I wasn't claiming that you made that claim.

However, for reproduction of recorded content, there's something to be said for mathematically precise reproduction in the final power stage. If you need processing to compensate for room/venue acoustics, personal preference, etc., that should be done as a (controlled) pre-processing step.

Captain Ned wrote:

Oh, and the only way to get repeatable frequency response curves from headphones is to use dummy heads with instrument-grade mics, but even these won't capture the full interaction of the soundwave with the ear's pinna (the outside bit we all think of when we hear "ear"). That, and they cost a fortune. The headphone/ear interface is just too chaotic and too individualised to predict how anyone will "like" what headphone X sounds like on their head.

...and this is also why headphone-based "surround" solutions do not work well for everyone. The canned DSP transfer function is necessarily a compromise, and does not accurately model how ambient sounds interact with each individual user's unique pinnae.

just brew it! wrote:

However, for reproduction of recorded content, there's something to be said for mathematically precise reproduction in the final power stage. If you need processing to compensate for room/venue acoustics, personal preference, etc., that should be done as a (controlled) pre-processing step.

I can get behind that, though "mathematically precise reproduction" without any listening tests is what gave us the '70s receiver wasteland.

In the Audiophilia Nervosa press there's a description of your ideal amp. It's an amp that finally lets you hear how crappy the upstream chain sounds.

Sidebar for those not of the vintage of JBI or myself on 1970's receivers (pre-digital, so FM tuner, pre-amp, and power amp all in the same box): The power spec/distortion spec wars were fought primarily in ad copy in the two magazines guaranteed to have widespread readership among males of stereo-buying age throughout the 1970s. Their titles both started with "P". I think the stereo mfgs of the time spent more on photography than the editors of those mags did for the actual "content".

Oi. the 1970's were a weird world.

Captain Ned wrote:

just brew it! wrote:

However, for reproduction of recorded content, there's something to be said for mathematically precise reproduction in the final power stage. If you need processing to compensate for room/venue acoustics, personal preference, etc., that should be done as a (controlled) pre-processing step.

I can get behind that, though "mathematically precise reproduction" without any listening tests is what gave us the '70s receiver wasteland.

Entirely possible that the numbers being measured were not the ones that mattered most. Though I'm sure enough equipment of that vintage still survives that a really determined person with deep pockets and new theories about what needs to be tested could collect fresh data.

Captain Ned wrote:

In the Audiophilia Nervosa press there's a description of your ideal amp. It's an amp that finally lets you hear how crappy the upstream chain sounds.

...and there may have been a bit of this going on as well.

Edit: And it all sounds a bit like tuning GPU driver performance to perform well in benchmarks, doesn't it? Some things never change.

just brew it! wrote:

Entirely possible that the numbers being measured were not the ones that mattered most. Though I'm sure enough equipment of that vintage still survives that a really determined person with deep pockets and new theories about what needs to be tested could collect fresh data.

The general explanation for the '70s distortion-spec wars was that massive amounts of negative feedback could clearly make the stats look better, but that it sounded like merde.

Something the current generation of specs-only "reviewers" must again comprehend.

Insofar as negative feedback is able to nullify non-linearities in the underlying components, it's generally a good thing. In the macro picture, it gives you that "mathematically perfect" performance.

I suspect that the problem was massive negative feedback, combined the race for higher power output, plus sub-optimal circuit design. Lots of negative feedback can give you very nice low THD/IMD numbers for large signals when you're pushing an amp pretty hard but not over-driving it. But at lower (i.e. more realistic) listening levels, crossover distortion in (sloppily biased) class AB output stages was problematic. IOW, it was analogous to "gaming the benchmarks" like I mentioned above.

just brew it! wrote:

IOW, it was analogous to "gaming the benchmarks" like I mentioned above.

Wasn't that the entire premise of the stats wars fought through ad copy in Playboy amd Penthouse some 40 years ago? The numbers had squat-all to do with real sound quality and lived only in the marketing office.

...and the "Class A or die" crowd is an over-reaction to that, I suspect.

With proper biasing, combined with sufficient open loop gain and slew rate, lots of negative feedback should give an optimal (or at least near-optimal) result. At least in theory.

Oh, and circuit stability when driving highly inductive loads is another complicating factor...

If it was simple, we'd have found an optimal solution already.

Sorry for the audio geek digression, folks. But I guess it was inevitable given the topic of this thread.

You up for a stab at Futterman-design tube amps with no output transformers? Gotta bring your best EE to this particular audio insanity.

Yes, OTL tube amps are insane... special kind of craziness there. It's a spectacle I prefer to watch from afar (both because I think the participants are certifiable, and because I couldn't afford to participate even if I wanted to).

just brew it! wrote:

Yes, OTL tube amps are insane... special kind of craziness there. It's a spectacle I prefer to watch from afar (both because I think the participants are certifiable, and because I couldn't afford to participate even if I wanted to).

Avoiding the inevitable shrapnel is another valid reason.

Captain Ned wrote:

just brew it! wrote:

Yes, OTL tube amps are insane... special kind of craziness there. It's a spectacle I prefer to watch from afar (both because I think the participants are certifiable, and because I couldn't afford to participate even if I wanted to).

Avoiding the inevitable shrapnel is another valid reason.

Do they really fail that spectacularly? Or do they normally just pop like a lightbulb? Mind you I've never had anything tube but have always been intrigued by them.

tanker27 wrote:

Do they really fail that spectacularly? Or do they normally just pop like a lightbulb? Mind you I've never had anything tube but have always been intrigued by them.

I have no idea whether they actually explode, but OTL designs apparently tend to push them well beyond their design limits, so it certainly seems possible to me.

We had tube-based TVs and phonographs when I was a kid. Whenever the TV stopped working dad and I would take a trip down to the local Rexall drugstore with a brown paper bag full of tubes, to use the tube tester. Yup, back in the day many drugstores had tube testers set up that you could just walk in and use, and you could even buy replacements for the more common types at the counter.

The tester had a bunch of tube sockets to fit all the various form factors, rotary switches to select the appropriate test parameters (which you had to read from a chart) for the model of tube being tested, and an analog meter. You set up the test, plugged in your tube, and read the results on the meter, which had "bad" and "good" ranges marked on it.

I remember being pretty impressed (and more than a little scared) by the high-voltage rectifier tube in our first color TV. Apparently those things got extremely hot and emitted non-trivial amounts of X-rays during normal operation. There was a special shielded metal compartment / containment vessel for it inside the TV, with some rather dire warnings on it... the message was essentially, "Don't open this compartment while the TV is powered up unless you want to DIE!"

JBI wrote:

Whenever the TV stopped working

But it was a richer, warmer, kind of unreliability!

Glorious wrote:

JBI wrote:

Whenever the TV stopped working

But it was a richer, warmer, kind of unreliability!

It was also a very early lesson for me in how MTBF works. The color TV had (IIRC) a couple dozen or more tubes in it. Even if individual tubes were fairly reliable, having that many tubes, each of which represented a potential single point of failure, translated to a rather dismal MTBF for the TV as a whole.

just brew it! wrote:

We had tube-based TVs and phonographs when I was a kid. Whenever the TV stopped working dad and I would take a trip down to the local Rexall drugstore with a brown paper bag full of tubes, to use the tube tester. Yup, back in the day many drugstores had tube testers set up that you could just walk in and use, and you could even buy replacements for the more common types at the counter.

Same here in VT. Remembrances of single-digit ages toting the bag for Dad as he furiously twisted dials on the tube-tester at Rat Shack.

just brew it! wrote:

We had tube-based TVs and phonographs when I was a kid. Whenever the TV stopped working dad and I would take a trip down to the local Rexall drugstore with a brown paper bag full of tubes, to use the tube tester. Yup, back in the day many drugstores had tube testers set up that you could just walk in and use, and you could even buy replacements for the more common types at the counter.

The tester had a bunch of tube sockets to fit all the various form factors, rotary switches to select the appropriate test parameters (which you had to read from a chart) for the model of tube being tested, and an analog meter. You set up the test, plugged in your tube, and read the results on the meter, which had "bad" and "good" ranges marked on it.

I remember being pretty impressed (and more than a little scared) by the high-voltage rectifier tube in our first color TV. Apparently those things got extremely hot and emitted non-trivial amounts of X-rays during normal operation. There was a special shielded metal compartment / containment vessel for it inside the TV, with some rather dire warnings on it... the message was essentially, "Don't open this compartment while the TV is powered up unless you want to DIE!"

I had totally forgotten about this! Yes I remember my dad and Grandpa doing the same thing for our TVs and Stereos! Except replace Rexall with Walgreen's, Sears, and Radio Shack. So I kinda answered my own question, I dont ever recall those tubes exploding.

tanker27 wrote:

[I don't ever recall those tubes exploding.

It would never happen in a device designed using the specced parameters of the tubes used. OTL tube amps run so far out of the designed operating parameters that anything is possible.

Tubes are high voltage/low current devices. Hooking them to a speaker, which is a low voltage/high current device requires a transformer to swap voltage for current and keep everything within the design specs. Transformers are also known to have bad effects on both extremes of the frequency range (c.f. hysteresis) so some enterprising tube jockeys got the great idea "if we use a bazillion tubes we can cut the amperage per tube to something under the splodey limit and ditch that fargin' transformer" and ran with it. OTL tube amps run the tubes far out of spec and they run hot. Real hot, As in don't need to turn the heat on hot. Since they're all Class A by definition, they pull the full bias current (and heat) from the wall regardless of signal load (yes, I'm approximating, but there's a reason Class A amp designers tend to live in places that have real winters). Cats love OTL tube amps.

Glorious wrote:

But it was a richer, warmer, kind of unreliability!

Ah, so you have read Stereophile.

 I've run tube amps for a long time. I dealt tubes for a while, and was able to put a pretty nice system together, fairly cheaply.

 I like em', a lot. They make digital input easier to take, and do sound wonderful.

Captain Ned wrote:

Glorious wrote:

But it was a richer, warmer, kind of unreliability!

Ah, so you have read Stereophile.

Yeah, when I'm mixing live sound I only use those kind of terms when I don't know what the actual problem is (yet).

just brew it! wrote:

Entirely possible that the numbers being measured were not the ones that mattered most. Though I'm sure enough equipment of that vintage still survives that a really determined person with deep pockets and new theories about what needs to be tested could collect fresh data.

The infamous THD @ 1kHz, perhaps? A fairly easy test for even a cheapky-built solid-state amp to pass, and not terribly relevant to full-spectrum audio reproduction, unless you spend your listening time playing back 1kHz sine tones.

Then there was the fact that "true" complementary P-channel silicon devices didn't start to really go mainstream until sometime well into the 1970s...

ludi wrote:

just brew it! wrote:

Entirely possible that the numbers being measured were not the ones that mattered most. Though I'm sure enough equipment of that vintage still survives that a really determined person with deep pockets and new theories about what needs to be tested could collect fresh data.

The infamous THD @ 1kHz, perhaps? A fairly easy test for even a cheapky-built solid-state amp to pass, and not terribly relevant to full-spectrum audio reproduction, unless you spend your listening time playing back 1kHz sine tones.

Indeed. Lots of negative feedback + gain stages that suffer from slew rate limitations could easily turn "acceptable" crossover distortion at 1 kHz into a train wreck at 10 kHz. Translation for those who haven't studied any EE: just because your frequency response extends out to 20 kHz doesn't mean you're doing it without massively distorting the waveforms. And that's the tip of the iceberg. Don't get me started on the "peak music power" versus "RMS power" boondoggle.

ludi wrote:

Then there was the fact that "true" complementary P-channel silicon devices didn't start to really go mainstream until sometime well into the 1970s...

Weren't many designs still using BJTs into the '80s?

just brew it! wrote:

I generally just go with "tweak the EQ until it sounds about right". At the end of the day it's all very subjective anyway, and my gear isn't high-end, so it's mostly about compensating for the shortcomings of my speakers/headphones.

Speaking of subjective listening tests (and sorry for the tangent), I did notice some improvements in the punch of the bass and clarity of the highs when I switched from onboard Realtek to a Xonar DSX recently. Not a huge difference, mind you... but noticeable. Onboard has come a long way in the past decade and a half, but IMO when VIA faded from the scene and Realtek became the de facto standard we took a small step backwards. I still have a couple of Asus motherboards from the late '00s that have VIA onboard on them, and they sound better than the more recent Realtek ones... especially when directly driving headphones. Unfortunately they're old enough that they are now relegated to server duty, where the better onboard audio implementation goes to waste.

Gotta agree on the Realtek note.  For most people, it's totally fine, but if you care at all about sound quality and have even something relatively decent to listen with, it really doesn't cut it.  I've been using a Creative Sound Blaster X-Fi Titanium Fatal1ty Professional Series (that's a mouthful) and aside from iffy drivers (found some better custom drivers by a dude named Daniel_K who also makes drivers for Xonar cards), I've been very happy with it for my PC audio for years now.  It's very much obviously better than any integrated audio I've heard.  It's nice to see some motherboard makers at least trying with some of the onboard audio in some of the higher end motherboards, but we don't see anything like that in laptops these days.

Acidicheartburn wrote:

... and aside from iffy drivers (found some better custom drivers by a dude named Daniel_K who also makes drivers for Xonar cards), I've been very happy with it for my PC audio for years now.

I use the bundled in-tree Linux drivers (I researched Linux compatibility of the Xonar DSX beforehand), plus the JACK audio stack, and a plugin to convince PulseAudio to co-exist peacefully with JACK instead of fighting with it for control of the audio hardware. Getting this (self-imposed) kludge of an audio stack set up properly was was kind of fiddly (which I expected), but once it is set up it does everything I want. It's basically the same software setup I've been using for a number of years with Realtek (and VIA before that) onboard, but now with a better DAC and better analog output circuitry.

just brew it! wrote:

Weren't many designs still using BJTs into the '80s?

Many still do. Shine a flashlight into my Yamaha RX-V473 7.1 channel receiver, and you'll find fourteen BJTs in a neat little row.  Complementary pairs are cheap and very well-matched now, while Class-D designs remain difficult and expensive to build with good high-frequency performance characteristics.  While many units DO have switchmode power supplies in place of conventional transformers, they tend to maintain a conventional linear output stage.

Well, I will have to defend the crab here. I've been both the victim of bad motherboard audio and good motherboard audio. In all honesty, in the last 10-ish years, it's all come down to the motherboard's DACs, op-amps, and circuitry noise isolation. The digital side, that is, the Realtek and C-Media chips, work perfectly fine. In fact, I'll even go as far to say that if I have a good output stage, I'd rather have a Realtek solution over anything else, simply because the drivers Just Work.
Then again, the best solution for PC audio folks that aren't interested in surround has been an affordable entry-level USB studio interface, really.