lumanauw said:
This is a common experience. I think that the sound of amplifiers
using loop feedback becomes poorer as they approach the level
of feedback which would cause oscillation. They are not actually
oscillating, but they are close.
You may recall your feedback experiment as one of my "chip amp"
techniques in which the input networks are used so as to throw
away some open loop gain.
I have looked for chaotic behavior such as higher noise and period
doubling near the point of instability, but have not yet seen it.
Yes, Mr. Pass, I know only few people mentioned this technique. First is yourself, and the second is Bob Pease from National.
In his book (Troubleshooting Analog Circuit), and paper LB-42 http://www.national.com/ms/LB/LB-42.pdf he mention this technique.
Mr. Pass, do you mean the change in sound is not due to ammount of feedback applied, but because of the amp itself near oscillating in follower mode?
This is difficult for us DIYer's. Every opamp I tried is stated as "Unity Gain Stable". I remember you have experience with OPA604 in unity gain too.
While using a resistive divider to reduce the signal works, I find myself bothered by the two-steps-forward-one-step-back philosophy it implies. If you amplify a signal 20dB in the previous stage, then throw away 20dB with a 1/10 divider, what have you gained? Nothing but noise and distortion. Ugh.
I'm not disputing that it improves the sound of the opamp under the stated conditions, but wouldn't it be better simply to build a discrete circuit with less gain in the first place? More elegant and even better sound.
This is something that you see in high gain preamps. If you've got so much gain that you're turning the volume knob way down to achieve a listenable level, all the preamp is really doing is adding noise and distortion. My old CJ Premier Three preamp's line stage is at 0dB gain at about 10 o'clock on the volume knob. (I've got a small piece of masking tape stuck there with a little hash mark. Very informative to know how much gain you're actually using.) If I'm below that level, I know that all the line stage is doing is goofing up the signal--might as well use my Line Drive or Nelson's First Watt follower preamp circuits (very nearly the same circuit) that don't supply any (voltage) gain at all. Obviously, above that level the line stage begins to earn its keep, but the thing's got something like 26dB gain, which means that there's a certain amount of yuck being added to the signal whether I use that much gain or not. Which I never do.
Go back and look at Nelson's volume circuit for the Aleph preamp...the one that below unity gain simply serves as a pass-through for the input signal. I like that circuit. Simple. Clever. Sign me up.
Opamps have too much gain and too little bandwidth for their own good. They're always tripping over their own shoelaces trying to add more gain so they can apply more feedback so they can correct the distortion generated by the extra gain. What's wrong with this picture? About 100dB too much gain and a couple hundred kHz too little bandwidth. Jeez, some of those things only have an open loop bandwidth of 10Hz or so. That means you have to apply 10000x worth of leverage (this translates to 80dB of negative feedback), minimum, to even begin to get anything like reasonable bandwidth. Whew! That's lotsa feedback!
There's no denying that opamps are cheap and convenient, but I'm reluctant to use them for audio signals. It's not that the pig sings so well...the miracle is that it that it sings at all.
Grey
EDIT: Note that the Line Drive and First Watt preamps are stable at unity gain.
I'm not disputing that it improves the sound of the opamp under the stated conditions, but wouldn't it be better simply to build a discrete circuit with less gain in the first place? More elegant and even better sound.
This is something that you see in high gain preamps. If you've got so much gain that you're turning the volume knob way down to achieve a listenable level, all the preamp is really doing is adding noise and distortion. My old CJ Premier Three preamp's line stage is at 0dB gain at about 10 o'clock on the volume knob. (I've got a small piece of masking tape stuck there with a little hash mark. Very informative to know how much gain you're actually using.) If I'm below that level, I know that all the line stage is doing is goofing up the signal--might as well use my Line Drive or Nelson's First Watt follower preamp circuits (very nearly the same circuit) that don't supply any (voltage) gain at all. Obviously, above that level the line stage begins to earn its keep, but the thing's got something like 26dB gain, which means that there's a certain amount of yuck being added to the signal whether I use that much gain or not. Which I never do.
Go back and look at Nelson's volume circuit for the Aleph preamp...the one that below unity gain simply serves as a pass-through for the input signal. I like that circuit. Simple. Clever. Sign me up.
Opamps have too much gain and too little bandwidth for their own good. They're always tripping over their own shoelaces trying to add more gain so they can apply more feedback so they can correct the distortion generated by the extra gain. What's wrong with this picture? About 100dB too much gain and a couple hundred kHz too little bandwidth. Jeez, some of those things only have an open loop bandwidth of 10Hz or so. That means you have to apply 10000x worth of leverage (this translates to 80dB of negative feedback), minimum, to even begin to get anything like reasonable bandwidth. Whew! That's lotsa feedback!
There's no denying that opamps are cheap and convenient, but I'm reluctant to use them for audio signals. It's not that the pig sings so well...the miracle is that it that it sings at all.
Grey
EDIT: Note that the Line Drive and First Watt preamps are stable at unity gain.
lumanauw said:
Yes. Any time you approach instability, there seems to be a sonic
degradation. This is, of course, my anecdotal opinion, and I have
not conducted double blind listening tests with statistical analysis
of a population of golden ears.
GRollins said:
Naturally, I'm just being Mr. Helpful to those who insist on using
integrated chip amplification. Not that they seem to appreciate it...
lumanauw said:
At some point in the dim mists of prehistory, man learned to use fire to cook food. Heating the food killed microorganisms that caused disease and spoilage and made the meat more tender. As a fringe benefit, it also improved the flavor.
Imagine the consequences if cavemen had obstinately refused to cook their food until the various processes were understood. To this day we still would be unable to cook our food because we still don't understand everything that happens. Maillard reactions, caramelization of sugars, denaturing of proteins, dehydration, oxidation, carbonization...the list is nearly endless, and we're nowhere near exhausting the various interactions.
For that matter it was long known that it was safe to drink beer, even when the water it was brewed from made people ill. And yet no one knew why. Even such a simple thing as boiling (an integral part of the brewing process) killing pathogens was not yet known.
And don't get me started on wine.
Those are only a few examples of many I could give (consider the history of aspirin, for instance), but the same attitude crops up frequently in audio. People mock what they do not understand and adamantly refuse to consider even the most rudimentary concepts proposed by those who listen because there's no known explanation for some of the observed results. The proposed explanations for some of the phenomena are ridiculed because they haven't been 'proven' to the satisfaction of those who complain. (Once they are proven, you can bet there's no credit or thanks given to those who endured years of abuse trying to bring attention to what they were hearing...go figure...) The Blowtorch thread seems to attract this sort of person, so if you want to watch this mindset in action, feel free to browse any random five pages. You'll see posts from certain individuals that clearly feel that anything they can't find in a book doesn't exist and that anyone who posits anything beyond their rather limited view of the universe is deluded; made a target for scorn and derision.
As for the "why" of the feedback problem, it's almost certainly related to the time delay involved. No matter how fast a feedback loop may be, there's always some delay involved and you're correcting something that's no longer there...in other words, you are introducing distortions.
Observationally, there's a reduction in subtle detail. Imaging, which happens to be of particular interest to me, suffers, but it's not the only thing. For those who prefer to immerse themselves in math, it's been shown that feedback converts lower order distortion products into higher harmonics. The absolute levels may be lower, but the higher harmonics are disproportionately more annoying. No, the "more information" so beloved of high feedback proponents...ain't. Finger noise on guitar strings doesn't really sound like that. It all boils down to having an idea of what real music actually sounds like. Declaring that it "must" sound better because the THD is lower is poor reasoning. The underlying assumption that THD covers everything is flawed.
It's not particularly good at discriminating transient phenomena like intermittent ringing, for instance. Music, with irregular waveforms and wide dynamics, is much better suited, but it's hell to work with on the test bench.
Grey
Mr. Pass,
Thanks, for the very nice article on distortion ... so much to learn. But, you make it so easy, it's really appreciated!
Really good thoughts ... I agree, if only I could express myself that well.
I once auditioned a Counterpoint amp. It was the 220 watt hybrid amp and it produced this huge sound stage, and very large imaging, but the sound was "caramel colored." If that makes sense. I read reviews the seconded my oppinion on that. I'm guessing distortion caused this affect?
I'm wondering if distortion can take away form the imaging and sound stage, maybe it can add to it as well.
Like you say, the only true reference is live acoustic music ... and reproduced on high quality audio playback formats, like SACD, or vinyl.
Thanks, for the very nice article on distortion ... so much to learn. But, you make it so easy, it's really appreciated!
GRollins said:
Really good thoughts ... I agree, if only I could express myself that well.
I once auditioned a Counterpoint amp. It was the 220 watt hybrid amp and it produced this huge sound stage, and very large imaging, but the sound was "caramel colored." If that makes sense. I read reviews the seconded my oppinion on that. I'm guessing distortion caused this affect?
I'm wondering if distortion can take away form the imaging and sound stage, maybe it can add to it as well.
Like you say, the only true reference is live acoustic music ... and reproduced on high quality audio playback formats, like SACD, or vinyl.
The accusation that distortion somehow "adds to" or causes image is usually advanced by those who look at low feedback designs (e.g. tubes) and conclude that since they image better and have somewhat higher measured distortion, that it's the distortion that's causing or enhancing the image. I find it difficult to imagine a mechanism whereby distortion--out of all the possible effects it could have--would just happen to alter the sound in such a manner that it sounds more like a real image heard in a concert hall. I'm not saying that it's impossible, just statistically unlikely.
I first heard this hypothesis twenty-five or thirty years ago. The high end, as a market segment, was still young. Solid state manufacturers were getting their butts kicked in the imaging department by Audio Research, Conrad Johnson, and resuscitated Dynacos. They had to say something, didn't they? The only two honest claims they could make were that solid state amps had better damping factors and that they had lower measured distortion. So they played to their strengths. No surprise, there. Unfortunately, then, as now, they were saying some pretty wild things.
The market voted. Audio Research and Conrad Johnson are still here. Nearly all of their solid state competitors are gone and the few that remain are...um...changed.
Grey
I first heard this hypothesis twenty-five or thirty years ago. The high end, as a market segment, was still young. Solid state manufacturers were getting their butts kicked in the imaging department by Audio Research, Conrad Johnson, and resuscitated Dynacos. They had to say something, didn't they? The only two honest claims they could make were that solid state amps had better damping factors and that they had lower measured distortion. So they played to their strengths. No surprise, there. Unfortunately, then, as now, they were saying some pretty wild things.
The market voted. Audio Research and Conrad Johnson are still here. Nearly all of their solid state competitors are gone and the few that remain are...um...changed.
Grey
People tend to get obsessed with the distortions they know. They choose one, make it the devil incarnate, and bend all their efforts towards eradicating it. This is the classic "when all you have is a hammer, everything starts looking like a nail" mindset.
I'm more interested in the ones we don't yet know or the ones that are incompletely understood. I find the "thermal distortion" thing promising. Lavardin? Is that the guy's name? One of these days, I'm going to go to the trouble to rig an experiment with a power resistor on the heat sink next to the transistor. Every time the transistor cools, I'm going to heat it up with the resistor. If it gets too hot, I'll cut back on power delivered to the resistor or adjust water flow to cool it. To be worthwhile, this will have to be done on a system with small thermal mass so as to respond quickly. Conversely, you could take it to the other extreme and use a really massive system that achieves such thermal inertia that it's difficult to make it change temperature no matter what you do. (My water cooled system already approaches this, I think, but it's an uncontrolled experiment as things stand now.) Note that Charles Hansen's three-ton block of aluminum amplifier (the MX-R) might be a fair step in the right direction. No, John Curl's Blowtorch doesn't qualify because--as I understand it--he's not using the chassis as a heat sink. It's just massive to block EM fields. That's not to say that line level amplification can't benefit from controlled temperatures.
I first started thinking along these lines when I was doing the Aleph-X. I started thinking about a block of metal that I called a "thermal capacitor" to promote DC stability. Later, it occurred to me that it would help with thermal distortion (assuming that such a thing really exists) as well. Fringe benefits including heat sinking ability make it even more attractive.
Interesting factoid--class A amps actually cool down the harder they're driven, unlike class B or AB amps, which heat up.
To the extent that there might be anything to thermal distortion, class A amps with lots of heavy metal might arguably exhibit less distortion than lightweight AB or B amps. Given that most of the delicate aspects of music are more noticeable in quieter passages, class A amps will cool down less, proportionately, than lightly biased amps heat up.
I believe the MX-R is class AB. (Nelson, can you confirm this? I've got an e-mail from Charlie somewhere that mentions his preferred bias levels, but I'd have to dig to find it--and I don't know whether or not it applies to the MX-R, anyway.) Charles's chassis design leaves me curious as to what would happen if one were to follow a similar approach with a class A amp. This leads me back to my water cooled system, which takes quite a while to heat up and quite a while to cool down...though it's far from portable and would not make a good commercial product. If you're willing to build everything in situ and leave it be, it's a beautiful thing for the DIYer, though.
Grey
I'm more interested in the ones we don't yet know or the ones that are incompletely understood. I find the "thermal distortion" thing promising. Lavardin? Is that the guy's name? One of these days, I'm going to go to the trouble to rig an experiment with a power resistor on the heat sink next to the transistor. Every time the transistor cools, I'm going to heat it up with the resistor. If it gets too hot, I'll cut back on power delivered to the resistor or adjust water flow to cool it. To be worthwhile, this will have to be done on a system with small thermal mass so as to respond quickly. Conversely, you could take it to the other extreme and use a really massive system that achieves such thermal inertia that it's difficult to make it change temperature no matter what you do. (My water cooled system already approaches this, I think, but it's an uncontrolled experiment as things stand now.) Note that Charles Hansen's three-ton block of aluminum amplifier (the MX-R) might be a fair step in the right direction. No, John Curl's Blowtorch doesn't qualify because--as I understand it--he's not using the chassis as a heat sink. It's just massive to block EM fields. That's not to say that line level amplification can't benefit from controlled temperatures.
I first started thinking along these lines when I was doing the Aleph-X. I started thinking about a block of metal that I called a "thermal capacitor" to promote DC stability. Later, it occurred to me that it would help with thermal distortion (assuming that such a thing really exists) as well. Fringe benefits including heat sinking ability make it even more attractive.
Interesting factoid--class A amps actually cool down the harder they're driven, unlike class B or AB amps, which heat up.
To the extent that there might be anything to thermal distortion, class A amps with lots of heavy metal might arguably exhibit less distortion than lightweight AB or B amps. Given that most of the delicate aspects of music are more noticeable in quieter passages, class A amps will cool down less, proportionately, than lightly biased amps heat up.
I believe the MX-R is class AB. (Nelson, can you confirm this? I've got an e-mail from Charlie somewhere that mentions his preferred bias levels, but I'd have to dig to find it--and I don't know whether or not it applies to the MX-R, anyway.) Charles's chassis design leaves me curious as to what would happen if one were to follow a similar approach with a class A amp. This leads me back to my water cooled system, which takes quite a while to heat up and quite a while to cool down...though it's far from portable and would not make a good commercial product. If you're willing to build everything in situ and leave it be, it's a beautiful thing for the DIYer, though.
Grey
GRollins said:
I own both tube and solid state gear. One of my favorite amps was a 1974 Dunlap-Clarke, Dreadnaught 500, which had a damping factor of 1000. I slightly modified it by adding more filter capacitance and bypass caps. Anyway, I had it driving our Thiel CS1 loudspeakers in our upstairs living room (hardwood floors with good acoustics) and it sounded just really natural and musical. It got the timbers right, and the sound stage and imaging sounded good to my ears. It was great until it went DC and destroyed one of the speakers.
I completely understand what you mean about tubes having a more "you are there feeling." Right now I have my record player hooked up to a tube phono preamp and this combo sounds really nice. Two months ago I was playing a old 1950s LP of Rimsky-Korsakov's Scheherazade (mono) and it just pulled me in, despite all the surface noise of this LP. Same, with Jimmy Hendrix's, Wild Blue Angel LP.
When playing SACDs on this system (all solid state) it sound's good but somewhat dry, and you don't feel as pulled in to the music. This in our main listening room which pretty dead. The amp I'm using is a commercial discrete op-amp based solid state amp. It has two gain stages with local feedback only.
This is a lousy comparison. I've used tube preamps in this system too, and it does and smoothness and musicality. But, my point is that the tubes in my system aren't taking any distortion out of the recording, or amps. They are adding to it. And given the state of most modern recordings, that's not such a bad thing. I mean the distortion spectrum does matter, I think.
Maybe adding some 2nd and 3rd harmonic distortion makes those higher harmonic distortions less offensive. And maybe it also increases the perceived imaging and soundstage. 2nd and 3rd harmonics are found in music, and are more natural sounding. Maybe they make room reflections less annoying and creating a subjectively larger space?
I mean the imaging and soundstaging of that Counterpoint wasn't like normal. The singers sounded 12 feet tall. How can one hear that? Well, I don't understand psyco-acoustics. Another mystery I guess.
I agree, isn't it terrible, the only thing worse is when "those who listen" draw conclusions from sighted, uncontrolled listening. Then they post to the world "I hear this and I hear that", and start long discussions of technical post-rationalisations for an effect that they would never have heard in a controlled environment.GRollins said:
It would be easy (for someone competent, not me) to plot that delay as a delta-V on the transient signal voltage. What would the distortion level be? I expect not much, but surprise me, my mind is open, I was surprised by NP's article.GRollins said:
Johnloudb said:
--Audio equipment (whether tube or solid state) never takes distortion out of the signal. It can only add. And, no, there's no good way to make distortions cancel out between two different pieces of gear. It's always additive.
Lest someone think otherwise, Nelson's "X" circuit depends on two precisely matched circuits being operated in opposition. By definition, preamps and amps are not the same circuit and do not operate in opposition.
--Back when I was in the retail end of things we sold Counterpoint for a while. I never particularly cared for it. For that matter, I've never heard a hybrid amp that I liked, regardless of manufacturer.
tnargs said:
--Your sneering attitude betrays you. You really ought to try listening sometime, you might be surprised...on the other hand, perhaps not, since you appear to have already made up your mind that there's nothing there to hear. (And the THD Is God crowd claim that people who listen have preconceived group-think notions! They must not own mirrors.)
While it's obvious that there are those who will follow any fad, the fact of the matter is that people who actually listen don't necessarily follow like sheep. My personal favorite example is the VPI Magic Brick. The concept made perfect sense. The product wasn't insanely priced. It wasn't difficult to use. It's just that I (and some others who actually listened) never bought into the mystique. The irony being that, although I never bought one, I now own several. People gave me theirs after they grew tired of them. They make dandy paperweights.
Your condescending presumption that people are incapable of independent thought is not only insulting, it simply is not true.
Incidentally, the Magic Brick saga also shows the value of long term listening on a familiar system. Had those people done so (or had the opportunity to do so--a separate issue) they probably would not have bought the product. It's only after extended evaluation that they concluded that the Bricks did nothing.
--If all this were 'easy' everything would have been solved years ago. Oh...wait...to the True Believer in THD and double blind testing, it was all solved years ago. How silly of me!
That doesn't explain how things like the audibility of passive components slipped under the radar until the late '70s. Oooops!
Grey
You should read you own posts. Pot, kettle, black.GRollins said:
Again, you are guilty of what you are criticising someone else for.GRollins said:
As for not listening, perhaps that is less so on this forum, as with few exceptions it generally has a much higher engineering standard than other audio forums. However, I can show many, many examples of herd mentality over many products and many years elsewhere and an investigative attitude is fairly rare amongst audiophiles.
People who listen have advanced many ideas over the years. Quite a few have been demonstrated to be true, in spite of the jeering and "it's all in your imagination" accusations advanced by those who can't be bothered to listen. Audibility of passive parts, absolute polarity (Douglas Self made a bit of a spectacle of himself in an earlier edition of his amp book deriding those who claimed absolute polarity was audible--but he was man enough to admit he was wrong in a later edition), time alignment in speakers (ironically, this particular battle has had to be fought more than once--you'd think people would learn, but no...), etc. etc. etc.
Always the same story: It's BS. It's all in your mind. It's delusion. It's sheep following the guru of the week.
But lawsy me, let the thing be proven and the mouthy ones run for the hills. Not a word of apology. No humility. And do they learn to keep an open mind on other matters? Of course not. Same nonsense all over again. "It can't be." "THD covers everything." Blah, blah, blah...
And what do they bring to the table?
Nothing.
Ever.
'Cuz, you see, since THD is all you need, audio reached the pinnacle of perfection in the late '70s with the .0001% distortion figures. And since audio is already perfect, the only thing to do is criticize those who are interested in making it better, since they're clearly imaging defects where none exist.
Grey
Always the same story: It's BS. It's all in your mind. It's delusion. It's sheep following the guru of the week.
But lawsy me, let the thing be proven and the mouthy ones run for the hills. Not a word of apology. No humility. And do they learn to keep an open mind on other matters? Of course not. Same nonsense all over again. "It can't be." "THD covers everything." Blah, blah, blah...
And what do they bring to the table?
Nothing.
Ever.
'Cuz, you see, since THD is all you need, audio reached the pinnacle of perfection in the late '70s with the .0001% distortion figures. And since audio is already perfect, the only thing to do is criticize those who are interested in making it better, since they're clearly imaging defects where none exist.
Grey
I was nuking my supper here at work and busted out laughing.
CD. Digital recording. "Pure perfect sound forever." Remember? Now that was a good one, although there are many here who are too young to know what I'm talking about.
Instantly...and I mean instantly on release of the first CDs, those who actually listened condemned the new format as brittle sounding, sterile, harsh, shrill, and a thousand other similar terms. (I've still got an old Chalfont digital recording of a classical piece that I keep around as an object lesson in just how bad early digital really was.) Those who defended the fledgling format made hysterical defensive claims:
--It can't be the digital format! Just look at the distortion specifications!
--It wasn't the digital format making those ugly sounds...it was the perfection of digital finally revealing the flaws in the (pick one) mic techniques, mixing boards, analog circuitry, LP reproduction (an entire sub-argument insisted that LPs ruined the 'perfection' of the digital format--the only way to hear it in all its glory was on CD), etc.
--Ooooh! Look! It's so quiet! The lack of noise is now revealing the flaws in the (pick one) mic techniques, mixing boards, analog circuitry, LP reproduction, etc.
--Ooooh! Look! It's so dynamic! The amazing dynamic capabilities of digital are showing how pathetic the (pick one) LP, phono stage, line stage, amplifier, speaker, etc. are.
And we all know who won the Digital Wars...right?
(HINT: It wasn't the measurements-are-everything, double blind, 'pure perfect sound forever' crowd. Major changes were needed...pronto!)
Sure enough, Red Book digital brought with it a host of brand new problems. It was positively infested with things that boogered up the sound...but didn't show up in conventional measurements. But goodness me, the measurements crowd was scathing in their condemnation of those who actually listened, calling them all sorts of names and ridiculing them in exactly the same condescending way that they do today when anyone says they hear something (or don't, as the case may be).
The ears won (again). The meters lost (again). And we all had to learn a passel of new terms like dithering, LSB error, jitter, and the like.
Did the horses' rear ends who mocked those who pointed out that the emperor had no clothing ever apologize for their behavior?
Huh!
Guess you know the answer to that one.
Grey
CD. Digital recording. "Pure perfect sound forever." Remember? Now that was a good one, although there are many here who are too young to know what I'm talking about.
Instantly...and I mean instantly on release of the first CDs, those who actually listened condemned the new format as brittle sounding, sterile, harsh, shrill, and a thousand other similar terms. (I've still got an old Chalfont digital recording of a classical piece that I keep around as an object lesson in just how bad early digital really was.) Those who defended the fledgling format made hysterical defensive claims:
--It can't be the digital format! Just look at the distortion specifications!
--It wasn't the digital format making those ugly sounds...it was the perfection of digital finally revealing the flaws in the (pick one) mic techniques, mixing boards, analog circuitry, LP reproduction (an entire sub-argument insisted that LPs ruined the 'perfection' of the digital format--the only way to hear it in all its glory was on CD), etc.
--Ooooh! Look! It's so quiet! The lack of noise is now revealing the flaws in the (pick one) mic techniques, mixing boards, analog circuitry, LP reproduction, etc.
--Ooooh! Look! It's so dynamic! The amazing dynamic capabilities of digital are showing how pathetic the (pick one) LP, phono stage, line stage, amplifier, speaker, etc. are.
And we all know who won the Digital Wars...right?
(HINT: It wasn't the measurements-are-everything, double blind, 'pure perfect sound forever' crowd. Major changes were needed...pronto!)
Sure enough, Red Book digital brought with it a host of brand new problems. It was positively infested with things that boogered up the sound...but didn't show up in conventional measurements. But goodness me, the measurements crowd was scathing in their condemnation of those who actually listened, calling them all sorts of names and ridiculing them in exactly the same condescending way that they do today when anyone says they hear something (or don't, as the case may be).
The ears won (again). The meters lost (again). And we all had to learn a passel of new terms like dithering, LSB error, jitter, and the like.
Did the horses' rear ends who mocked those who pointed out that the emperor had no clothing ever apologize for their behavior?
Huh!
Guess you know the answer to that one.
Grey