The textbooks will show you the complex-number algebraic equations that describe a gyrator, but how do we translate that to any kind of intuitive understanding? This is the sort of thing where I find LTSpice can help. It won't make the math any easier, but at least you can see what the circuit does, and the effect of changing any of the components.
Every time you mention buzzard hunching, you remind me of Snoopy.
And yes, LTSpice will probably lead to even more buzzard-hunching...
-Gnobuddy
The last time I paid attention (circa Windows 98), Windows used two separate invisible ASCII characters at the end of each line of text - a carriage return, and a line feed. If you've ever used an ancient mechanical typewriter, you'll be familiar with the concept - one key to move back to the start of a line of text, another key to advance to a new blank line.
This made good sense on a mechanical typewriter, because it let you re-type over the same line of text to correct errors or insert changes. Paint some White Out (TM) over the mistake, retype over it, and you don't have to type out the entire page again.
Unix (and its descendants, including Linux and FreeBSD) don't bother with archaic typewriter analogies, and instead use a single "newline" character that performs both functions. White Out doesn't work, and we don't re-type along the same line with a computer, so why do we need a carriage return without a line feed?
About 20 years ago, this sometimes caused headaches when transferring plain text files from Windows to Linux, or vice-versa. In Linux, you'd sometimes see garbage characters at the end of lines. In Windows, some text editors would fail to recognize the Linux newline character at all, and all of the text would show up as one single long line.
I haven't seen those problems since the very beginning of the 21st century, though. I'm surprised that the issue would re-surface here on diy Audio nearly two decades later.
(I don't know if that's actually what's happening, but it is a plausible hypothesis.)
-Gnobuddy
Attachments
Ms. Buzzard looks pretty mad about that overpriced Apple monitor.
-Gnobuddy
We've had numerous op-amps mentioned here, and I thought it might be a good idea to do some actual noise calculations. That will let us compare some of these op-amps and find out which is best, and if the best is sufficiently better to be worth using.
Let's start with the noise from the guitar itself. Every resistance generates a noise voltage across itself, because the electrons inside move erratically due to thermal energy. At any one instant, there will be more electrons at one end (and therefore a more negative voltage at that end). A split second later, there will be more electrons at the other end, and now that end will be negative. And so on, resulting an an erratic AC voltage across the resistor - thermal or Johnson noise.
Any guitar being used to play metal is pretty much guaranteed to use humbucker pickups. Typical DC resistance of the windings in these is somewhere in the 10k - 20k range. That resistance will generate thermal noise voltage, like it or not.
Even worse, the guitar will have a volume pot inside. To avoid loading down humbucker pickups (which dulls the sound), this will typically be a 500k pot. Set half-way, this is equivalent to two 250k resistors, one connected directly to ground, the other connected to ground through the resistance of the pickup coil. So we have one 250k resistor, in parallel with a second resistor of around 260k - 270k. For all practical purposes, this calculates out to about 125k, plus or minus a few percent.
So, worst-case, the guitar itself will make as much hiss (noise voltage) as a 125k resistor.
The actual amount of thermal noise voltage generated by a resistor depends on the range of frequencies we include in our measurement. A generous number is to use a 10 kHz bandwidth for guitar (actually there's nothing good from an e-guitar above maybe 5 kHz.)
If we do the calculation, the guitar itself, with the volume pot backed off to half-resistance, will generate about 4.6 micro-volts of noise over a 10 kHz bandwidth.
So there's our baseline. We cannot count on the guitar itself making less than 4.6 uV of hiss. As long as our guitar amp is quieter than this, then the hiss from the guitar itself dominates the total noise, and the amp itself will be blameless.
So that's our goal - our amplifier (or guitar pedal) needs to have an equivalent input noise of less than 4.6 uV over a 10 kHz bandwidth.
In my next post, we'll take a look at some of the op-amps mentioned here, and see how they stack up against the guitar's own irreducible thermal noise.
-Gnobuddy
Okay, on we go!
Op-amps have equivalent input noise voltage, and equivalent input noise current. If we divide the former by the latter, we get an imaginary resistance (by Ohm's law). It's considered a sort of figure-of-merit for an op-amp, and may be notated Rs.
That imaginary resistance tells us, roughly speaking, the biggest source resistance we can use with that op-amp: any bigger, and the op-amps input noise current will dominate the noise, which we don't want. But as long as our actual signal comes from a source impedance much less than this, we're good on the input-noise-current front: the op-amp we're using has a sufficiently low input noise current to work well.
For the TL07x op-amp, I calculated Rs at about 1.8 mega ohms. For the newer and shinier OPA164x, Rs calculates out to 6.4 mega ohms.
Since we're anticipating a source resistance of 125k (max) from our guitar, both the TL07x and the OPA164x are good choices - both of them have such low input noise current that we can be sure this particular source of noise will not be a problem.
That leaves the other type of op-amp input noise: input noise voltage (not current).
For the TL07x, specified at 18 nV/Root(Hz), the input noise voltage in a 10 kHz bandwidth calculates to 1.8 uV.
For the OPA164x, specified at an excellent 5.1 nV/Root(Hz), the input noise in a 10 kHz bandwidth calculates to 0.51 uV. Quite a bit less than the TL07x, at first sight.
Recall that the guitar itself can put out up to 4.6 uV of noise. So the first thing we can immediately see is that both the TL07x and the OPA164x are much quieter than the guitar itself. And this means the OPA164x will not be a dramatic improvement over the TL07x, because the guitar's own noise will dominate in either case.
Let's do the math to quantify this. Noise sources add as root of the sum of squares; we square each noise source, add them, then take the square root. Doing this with the TL07x's 1.8 uV and the guitar's 4.6 uV gives us a total noise of 4.94 uV (barely bigger than the guitar's own noise.)
Doing the same thing with the OPA164x's 0.51uV and the guitar's 4.6 uV, the total noise works out to 4.63 uV. Once again, barely bigger than the guitar's own noise.
What would be the improvement in noise by switching from the TL07x to the OPA164x? The noise drops from 4.94 uV to 4.63 uV, which is a drop of 0.56 decibels. The human ear can barely detect a 1 dB change, so 0.56 dB is such a small reduction in hiss that we won't even hear it.
So there's our answer. The OPA164x is a marvel of modern semiconductor technology, with far superior noise specs to the old faithful TL07x. But those better specs don't translate to better performance when connected to a conventional electric guitar, with passive pickups and a 500k volume pot.
Things might be different if the 500k volume and tone pots in the guitar are removed, and an op-amp preamp is wired directly to the pickup coil windings. With the 125k source resistance reduced to some 10k - 20k of the pickup winding, the guitar will make considerably less thermal noise, and the OPA164x might very well turn out to be much quieter than the hoary old TL07x in this situation. (Do the math to be sure!)
By the way, I left the TL06x out of these calculations because (a) it has about twice the noise voltage of the TL07x series, and (b)I couldn't even find any input noise current spec for the TL06x.
What the TL06x does well is run on a 9V battery - it draws much less power than the TL07x and other op-amps.
This isn't as big a deal in a guitar pedal these days, as the rising prices of 9V flat batteries, and the falling prices of 9V switching power supplies, have convinced most guitarists to switch over to external power supplies for their guitar pedals, rather than internal 9V batteries.
-Gnobuddy
That's fine, just think of it as a way to make a virtual L-C network without using a physical (more expensive, bulky) inductor. My thought was actually to reduce the NFB resistor in the associated op amp circuit, not play with the values of the gyrator per se. Maybe knocking down the gain in the stages but not playing with the passbands or Q, might tame the noise a bit.
The plan that seemed to be the best bet, was to replace the first FET stage with a quiet opamp with a gain of about 4, then reducing the gain of the second stage accordingly. I hope the problem will be solved without messing with that part of the circuit. Perhaps a better opamp in the second, but my intuition says that won't do anything, especially with the reduced feedback resistances in the plan.
@Gnobuddy: Where is the voltage coming from when the guitar is not outputting anything to create voltage noise in the resistances?
@Gnobuddy: Where is the voltage coming from when the guitar is not outputting anything to create voltage noise in the resistances?
I wrote a driver for the Diablo daisywheel. You could easily microcode some elaborate text enhancement. At a low level you would backspace and re-type for bold, or backspace for underline. But at some point it became much faster to CR and do the enhancing on a second pass. (CR was much faster than 10 BSes.)
CMP/MSDOS tended to drive things up to a Diablo. And print-filters (pipes) were awkward. Unix tended to fast-drafts and then the final product would be piped to a print-filter on a much fancier type-set machine. The thinking was different.
If you think Microsoft wasn't thinking it through, I won't disagree.
Resistances are built out of atoms, right? Those atoms each donate electrons to create a "sea" of electrons inside the resistor. But thermal energy - the temperature of the resistor - won't let the electrons lie in peace. Instead, they are battered back and forth randomly and continuously.
Moving electrons equals electric current flow. Current flow through a resistance equals electric voltage. So as the electrons flutter hither and yon, there is an associated fluttering and irregular noise current through the resistor, and a corresponding fluttering and irregular noise voltage across it.
Weird, right? But true. Every resistor generates its own noise, just sitting there connected to nothing at all!
If you could cool the resistor to absolute zero temperature, there wouldn't be any heat energy to disturb the electrons, so the noise goes away. We can't get to absolute zero, but immersing your electronics in (very cold) liquid nitrogen goes a long way to reducing noise. Immersing really sensitive electronics in liquid helium (even colder) goes even further.
NASA cools down their satellite radio receivers that pick up weak signals from satellites far away in space. I know liquid nitrogen cooling was routine. Liquid helium is much more expensive, and will destroy many electronic components, but I believe it was used in especially sensitive electronics. Those secret military receiving stations we never hear about must do the same things.
If you think about it a little deeper, this thermal noise in resistors is actually a quantum physics effect: electricity isn't a perfectly smooth "fluid", as people thought in the early 1800s. Instead, it's a stream of little lumps (quanta) of electric charge which we call electrons. And it's the "lumpiness" of the charge that manifests as thermal noise in a resistor!
There are more easily visualized equivalents to this phenomenon. For instance, the air molecules around us are shooting rapidly this way and that, all the time, because of heat energy. If you put a microphone diaphragm in the room, the little air molecules bombard it just like rain pelting down on a tin roof. As a result, the microphone diaphragm gets shaken about ever so slightly - there will be a tiny random noise generated by the impacts. Even if there is no sound at all in the room!
If the microphone has a large diaphragm, there are more impacts (from air molecules) per second, and the noise gets "smoothed out" a bit - so large-diaphragm mics generate less of this acoustic thermal noise than small-diaphragm mics do.
Note that we're not talking about noise in the microphone electronics - the type of noise we're talking about here comes from the physics of air molecules at non-zero temperature. This too is a quantum effect.
It happens in water, too. After the microscope was invented, botanist Robert Brown noticed that if you put tiny grains of pollen in water and looked at it under the microscope, the pollen grains didn't stand still, but instead moved slowly and erratically about. He reported this in 1987, but nobody could explain it at the time. Maybe it was some sort of black magic, or maybe the pollen was alive in some way?
Eventually Einstein (yes, that Einstein!) came along and figured it out. Yup, it's exactly the same thing as the thermal noise in a resistor, or the thermal acoustic noise in a microphone diaphragm. Water molecules zip about due to heat energy too, and when they hit the pollen grain, it moves a tiny bit. Because the water molecules hit the pollen erratically from every direction, the pollen grain moves slowly and erratically about.
-Gnobuddy
Fascinating!
Mebbe Tim Paterson (who wrote the original DOS) owned a Diablo, and Bill Gates wasn't smart enough to rip out the relevant code?
-Gnobuddy
Refocus
Agreed. So when the input is grounded on the fuzz box, it is the 10k, the 1M, and the battery creating input noise to the first amplifying device.
The hiss is virtually the same with the guitar plugged in or not, or if a shorting plug is inserted in the input. The exception is backing off the volume pot to around 8. This resultant added hiss is negligible and does not present a problem.
It is the consensus that even if the noise in the first stage went to zero, when DIST (overall distortion) is up, and the EQ knobs, MID, and HIGH are up, it is guaranteed this state of affairs will bury an noise in the first stage and most likely the second as well.
I will settle for a 10% improvement without the controls even near maxed out. I have 50 USD to spend on this mod.
Agreed. So when the input is grounded on the fuzz box, it is the 10k, the 1M, and the battery creating input noise to the first amplifying device.
The hiss is virtually the same with the guitar plugged in or not, or if a shorting plug is inserted in the input. The exception is backing off the volume pot to around 8. This resultant added hiss is negligible and does not present a problem.
It is the consensus that even if the noise in the first stage went to zero, when DIST (overall distortion) is up, and the EQ knobs, MID, and HIGH are up, it is guaranteed this state of affairs will bury an noise in the first stage and most likely the second as well.
I will settle for a 10% improvement without the controls even near maxed out. I have 50 USD to spend on this mod.
An ideal battery has zero resistance and doesn't make any noise. Real batteries do make some noise, but we always put a big fat electrolytic cap across the power supply rails to suppress it to negligible levels. So the battery is unlikely to be the culprit.
As far as AC signals (and noise) are concerned, the 10k and the 1M appear in parallel to each other, as we've discussed a few times in this thread. So there is effectively a single 9.9 k resistor - effectively, a 10k resistor - feeding thermal noise into the gate of the JFET.
The JFET will then add some noise of its own. And the JFETs 10k source resistor will add a little more.
This tells me the schematic you posted is wrong. The input jack must be internally wired as a shorting jack, in such a way as to short the input to ground when nothing is plugged in. The posted schematic doesn't show this, however.
This explains why you don't hear any change in hiss when a shorting plug is inserted into the input. It also explains why the hiss level goes up when a guitar is plugged in and the pot backed off to 8 (this is probably about the point where the source impedance is highest - volume pots are usually "audio taper" or log pots, and they hit half-resistance when the pot is barely turned down from maximum.)
Now, that is interesting. If I'm understanding you correctly, you're saying that there is negligible change in hiss whether the input is shorted (zero impedance), or has 125k source impedance (4.6 uV of thermal noise.)
If correct, this means the MT-2 has equivalent input noise equal or greater to the 4.6 uV thermal noise from the guitar with the volume pot backed off to 8 (all numbers here are ballpark, not exact.)
That noise isn't coming from the 10k (too small), or the 1 Meg (it's shunted by the 10k). So most of that noise is either from the JFET, or from the first op-amp itself.
I looked up the stock op-amps datasheet, which shows an input noise voltage of around 10 uV/root(Hz). Not spectacularly good by today's standards, but not terrible, either - in fact, it's less than the thermal noise of a 10k resistor. And much less than the thermal noise of a 125k resistor.
This is a bit of a head-scratcher. If it meets its datasheet specs, the op-amp is supposed to be much quieter than a 125k resistance. But the MT-2 itself isn't. So where is the noise coming from? Surely that JFET can't be that much noisier than the 4558 op-amp, can it?
So I looked up the 2SK184 datasheet. No smoking gun there, either. The JFET is noisy at low frequencies (10 Hz), but e-guitar starts at 83 Hz, and most of it is above 160 Hz. And the 2SK184 is reasonably quiet up there.
I'm somewhat baffled at this point. Apparently the MT-2 is noisier than a 125k resistor, but it shouldn't be; neither the JFET nor the op-amp is supposed to generate that much noise.
FWIW, I'm not part of that consensus.It is the consensus that even if the noise in the first stage went to zero, when DIST (overall distortion) is up, and the EQ knobs, MID, and HIGH are up, it is guaranteed this state of affairs will bury an
I haven't done any calculations for this specific circuit, but it would be very unusual for the second stage noise to exceed the noise from the input stage.
Most likely, what the EQ is doing is amplifying the noise from the first stage. (Not adding lots more noise from the EQ stage itself.) Adding the usual "smiley face" metal EQ curve will boost hiss from the input stage and make it sound worse. (But the hiss isn't coming from the EQ stage, it's coming from the input stage.)
At this point, I'm wondering how you can even get that, because I haven't been able to figure out why your pedal is as noisy as it apparently is.
If you have a clean-boost pedal, or a graphic EQ pedal (which usually also has a small clean boost included), or if you can borrow one from another guitarist, I urge you to try the simple thing first: chain the pedals, clean boost before MT2, turn down the distortion / gain on the MT2, see if noise performance is better or not. If not, save your $50 to spend somewhere else...
I mentioned the ancient Phillips DNL. As best I can recall, it is a simple circuit, and quite effective. The original was stereo (for use with a Phillips stereo cassette tape player), but you only need one channel for guitar, so even easier to build.
I'm wondering if you might get much more hiss reduction by sticking a DIY Phillips DNL after your MT-2, than by tinkering inside the guts of the MT-2 itself.
-Gnobuddy
WAY off-topic.....
> Mebbe Tim Paterson
Can't blame him. He specifically copied CP/M traditions. We should note that each application really decided what line termination. However it was common to copy a file to the printer. So it comes down to what the more common printers supported/expected.
> Mebbe Tim Paterson
Can't blame him. He specifically copied CP/M traditions. We should note that each application really decided what line termination. However it was common to copy a file to the printer. So it comes down to what the more common printers supported/expected.
Since the box had been received by the seller/repairman with no output, and his replacing all the opamps with new ones made it work again (according to him), it is possible the unit was connected to a higher voltage power supply, which may have damaged the FET slightly.
To do:
See if there's a big cap across the rails.
See if the input jack is a shorting type.
Replace the FET with the 1641 and low noise resistors. (There are already in the mail).
See what happens.
If no noticeable improvement, replace the first opamp (second stage in chain) with a 1611 hastily ordered (also in the mail). See what happens. If no go, investigate the Phillips DNL circuit in any case. Sounds interesting.
Or go back to work on my soft treble gate. I don't know if this buzzard has it in him, though, or even the gate circuit. At that point, I envision capitulating and acquiring a pedal gate.
Thanks for your comments!
I seem to distinctly recall an article (circa 1980) written by one Walter Jung that describes the clearly obvious improvement made to a Dynaco preamp by substituting film caps for the ceramics and electrolytics that were stock. Walter is no "technically ignorant audiophile" as he was one of the chief designers at Analog Devices and clearly one of the most brilliant electronics minds in audio.
Absolutely brilliant people - including scientists and engineers - often hold superstitious beliefs that are contrary to all physical evidence.
As just one example, Linus Pauling was a genius, and a Nobel Prize winner in chemistry. He believed mega-doses of vitamin C protected one from the common cold. As the years rolled by, there was a growing mountain of evidence that this simply was not the case. Pauling never lost his superstitious belief in vitamin C's magic curative properties, though.
Walter Jung was an analog electronics genius. I would gratefully take electronics lessons from him any day. But if he claimed to have seen a unicorn in his backyard, I would want to see photographs, video, fur, scat, and a DNA analysis showing a hitherto unknown creature before I believed in his unicorn.
Magic capacitors are in the same category. There is a simple objective way for true believers to show that magic caps work, which doesn't involve magic ears, faith, appeals to genius, and unverifiable subjective experiences.
I explained that method in an earlier post in this thread. The equipment required is reasonably simple. Any number of research labs could easily do the experiment. In fact, plenty of research labs have studied capacitor behaviour - some caps are indeed better than others, but only at very high frequencies (far into the RF band), or at nearly DC (far below audible frequencies.)
If the magic caps work, why hasn't anyone done the experiment, and published the data? Play Beethoven or Thin Lizzy or Guns 'n Roses or Diana Krall, or all of them, through two RC networks with identical time-constants; one with a magic cap, one with a plebian plane-Jane cap. Subtract one signal from the other, amplify the result, show that there is a difference, and that the difference signal is no more than, say, 40 dB below the original signal.
Because there is no Yeti-DNA in sight, I don't believe in Yetis. If any number of celebrities swore they'd seen Yetis, I still wouldn't believe they existed until I saw the DNA analysis results.
It's the same story with magic caps - it doesn't matter how many audio celebrities say they heard the sound of the magic Yeti-capacitor. With no objective data showing magic audio properties of very expensive caps, I don't believe in magic caps.
-Gnobuddy
How about the Cyril Bateman series of articles on capacitors?:
Cyril Bateman's Capacitor Sound articles | Linear Audio NL
I think it shows quite objectively the difference between capacitors.
"Around the world, professional "golden ear" audiophile experts have praised the sound quality of systems using the VAR series over the exact same systems using other precision resistors."
Vishay Precision Group (VPG) Introduces 60 W, 0.05 ppm/Deg. C TCR Resistors for High-End Audio Applications
During the 1960s tens of millions were blessed by music created with antiquated equipment, in mono, and played through 19 USD (today's dollars) Japanese transistor radios.
The artists, record companies, producers, and radio stations also made tens of millions in the process.
The only "golden ear" you really need is one similar to Phil Spector's, if you want to make some money.
I do appreciate the scientific approach, however, of Bateman's "Capacitor Sounds 4 - capacitances from 100 nF to 1 mF."
It appears from the graphs of most components, distortion and noise differences occur exponentially at either end of the spectrum, or both.
Virtually nobody will hear the difference between a ceramic disk cap and a polypropylene. There are too many variables starting with the way an instrument is played or a recording is produced, and ending with room acoustics. Players may very well be compensating for shortcomings in the instrument, amps or speakers, unbeknownst to them.
Even knowing this, I would still use foil resistors and PP caps just for "peace of mind". Yet I hate "tropical fish" caps, even though I never used one. Go figure.....
Vishay Precision Group (VPG) Introduces 60 W, 0.05 ppm/Deg. C TCR Resistors for High-End Audio Applications
During the 1960s tens of millions were blessed by music created with antiquated equipment, in mono, and played through 19 USD (today's dollars) Japanese transistor radios.
The artists, record companies, producers, and radio stations also made tens of millions in the process.
The only "golden ear" you really need is one similar to Phil Spector's, if you want to make some money.
I do appreciate the scientific approach, however, of Bateman's "Capacitor Sounds 4 - capacitances from 100 nF to 1 mF."
It appears from the graphs of most components, distortion and noise differences occur exponentially at either end of the spectrum, or both.
Virtually nobody will hear the difference between a ceramic disk cap and a polypropylene. There are too many variables starting with the way an instrument is played or a recording is produced, and ending with room acoustics. Players may very well be compensating for shortcomings in the instrument, amps or speakers, unbeknownst to them.
Even knowing this, I would still use foil resistors and PP caps just for "peace of mind". Yet I hate "tropical fish" caps, even though I never used one. Go figure.....
Thank you for the information and the link! It was quite helpful.
A preamble: Thousands of carefully conducted double-blind listening tests were conducted by trained researchers during the era when Hi-Fi was still a science, before it degenerated into superstition. The results were then statistically analyzed to verify if they were the result of random chance, or not.
As far as I know, after all those experiments, nobody has ever shown that people can detect less than about 1% - 0.5% THD. And that is only when listening to pure sine-wave test signals. (When listening to music, much higher levels of distortion are completely undetectable.)
Allowing a big fat safety margin, researchers concluded that THD below 0.1% would be the gold standard for Hi-Fi. Any device with such low THD has inaudible distortion - we can't hear any distortion at all, so the device is audibly perfect.
I clicked randomly on one of Bateman's papers from the link dotneck335 kindly posted. Bateman proudly writes about test data from an evil ceramic capacitor: it produced 0.00074% distortion!
This is just completely ridiculous. The measured distortion is more than a hundred times below the Hi-Fi gold standard of 0.1%, and nearly 700 times lower than the lowest distortion people have ever demonstrated the ability to actually detect.
Bateman then applies a mix of AC and DC to the cap to see if he can provoke more distortion. He did, to the tune of - wait for it - a whopping 0.002%. That's fifty times below the gold-standard 0.1%, and 250 times lower than the smallest distortion people have actually been able to hear.
So Bateman's work shows exactly what I've been saying: feel free use the cheapest ceramic caps with confidence in audio circuits, because their performance is audibly perfect; the microscopic distortions they produce are fifty to a thousand times below anything that human beings can actually hear.
Really, where else except in electronics can you get a wonderful bargain like that? A perfect component, for pennies? We are so lucky!
The only audibly flawed components in a modern audio chain are the microphones used to make the recording, the loudspeakers used to reproduce it, and the listening room itself. Even halfway-decent audio electronics has been audibly perfect for decades; that's why I buy 20-year-old Hi-Fi receivers from thrift stores, but put money and effort into buying the most accurate speakers I could afford (and they are still audibly flawed.)
-Gnobuddy
Attachments
Could you provide a link to those "tests"? I'm not at all sure what measurement differences are audible....I once compared two professional power amplifiers (a Crown and a Studer) in a studio setting, both rated at less than 0.1% THD. I fully expected to not hear a tinkers dam's worth of difference. Boy, was I surprised! One sounded MUCH better to my ears than the other! Perhaps this was not a classic scientific research test, but it did open my ears to the fact that not all audible differences are measurable, and not all measurable differences are audible.
Go to the technical reference section of your local big university library, and you will probably find dozens of dusty old peer-reviewed science and engineering research journals with reports on such listening tests.
I've read dozens of these during the years when I was actively interested in Hi-Fi, between roughly the ages of 15 and 25.
There were hundreds of such listening tests conducted, done over many decades, by big research labs like Bell Labs, Phillips, and many other audio companies as well as independent researchers. A huge body of knowledge was accumulated, and vast numbers of research papers published. That knowledge was then used to set the standards for Hi-Fi, and to manufacture audio components that achieved those standards, or exceeded them when possible.
Later some of those research results got condensed into books, and incorporated into Hi-Fi specs that went beyond research labs and became available to the public. That's where things like the once commonly quoted "20 Hz to 20 kHz, +/- 3 dB" frequency response standard came from - that wasn't perfect, but it was very, very good, and it would be hard to do audibly better.
All of this was pre-Internet, and I have no idea how much of all this research data ever made it to the 'Net. That is one of the bad side-effects of the Internet - vast amounts of valuable information are completely forgotten, because nobody bothers to go look up the books and magazines that originally contained it.
I can do a Google search to see what, if anything, of all this stuff has made it to the Internet. But it's Friday night, I'm tired from a long hard work week, and I have no interest in doing it right now.
That's the trouble with uncontrolled subjective evaluations. For starters, the test wasn't double-blind. It wasn't done on the basis of at least 30 listening tests (to be statistically significant, you need a large number of tests.) It wasn't analyzed to see whether your conclusions were statistically significant.
There are other things I don't know about the circumstances - such as, was one amp driven into clipping? Were both amps switched into the exact same pair of speakers? Were both amps EQ'd absolutely flat? Were both amps set to generate *exactly* the same SPL? If not, any of these things could have led you to prefer one over the other.
All this means that your subjective experience cannot be taken as an indication that 0.1% THD is audible. In fact, as I mentioned earlier, there is a vast body of evidence that this is not the case.
We cannot trust our senses. It has been shown, for instance, that when people compare two loudspeakers in plain sight, there is a strong tendency to believe that the larger one sounds better. Similarly, if the prices of two products are known, people will prefer the more expensive one. All of these placebo-effects have to be accounted for, otherwise we just fool ourselves.
(I've fooled myself too, with a pair of supposedly identical speaker systems that had about a 0.3 dB difference in measured frequency response (I had access to professional measuring equipment at the time.)
I could clearly hear a difference, but the textbooks said I shouldn't. So I called in two friends to put me through a series of double-blind listening tests. After 25 double-blind listening runs, the difference I was convinced I was hearing evaporated into thin air; statistically, my results were the same as flipping a coin. Humiliating, but very, very enlightening: that experience taught me never to trust crudely done subjective evaluations.
The latter is certainly true, as our measurement equipment is far more trustworthy than our ears. For example, a fifty-cent Panasonic WM-61a electret microphone has a much flatter frequency response than the human ear.
The former simply isn't true. There is nothing magic about human hearing. If there is an audible difference, it can be measured. If it can't be measured, you're doing the wrong measurement.
We live in a world where "fake news" and superstition is increasingly taking over. It is hard to know what to trust on the Internet, where everything is temporary and mutable, and yesterday's myth becomes today's "fact". But the old research journals and engineering textbooks live on, at least some of them. That's where you can find real peer-reviewed information, based on the real scientific method, conducted in real research labs, by real researchers, with real educational qualifications, using real research-grade equipment.
-Gnobuddy
So, let's see........I think that we pretty safely assume that ANY halfway-decent solid state power amplifier manufactured in, oh, say, the last 40 years or so has less than 0.1% THD. So what you are saying is that ALL POWER AMPLIFIERS SOUND EXACTLY THE SAME. I guess I just don't believe that, and I doubt that most on this Forum would agree to that. I am really not sure what measurement would account for differences, but I don't believe they all sound the same.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.