So where's the part about building a solid state amp?
I'm inclined to just start designing one myself, TBH. 🙂
A "musical instrument" amplifier might be more inclusive to "new blood" as well. Less of a p** contest (not pointing fingers) to see whose simulation can deliver the most/least [irrelevant parameter] that flies in the face of actual listening experiences, but I digress.
I'm inclined to just start designing one myself, TBH. 🙂
A "musical instrument" amplifier might be more inclusive to "new blood" as well. Less of a p** contest (not pointing fingers) to see whose simulation can deliver the most/least [irrelevant parameter] that flies in the face of actual listening experiences, but I digress.
I wouldn’t say moot. For a given amount of digital hash running around on the PCB and getting into everything, would you rather have 33 volts of analog signal swing available to “digitize” or only 3.3?
And I don’t know about the rest of you all, but “digital noise” sounds different to me than the hiss you get from an old school amplifier that has a limited BW and limited amount of NFB. The way it colors the noise spectrum is different. To truly model an old analog tube amp you’d have to inject noise, but it wouldn’t be the same noise your sound card generates.
Here, the “solid state amp” is assumed to be trivially easy, with the bulk of the work being accomplished in the digital domain ahead of the amplifier. Seems to me to be a better solution than forcing transistors to do things they aren’t necessarily good at.
Not an argument - just an observation. Many solid-state low voltage guitar circuits clip harshly, while the 3.3V Fender Mustang Micro does not.
As far as dividing down the guitar signal before it ever gets into the amplifier input stage, as I wrote earlier, "This is traditionally thought off as a bad thing to do, as it makes signal/noise ratio worse - but maybe it is the right solution for solid-state amps running on low voltage?"
If there is something to be learned here, it is that modern solid-state devices have such a low noise floor, that you can probably get away with dividing the incoming guitar signal down at the input, and still have adequate S/N ratio. And doing so, avoids problems with overloading low-voltage analogue and digital stages.
I will add that a vacuum tube is handicapped by fundamental physics compared to , say, a JFET; because of the high cathode temperature of a vacuum tube, it generates roughly twice as much thermal noise as a room-temperature FET. This is because the thermal energy of electrons is directly proportional to the temperature (1/2 k T, where k is Boltzmann's constant and T is the absolute temperature, i.e. temp in Kelvin.)
And this is for a mathematically ideal vacuum tube (which doesn't exist). A real tube has additional noise mechanisms which make it even noisier. Low frequency flicker noise, for example.
Some back-of-the-envelope calculations I did a while ago suggest that a well chosen JFET is probably the optimal low-noise input device for an electric guitar. BJTs are noisy due to the high source impedance of an e-guitar, MOSFETs are noisy, period, and vacuum tubes are handicapped by the high cathode temperature.
It may not matter much in real life, though. History demonstrates that vacuum tube input stages have been sufficiently quiet to make virtually all the great e-guitar music of the past 90 years. We often forget that there were there were electric guitars in 1932.
-Gnobuddy
Certainly some JFETs were best in class concerning noise in a hi-z input circuit. On the other hand the noise voltage density of a 200~500kOhm input is in the ballpark of 60nV/sqrt(Hz) and nowadays there are FET-opamps with less than 10nV/sqrt(Hz) available. An ADC powered with 3.3V may accept 3Vpp input i.e. 1Vrms. 16bit resolution yields about 96dB dyn range corresponding to an input referred noise close 15uVrms. For input voltage peak >3.3V a buffer stage powered with 9V or more followed by a pad reducing level about 6db should do the trick.
But for a first iteration on the way to a DIY guitar amp I propose to keep things simple, programming a DSP is the advanced course.
That was exactly my point and none other.
Including my later clarification.
33 Volts signal?
We are talking Guitars here.
This thread or at least what gnobuddy mentioned is about a 3.3V fed Digital system, in which case "you have what you have"
Yes, and one has to make absolutely sure the input voltage from the guitar never, not even for a moment, exceeds 3 Vpp.
One diyAudio member demonstrated his ability to produce 10 Vpp straight from his guitar, and proved it with 'scope captures. This is undoubtedly rare. But I didn't think it was even possible.
What this means is that if you power just about any SS guitar amp from a 9V DC source, its input can be driven into clipping. Doesn't matter if it is analogue discrete, analogue op-amp, or the input to an analogue-to-digital converter.
One interesting thing about a JFET in source-follower configuration, is that it will gracefully accept an input signal with peak value somewhat larger than the quiescent DC voltage at the source. Negative peaks will be very gently squashed, with no audible harshness to my ears.
I threw together a quick LTSpice sim (first attached image). In LTSpice, the 2N3819 biases up to +2.56 volts DC at the source. One might expect that any input signal with a negative peak value more than 2.5 volts would cause the output to clip, but the nonlinearity of the JFET near cut-off prevents this from happening. In fact the circuit will accept (-3 V) peaks at the input without any clipping at the output, just a gentle compression of negative output peaks.
If we take the extra trouble to bias the JFET gate positive for optimum input headroom, a JFET powered by only 9 Vdc will gracefully accept a 10 Vpp input signal, without harsh clipping. Both positive and negative peaks are gently squashed, rather like Leo Fender's accidentally-misbiased DC coupled cathode follower in his Bassman amp.
This is the only solid-state input configuration I have come up with that copes so well with unruly guitar input signals.
As I said before, I think a 10 Vpp guitar signal is very rare, though we now know it is not impossible. But the point is that a JFET input stage copes rather gracefully with excessive input signals, even if they are not quite as big as 10 Vpp. The same cannot be said of op-amps, or LTPs made with bipolar transistors.
The source follower (common drain amp) in the attached images doesn't give you any voltage gain (nominally a wee bit less than 0 dB), but it does avoid harsh clipping. The output could be padded down to keep subsequent gain stages from clipping harshly.
Adding a drain resistor lets us get a little voltage gain out of the JFET - maybe 6 dB - 12 dB with older JFETs. Input headroom before harsh clipping is no longer a full 10 Vpp, but still "big enough" for practical use.
Subjectively, and without any empirical data to prove it, I repeatedly found that using a JFET input stage, with typical voltage gains in the range of 6 dB - 12 dB, running on +18 Vdc, substantially reduced audible harshness in my attempts at building SS guitar preamps.
But, as you say, not everyone is bothered by that sort of harshness, so not every guitar player would care.
Indeed! IMO, the (audio) noise floor from A/D converters stopped being a real issue a long time ago. When Sony & Phillips combined forces to develop the CD player, they picked 16-bit numbers for very good reasons. Dynamic range is beyond anything necessary, and distortion is far below what the human ear can detect.
I bought my first CD player in 1990 or early 1991, as soon as I could scrape together the money from my tiny student scholarship. At the time I also had a record player, and a nice TEAC cassette deck with Dolby C and Dolby HX Pro headroom extension.
Using the CD player as source, I made a recording with metal tape in the TEAC, with both Dolby C and HX Pro engaged to maximise dynamic range, and recording levels carefully set. On playback, I could still hear the hiss (noise floor) from the taped recording.
Contemporary lab tests showed the noise floor from a good cassette deck under these conditions was at about (-60 dB). Quiet, but audible. More audible than I had realized.
As for the record player, vinyl surface hoise (hiss) was also at about (-60 dB), but of course there were the inevitable clicks, ticks, and pops from dust particles in the LP grooves on top of that. Noise from the vinyl was obtrusive, and annoying. I have a good ear for pitch, so the inevitable wow near the end of the record (from imperfectly punched centre holes in the LP) was yet another perennial annoyance.
And the CD player? Blissful perfection. NO audible noise, and I had a good pair of young ears back then. Also no wow, no flutter, no pops, no ticks, no gradual quality degradation with every replay, no audible amounts of THD, and a ruler-flat frequency response beyond the highest frequencies a human ear can hear.
And that was with "only" 16-bit A/D converters.
Digital audio recording and reproduction reached audible perfection a long time ago. Digitally simulated guitar amplifiers, no such luck - they only got close very recently.
But the $76 Flamma Preamp is a mighty contender these days. Why bother to build an analogue SS guitar amp unless you can at least match the sound quality of the Flamma? And that is a very high hurdle to clear.
No doubt. We had horrible-sounding digital models of tube amps literally for several decades, and those were developed by professionals. Even with sufficient processing power available, it was clearly quite a difficult job to develop really good-sounding DSP models.
There is no guarantee at all that amateur DIY hobbyists can develop good-sounding digital models of tube amps.
-Gnobuddy
Attachments
Since I am the only one who posted any simulations, it's pretty clear who you are aiming your insults at.
My "actual listening experience" is that every SS guitar amp using op-amps or other high-gain SS input stages sounds harsh. I do not know the reason, but suggested one plausible one: the low overload margin and harsh clipping inherent with high-gain, or high-feedback, or both, solid state gain blocks.
Then I posted a simulation that demonstrates one example, a circuit favoured by one of our members, but which sounded harsh to me when the builder posted a sound clip. The simulation shows that I'm not just making stuff up - it demonstrates how that particular circuit (and many, many other SS preamp designs) copes very poorly with the normal dynamic range of picked guitar notes.
You are, of course, welcome to your opinions. But a little civility to go with them would be nice. We are all trying to cope with very difficult circumstances these days. Last I checked, one out of every 375 Americans has now died of COVID-19, based on official figures for the current death toll and US population, and that number changes for the worse every day. Most of the rest of the world is not in quite as bad a situation, but it is more than bad enough.
If ever there was a time to try to be kind, this is it.
Please do. And please make it less-horrible than the thousands of horrible-sounding SS guitar amps that have already been inflicted on our bleeding ears over the last fifty years. 🙂
There is a long thread on diyAudio about the last SS guitar amp that I myself put significant development and construction time into. That one was designed and built by me as a gift for a senior citizen friend with a physical handicap, who is unstable on his feet and walks with two canes most of the time.
For those reasons, the amplifier had to be relatively small, and relatively light. As well, my budget was tight, so it had to be built with very little money. All three of those things dictated that the amplifier had to be solid-state, and I had to be creative about what parts I used.
I ended up with something that worked very well for clean guitar tones, with no audible harshness to my ears. In order to achieve that, I had to use a JFET input stage, as well as invent a "de-nastification filter" to take out harshness.
Since the intended user tends to get confused by multiple controls and tends to mis-set them, I also designed that amplifier to have no settings that produced bad tone. That involved using a very different type of tone control circuit, one that works in conjunction with the de-nastification filter to provide good guitar clean tone at all control settings.
Maximum voltage gain of the preamp was also deliberately limited, so that it is literally impossible to produce harsh overdriven sounds from this particular amplifier with a guitar plugged directly into the input.
diyAudio moderators split off my posts on this amp from a different thread in which I had been posting, created a new thread, named it something I would not have chosen, and stuck my posts there. Here is the thread, from late 2017/ early 2018: https://www.diyaudio.com/community/threads/tube-emulation-eq.314730/
If you are in a scoffing mood, I'm sure you will find plenty to scoff at there, as well.
A good thought. By all means, pursue it. It has been tried many times before, and some of the earliest Internet electronics bulletin boards and online forums were based around this idea, going back well over three decades; for example, the now (virtually) defunct AX84 forum, whose entire raison d'etre was to collectively design and build good DIY tube guitar amps.
The simple fact of the matter, is that most of us who tried it, have found that it is easier to get good guitar sounds out of simple tube circuits, than it is to get good sounds out of simple solid-state circuits.
"New blood" might have less fussy ears, which might help your cause; young guitarists often grew up with very heavily distorted e-guitar music, and are more likely to be satisfied with harsh clipping from back-to-back solid state diodes or similar crude-sounding devices. Line 6 has sold lots of harsh Spyder amplifiers to that demographic for the last two decades.
Then again, e-guitar itself has fallen out of fashion quite a lot lately, and DIY electronics even more so, particularly the analogue flavour. I have my doubts you will succeed in attracting much new blood to this forum via a DIY SS e-guitar amplifier.
-Gnobuddy
I'd like to try it actually. Transistors have plenty of X vs Y curves displaying soft exponentials or quadratic functions. It would be a matter of capturing that performance, as well as good interaction with the source and output.
"Too clean" is a legitimate concern by people who seem to be ignored necause they are musicians, which is the irony of ironies.
Gnobuddy --
The JFET behaviour intrigues me. Even with non-clipping signals, I have a hunch that low level noise will be better behaved if the extremes are also taken care of. After all, gaussian noise has a statistical spread, so the noise will contain all the 'dirt' on the amplifier does at every level.
A bit surprising to read that. Could you pls post some links. It is funny that after 5 decades playing my strat I prefer my old acoustic guitar for about 2 yrs. And I hear similar from the collegues. Is this a global trend after all?
Btw, I observed P.U. voltage peak of 5~10Vpp with the humbuckers of my then Gretsch Semi-Acoustic. Moreover I found out these peaks to be quite short and clipping them does not make that big difference you claim. Just my opinion.
The tricky bit is that the "scaling voltage" in the exponential is only 26 millivolts. In other words, every 26 mV increase in Vbe causes nearly a tripling of collector current!
This follows from the Shockley equation that describes an ideal BJT transfer function. Modern silicon small-signal BJTs accurately follow that equation over several orders of magnitude of collector current.
For our present purposes, this enormous voltage sensitivity can be a problem.
For example, if you have a common-emitter stage with the collector biased to half-Vcc, doubling the collector current will result in saturation (Vce=0), and a flat-bottomed output waveform. From the Shockley equation, it only takes an 18 mV positive peak in the input signal to achieve this!
To place that in context, we know from 'scope captures that a typical guitar can put out 1500 mV positive peaks under reasonable playing conditions (hard strumming). Even gentle playing may put out 100 mV peaks. Yet, our simple common-emitter stage is already hard-clipping at only 18 mV positive peaks!
Interestingly, in the very early days of semiconductor guitar devices, this approach was used more than once. The "Dallas Rangemaster" schematic is one good example ( https://www.electrosmash.com/images/tech/rangemaster/dallas-rangemaster-schematic-parts.png ).
The Rangemaster consists of a single common-emitter germanium transistor amplifier stage, with no negative feedback anywhere. Input impedance from the leaky, low-beta Ge transistor would have been far too low to be acceptable today, and the much-too-small input capacitor stripped out most of the bass frequencies from the guitar, while the much-too-low input impedance stripped out a lot of the high frequencies as well.
Brian May (of Queen) is one famous guitarist who used a Rangemaster. There is a current thread elsewhere on diyAudio where there has been some discussion of May's famous guitar sound, and where it came from.
Those leaky old Ge transistors didn't follow the Shockley equation as closely as the better transistors of today do. They don't saturate as hard as modern Si devices do, either. Those things may have had something to do with the fact that the Rangemaster didn't sound harsh in Brian May's recordings.
Typical JFET curves are spread over a considerably larger voltage spread than BJT curves. This means they tend to have low voltage gains too, as device transconductance is far lower than a BJT.
These are qualities that pushed the FET out of favour for small-signal voltage amplification a long time ago. But for an e-guitar amplifier, these "weaknesses" might actually be assets.
I have never heard an FET-based SS guitar amp that sounded just like a good tube amp - but they also tend not to sound harsh, and IMO it is fairly easy to get some relatively sweet-sounding distortion out of an FET or two.
There is a real conundrum here. Musicians, by nature, tend to be superstitious about their instruments and gear, and so scientists and engineers have to take their input with a grain of sand, or do careful controlled, double-blind experiments so they don't end up believing in musical Loch Ness Monsters.
On the other hand, MRI scans of musican's brains literally show that they have been significantly re-wired to favour auditory inputs over other sensory inputs. Human brains are quite flexible, and years of focusing on music can and does change them. Great musicians certainly hear things the rest of us cannot.
It makes perfect sense that musicians tend to be superstitious and subjective. If the musician doesn't see rainbows and hear angels when she plays her instrument, her audience certainly is not going to see and hear those things.
Me, I have a foot in each camp. My education is solidly in objective science. As an amateur musician (though definitely not a great one), when I play music, I do my best to switch to the "rainbows and angels" subjective, illogical, non-critical, right-brain-dominant, state of mind. It's the only state of mind from which you can create good music.
Concerns over superstitious beliefs aside, the "clean tone is too clean" complaint is no mystery, and we don't have to invoke superstition to explain it. A single half-12AX7 triode in a simple common-cathode amplifier stage can generate 5% low-order THD, and decades of experiments in the Hi-Fi world have shown that a significant number of people can hear as little as 1% THD on a pure sine wave audio tone.
5% THD is greater than 1% THD. If people can hear 1% THD, they can certainly hear 5% THD. Q.E.D - "and thus it is demonstrated"!
An e-guitar signal isn't a pure sine wave, not even close. It contains lots of harmonics, even if you play a single string. This means slight nonlinearities in the amplifier also generate intermodulation distortion (IMD), which is even easier to detect than THD alone.
So there's no doubt that people with a good pair of ears (no magic golden ears necessary) can indeed hear the slight change in guitar clean tone that results from going through, say, four cascaded stages of vacuum tube amplification, each one adding a few percent THD.
Going through four or more tube stages is quite typical of many of Leo Fender's tube guitar amp designs. Newer, higher-gain tube amps may have a lot more stages than that. So we're talking about maybe up to 20% THD in a guitar "clean" tone!
Some years ago, I spent quite a lot of time coming up with a design for a 2-watt tube guitar amp, the hope being to achieve good tone at low enough SPL to be usable in an apartment. Without an oscilloscope at the beginning, I tweaked my circuit by ear until I managed to get a good clean tone. When I bought a 'scope later, I was surprised to see around 20% THD, low order, mostly 2nd harmonic, from my carefully tweaked design. This, remember, was what my ears wanted in order to get an appealing clean guitar tone. At the time, I was very surprised by how much distortion still sounded "clean".
That was a low-power amp at relatively low SPL. Our ears are more sensitive to THD at somewhat higher loudness levels, and it is quite likely that an amp as loud as, say, a Fender Twin, needs much less than 20% THD to audibly colour the clean tones.
No wonder op-amp stages with 0.001% THD sound "too clean"!
-Gnobuddy
My frustration wasn't aimed at you at all, or anyone in particular. If anything, it was just a vibe I picked up from hanging around too much 'hi fi' stuff. I kind of want to get away from all that.
Probably all of that. Negative feedback is a double-edged sword. How often have you seen development driven by marketing gimmicks instead of engineering? One glance at Fletcher Munsen curves will tell you that THD is deeply flawed. Not to mention global feedback, routing the output back to the input with a non-linear, inductive, full duplex voltage-generating speaker system plugged in? What could possibly go wrong? This is probably one reason for the enduring appeal of tube amps with their low gain and low feedback.
I'm sure they're just doing their best. This is also the closest to an "active speaker" forum, carefully tucked away because holistic thinking is hard.
There seem to be plenty of soft distortion curves in the datasheets, but I don't know the nature of the problem just yet. Maybe it's a matter of zooming in on those inflexion points, where the gain sweeps a nice wide arc?
Well it's a matter of descriptive language as well. Is 'pure' or 'sweet' cleaner than clean? And why cleanliness? My earlier idea was that what musicians (I also fall into the amateur category) were hearing was somehow being translated into an image of the 'chore' of cleaning something up. That's pretty peculiar and to me it communicates in a subtle way that they can hear the residue left behind by linearisation techniques, like a detergent with a friendly aroma to say "this place has recently been cleaned".
Well, for starters, when was the last time you heard something that was recognizably electric guitar on a recent popular hit song? How many kids in your area still have garage bands? If there are any, how many include electric guitars? How many guitars do you see students carrying around at the local high school? How about the local college campus?
During the last 20 years I've worked, full-time, on three different college campuses, with tens of thousands of students on each one. In fact, the first two campuses together had a combined enrollment in the vicinity of 100,000 students total. Yet, I have only seen someone carrying a guitar on campus a handful of times - and usually that person turns out to either be a music major, or someone hired to perform somewhere on campus.
In 2017, the Washington Post published an article titled "While My Guitar Gently Weeps", subtitled "The slow, secret death of the electric guitar, and why you should care". The article is now behind a paywall, but it was publicly accessible for several years. The paywalled article is at https://www.washingtonpost.com/graphics/2017/lifestyle/the-slow-secret-death-of-the-electric-guitar/
Fortunately, the Internet Archive does have a readable copy of the original article here: https://web.archive.org/web/2017062...f-the-electric-guitar/?utm_term=.f8162582740e
I think the article makes a lot of compelling points. There are no "guitar heroes" today, Taylor Swift has sold more guitars recently than any other public figure, most remaining famous guitarists are now old men with white hair, there are almost no guitar sounds heard on contemporary hit songs, et cetera. Also, the days when you could become popular and earn a good income if you knew how to play guitar, are long gone. Both the death of the music industry, and the death of guitar music itself, have contributed heavily to this.
If we look back at history, other instruments have had huge surges of popularity, only to fade away after a while. For instance, America and Canada went crazy for mandolins, starting from roughly 1880, and continuing until WWI. Then that craze died away completely.
Much more recently, until roughly the 1980s, we heard saxophones on a lot of popular music. Some saxophone players became wildly popular public figures, just like electric guitar heroes two decades earlier. Remember Candy Dulfer, for instance?
But what happened after that? For the last forty years, saxophone players and their instruments have both disappeared. I cannot even remember the last time I saw a live music performance that included a saxophone. In all the hundreds of music hams I've been to during the last decade, I have never once seen a single saxophone.
Of course there are are still mandolins and saxophones out there, just like there are still horses and horse-buggies out there. But they are no longer in the public consciousness. Mandolins cling to life in a few folk / America genres, while saxophones seem to be mostly gathering dust in closets. There are probably more children taking (horse) riding lessons, than taking saxophone lessons.
Let's not even talk about classical orchestral instruments. Classical music is completely dead in this continent. Schools no longer have orchestras or musical instruments. K-12 students no longer take "band" class. Local junior colleges have closed down their music departments for good. I often see ads for free (acoustic) pianos in the local free newspapers (Craigslist, etc).
One free piano ad even added some emotional blackmail. The ad stated that the owner of the piano had been running his add for weeks already, needed to move to a new apartment soon, and was going to chop his piano into pieces with an axe, and stuff it in the dumpster, unless someone came and took it away within the next two days. You literally can't even give away a piano around here.
I think the electric guitar is well down the same road to extinction. Few young people find its sound inspiring these days, and you are much more likely to be popular with your peers if you rap or DJ, than if you play e-guitar. Electric guitar is now grandpa's instrument. Not to mention, if you live in an apartment, you can't use the most notorious capability of e-guitar: you can't play loudly!
Personally, I suspect the decline in socialization has a lot to do with the decline of the e-guitar, and that has been going on for decades. Electric guitar is not a great-sounding solo instrument. It is at its best when playing with other musicians, as the Beatles demonstrated so vividly sixty years ago. But most young people no longer "get by with a little help from their friends"; instead, they get by with a little help from Tik-Tok, Tinder, Instagram, You Tube, and so on.
If your "friends" are just digital avatars on your phone screen, you can't make music with them, and e-guitar is not an inspiring instrument to play alone, all by yourself, in your parent's tiny apartment, surrounded by short-tempered neighbours.
Acoustic guitar is doing a little better. It is easy to learn to play a few chords, it sounds passable all by itself, and if you are a young girl who is a Taylor Swift fan, you may want to buy an acoustic guitar just to look like she does on stage. Learn three chords, and you can sing along with most of her songs. Also, Country music is still huge, at least in North America, and acoustic guitars are still a part of that genre.
I'm not sure how the pandemic has affected the e-guitar. I suspect a lot of people have bought an electronic keyboard, drum set, acoustic guitar, or electric guitar while they were stuck in lockdown. I also suspect that many of the buyers hardly ever touch the instruments they bought, and most just sit quietly in their cases in the back of the closet. These instruments play a role more like a cheese-grater than the way previous generations thought of their musical instruments; the guitar or keyboard is no longer a major part of your everyday life, instead, like the cheese grater, you pull it out for a few minutes when it is needed, then you put it away.
It's very interesting that factories in China continue to pump out a stream of electric guitars, a lot of them quite good instruments at surprisingly low prices. I think this may continue to fuel the "cheese grater" end of the guitar-buying spectrum. If you can get a playable e-guitar for $100, why not buy one and stick it in the closet; maybe you will actually use it some day.
Here is my favourite Taylor Swift "performance":
-Gnobuddy
Just as DSP became more prevalent in guitar amps and effects, it also had a big impact on the technology of creating ''music'' in general. You could piece together samples and loops of drums, piano, guitar and vocals, all on your ''home studio'' laptop. Need a brass or string section? Just download the app for it and get the sample pack. As is evident on a lot of those types of recordings, it can sound very sterile, mechanical, electronic .. but to a large extent the technology may have steered many young people into creating that style of music. Very little requirement for playing an instrument, it comes down to being skillful with their software.
Thanks, one more data point. You are only the second guitarist I have heard from who reports such a high voltage from your guitar, though 3 Vpp - which is about 10 dB less - seems very common.
I don't "claim" this is the case, i.e. I have no proof. It is only a guess on my part. It would take controlled, double-blind listening tests with a large number of subjects to prove it.
The initial transient at the start of every guitar note occurs because the moving guitar pick (plectrum) hits the string, traps and moves it far off-centre, then the string slips free, and starts vibrating at its resonant frequency (and harmonics). The initial transient has a lower frequency than normal string vibrations (the pick moves more slowly than the vibrating string). The initial transient is also observed to often have far more peak amplitude than the rest of the note, depending on the playing technique being used.
Now let's think about the initial transient in context, i.e, mixed into the actual notes from a guitar. The Guiness Book of Records certified one guitarist - Sergiy Putyatov - as capable of playing 27 notes per second: https://www.guitarworld.com/news/video-sergiy-putyatov-fastest-guitarist-world
At his record-breaking playing speed, then, Putyatov's notes each last 37 milliseconds. People have no trouble hearing individual notes at that speed. You can verify that for yourself - here is Putyatov playing blindingly fast scales:
Let's consider the lowest note on a guitar in standard tuning, the musical note E2, at 82.41 Hz. One cycle lasts just over 12 mS. (Putyatov can actually play individual notes so fast that the low "E" string on a guitar only has time to vibrate three times before the next note arrives.)
Back to the the initial transient that starts off each guitar note. It is slower than the rest of the vibration; how long does the transient last when low E is picked? I don't have an oscilloscope capture handy, but let's guess it lasts about as long as two cycles of steady-state vibration, i.e. 24 mS.
Now let's say we have a solid-state guitar preamp that clips hard during this initial transient (but only during this transient, for discussion purposes). That means it clips hard for 24 mS at the start of each note.
If Sergiy Putyatov was playing at his record-setting speed on the open low "E" string, then, the first 24 mS of every note is distorted harshly. Each note is only 36 mS long. This means that two thirds of the duration of each note is harshly distorted.
Does it seem plausible to you that, if every note is distorted into a square wave for two-thirds of its duration, nobody is going to be able to hear this?
To me, that is not plausible. We already know we can hear individual notes in a stream of notes coming along at 27 times/second (listen to the video). How can we then believe that we cannot notice if two-thirds of the duration of each note is clipped hard?
Okay. Putyatov playing at top speed on the open "E" string is a pretty extreme hypothetical scenario. What about guitarists with slower fingers, playing slower music, like the rest of us?
Look back at the simulated waveforms in my post #26 in this thread. I simulated one note held for a full 700 mS, which is a very long, slow note in musical terms. LTSpice thinks the output of the LTP stage clips hard for well over 200 mS - more than one third of that very long, slow note!
Two hundred milliseconds is a very long time in the world of music. There is no doubt we can hear hard clipping lasting for 200 mS at a time. So even the slowest of slow guitarists, playing the slowest of slow music, is going to be clearly able to hear the long period of hard clipping at the start of every guitar note fed through a BJT long-tailed pair.
Okay. We've looked at extremely fast playing, and extremely slow playin. What about the middle? What about the top speed of more average player?
I'm a middle-aged amateur guitarist, and not a particularly fast one. I just checked, and today, without a warm-up, I can comfortably play sixteenth-notes with the metronome set to 175 bpm (175 quarter notes per second). I was playing a simple ascending four-note chromatic scale for this exercise.
I might do a bit more with my hands well warmed up, but today I sounded erratic and was missing some notes when playing sixteenth-notes at 180 bpm, so let's go with 175 bpm.
The metronome was set to 175 quarter notes per second, but I was playing sixteenth-notes, which are four times shorter, so I was playing 700 notes per minute. That's 11.667 notes per second.
There are 60,000 milliseconds in one minute. That means each note I was playing lasted (60000/700) seconds, or just under 86 milliseconds.
Note that Mr. Putyatov plays scales, crossing across multiple strings of the guitar, more than twice as fast as I can play multiple notes on the same string. What I'm capable of doing is nothing special. It's like running the 100-metre sprint in 25 seconds, while the world's fastest athletes run it in just under 10 seconds. You don't need to be an athlete to run 100 m in 25 seconds - just about anyone who is not physically disabled can do that.
That said, what we've just found out is that even an ageing amateur guitarist like me, can play fast enough that, if my guitar amp clips hard for 24 mS (the hypothetical length of our initial big transient), then it will be clipping for nearly one-third of the duration of each note I play.
Is it plausible that our ears cannot hear that a full one-third of the duration of each note we play is harshly clipped?
Again, I have no objective hard scientific data that's good enough to stand scrutiny in a peer-reviewed research paper. All I can offer is the thought experiment above. To me, it is not plausible that if a full one-third of the duration of every note is distorted heavily, clipped hard, we still cannot even hear that at all.
The remaining question is about the word "harsh". If your hard-clip a guitar waveform, is the resulting sound "harsh", or not?
Here we are on entirely subjective territory. To me, a square wave is a very harsh sound. You may or may not agree. But, speaking for myself, I most definitely do not want to hear square waves from my guitar.
I think that simple vacuum tube guitar amps often sound better than SS amps simply because of this "hard clipping" problem. Typical small-signal triodes have very low transconductance compared to BJTs, so they are better able to "eat" the full signal from an e-guitar without clipping. In addition, there is no NFB in a simple common-cathode gain stage, so when clipping does occur, it's relatively soft compared to an op-amp or BJT LTP stage.
To be doubly clear, I don't think there is any sort of magic in vacuum tubes. They just happen to work better with guitar pickups. And this is not a coincidence, as guitar pickups were originally designed in the very early 1930s, to work with the only type of amplifying device available then: vacuum tubes. George Beauchamp was, without question, a brilliant man, and he knew what he was doing when he designed the pickup, and the guitar, that became known to the world as the 1931 Ro-Pat-In / Rickenbacker A-22 Electric guitar, aka "Frying Pan".
-Gnobuddy
????
Are you talking about the young woman with the heavy black eye makeup at the beginning of the clip? That is the remarkably mediocre Ms. Taylor Swift herself.
Are you talking about the singing? That is Ms. Swift's voice, after heavy vocal pitch-correcting with Autotune/Melodyne, added background vocals, and who knows how many other studio tricks. Swift can barely sing live - there are plenty of embarrassing clips out there to prove it.
Are you talking about the best part of the clip? The goat? In that case, yes, you are absolutely correct. That is not Taylor Swift, but a goat. 😀
-Gnobuddy
