Yes indeed, the resistor throws away a lot of power that would otherwise go to the speaker and create sound, but it's simple and it always works. I worry about changing the feedback of a power amp, as it's too easy to convert an amplifier into an oscillator. With changing an amp there's a "design phase" with the emphasis on phase and keeping a close eye on phase shift vs. gain.
I'm not surprised you got such a huge peak from an EB0 - the pickup is almost halfway up the string, so the maximum initial voltage as the string is picked/plucked before it settles down to the note. The peak from the bridge pup of a Jazz would, I expect, be a lot less
How about an amp running Class A? Have a cheap Chinese JLH board that cost about $5, Have a pedal as a quick preamp, does not sound bad.
Yeah, why not! A 5W or so guitar amp is great with a good size speaker. Only a few transistors more to do a simple preamp.
I was thinking of doing a Blackface 6V6 amp and drive the SS amp with a buffer coming off the second triode cathode resistor. One channel for the tube side, another for the SS side, into a pair of remore speakers and a set of headphones. Should be interesting to see which ear is the happiest.
FWIW I just hooked up a multi 12ax7 preamp (loosely based on the AMPEG 15W bass amp inc. phase splitter from what I hear - a vintage-ish Mozztronics BP-1) with several bright locations, a parametric mid (and adjustable treble/bass corners) attached to a 60W SS amp and a proper (bass) guitar speaker & box. Lots of good tones available. Not cheap to build: 3 x 12ax7 plus more than a dozen front panel switches and knobs. Plus more on the back panel of the 1RU box - not a lot of empty real estate. But easy to dial in good blues to basic rock tones (it's not high gain ).
Last edited:
Very interesting work. Thank you for finding and posting it.
But the key question is: "What does it sound like?"
My cringing ears still remember that for the last several decades, every digital modelling amplifier and multiFX pedal sounded harsh and horrible, including lots of quite expensive products that sold very well to a specific market segment (guitarists with tin ears ).
IMO it's only quite recently that the tube amp DSP models became refined enough to sound good, and affordable DSP hardware became powerful enough to run the more refined tube amp models.
So it's taken some 25 - 30 years for the industry giants to come up with credible-sounding DSP tube amp emulations. They've been throwing money and smart engineers at the problem literally for decades, and it still took them this long.
Per se, there is nothing uber-complex about learning to write DSP code. I learned to write DSP code in assembly language on Motorola DSP56000 evaluation boards circa 1998. There were no sophisticated drag-and-drop programming tools available to me, and I hand-wrote every line of code. It didn't take long for me to learn enough to create a prototype of what my employer was after - custom FIR filters that flattened the frequency response of a proposed flagship loudspeaker product.
But that was a VERY long way from creating good-sounding models of an entire nonlinear tube guitar amplifier.
I have a spare Raspberry Pi 4 lying around, and if I were in better health I might try dropping that software on it. As it is, I think I'll wait for someone else to do it; I'm not very optimistic about the sound quality that will result. I would bet my two cents that the $68 USD Flamma preamp you bought beats the pants off the RPi+DSP in terms of sound quality...
-Gnobuddy
I worked for him in the late 1990s - it was my first job out of grad school.
Barr was a very, very smart man, who sadly took no care of his health, and who entrusted the running of his company to the hands of buddies who really didn't seem to know what they were doing.
The brand-name lives on today, but the real Alesis declared bankruptcy early this millennium, was dismantled, and saleable assets sold for pennies on the dollar.
-Gnobuddy
He was brilliant, no doubt at all. You are a lucky man to have worked for him. All my amps are fitted with his Spinsemi FV-1.
Gnobuddy - Welcome back! Long time no hear.
Makes you wonder why that is and if it's the same basis as why the original digitization of audio in HiFi was oft perceived as ... suspect.
Perhaps they figured "it's only a guitar; h*** we could use telephone bandwidth and hence, sample rate". Which would result in sub par performance, compared to an analog circuit.
Perhaps it's jitter; "so what if the samples are taken at a er, somewhat discontinuous rate - it's only a guitar - no one will ever hear it"
Then, as component costs dropped relative to what they wanted to spend on parts count, they could up the digital performance to be the same as modern HiFi; 96 khz etc.
Thanks for the welcome! Long COVID has pretty much had me flattened for a while, which is why I've been gone so long.
Real tube amps have a lot of subtle complexities to them. Each tube stage has gentle nonlinearity over some range of input signal amplitudes, and more abrupt nonlinearities if pushed hard into saturation or cut-off. Grid current starts to flow in some regions of operation. Bias shifts occur that alter the nominal DC operating point as signal strength is changed, causing duty cycle modulation when overdriven. In pentodes and beam power tubes the screen grid voltage can be modulated by changes in output power, adding gentle compession due to "sag". There is EQ ("voicing") between every two or three stages, changing the way the overdrive reacts at different guitar frequencies. There's probably other stuff I can't remember (or don't know about).
I don't think any attempt was made to model most of these things in the early modelling amps. I think an attempt was made to model the overall EQ of an amplifier and speaker, and then some sort of crude nonlinearity was bolted on. Not very different than a pair of anti-parallel clipping diodes across and op-amp feedback resistor, and just as harsh and unpleasant.
As we've seen right at the start of this thread, you can get lots of unpleasant harsh clipping in the (analogue!) solid-state input stage, before the signal ever reaches the ADC to be digitized. If that was happening in these older modelling amps, it would be a case of garbage in, garbage out. The signal would already be harsh and unpleasant-sounding even before it entered the ADC.
In the case of the early Line 6 products, there was a really nasty nasal tone to all of them. Speaker / cab emulation gone badly wrong? I don't know, but just that unpleasant nasal honk alone was enough to make the amplifiers sound nasty. And then you add the harsh overdrive on top of that.
It's been a while since I heard a modelling amp quite as nasty as early Line 6 Spyders and Pods, but there are still big-name modelling amps that sound harsh (to me). The best part of two decades after my poor experiences with Line 6 products, I bought a first-gen Katana 50. To me, the Katana has two usable channels: the Acoustic channel (for electro-acoustic guitars), and the Clean channel, which produces the same sort of subtle tubey colouration you get from a good Fender Blackface amp. Millions of happy Katana owners love the higher-gain voices, but they all sound rather harsh to me. And this is Roland/Boss we're talking about - a giant in the industry, with a solid reputation earned over decades.
Is it just me? Well, some months ago I stumbled across a Youtube Channel called "The Studio Rats". They did a video or two on Katanas, and Paul Drew, their extraordinarily talented guitarist, said the same thing I've been describing (the best channel is the clean one). Because he didn't like the sound of the Katana's overdriven tones, Drew left the Katana on its clean channel, and used effects pedals to get his distorted and overdriven sounds.
So it's not just me. Paul Drew spends his days playing through a variety of good tube amps, and he hears something he doesn't like in the Katana's higher-gain amp models, just as I do.
On the plus side, my little Fender Mustang Micro doesn't seem to suffer from the harsh input-stage analogue clipping described at the start of this thread. And it runs on a single lithium battery. The supply rail can't be more than about 4 volts DC, and may be only 3.3V (a voltage used by most contemporary microcontrollers and DSP chips).
How does the Micro manage to avoid harsh input clipping? I don't know, but my guess is that someone in the industry finally addressed that same harsh clipping that's been bothering me for roughly the last 35 years, and simply built in a voltage divider right at the input, before the guitar signal ever hits an op-amp. Bad for signal-to-noise ratio, but at least you can keep away the clipping op-amp harshness before the signal hits the ADC.
On the minus side, to my ears, the Micro amp models are adequate, but not great. My ears think the same is true for their very overpriced Tone Master amps. Fender Corp still doesn't have DSP tube amp models that are quite up to snuff. Rather ironically for a company that gained its reputation by building amplifiers with clean tone, they can't get their own in-house Fender clean tone right - their models sound too thin and too bright to me, lacking that subtle "tubey" warmth.
So there are still plenty of "meh" modelling guitar amps and multiFX pedals sold today, even at price points pushing well north of $1000 CAD. They don't sound as nasty as the old Line 6 stuff, but they don't sound particularly good, either.
And then there's the little Flamma FS06, which still sells for just under $100 CAD, and somehow manages to sound better than some $1500 products. Sure, I would love it if it had separate control knobs for the clean and dirty channels in each model, and a presence knob would be nice, and not every amp model it contains sounds like the real thing. But they all sound good, I haven't heard any nasty noises from it, there are sublimely good tubey clean tones from the Fender and Two Rock models, and very nice progressive, touch-sensitive overdrive from the AC30 and Two Rock models.
All of which points to the simple fact that it's clearly not a simple and obvious job to write DSP tube amp code that actually sounds good. Most of the industry giants still seem to be struggling with it, more than 25 years after the first Line 6 products hit American music stores.
Which is why I'm not too optimistic about getting good guitar amp sounds from the (otherwise very interesting) project to put DSP models on a Raspberry Pi. Roland/Boss hasn't got their code quite right in over two decades of trying. Ditto for Fender. Ditto for Yamaha. After a quarter of a century of trying, Line 6' still can't even match the Katana for sound quality. Will an open-source project do any better?
-Gnobuddy
I don't know the answer, but I think the problem was much worse than any subtle artifacts from digitization. After all, plenty of analog solid-state guitar amps suffered (and some still do) from the same sort of ear-cringing harsh sounds.jjasniew said:
Real tube amps have a lot of subtle complexities to them. Each tube stage has gentle nonlinearity over some range of input signal amplitudes, and more abrupt nonlinearities if pushed hard into saturation or cut-off. Grid current starts to flow in some regions of operation. Bias shifts occur that alter the nominal DC operating point as signal strength is changed, causing duty cycle modulation when overdriven. In pentodes and beam power tubes the screen grid voltage can be modulated by changes in output power, adding gentle compession due to "sag". There is EQ ("voicing") between every two or three stages, changing the way the overdrive reacts at different guitar frequencies. There's probably other stuff I can't remember (or don't know about).
I don't think any attempt was made to model most of these things in the early modelling amps. I think an attempt was made to model the overall EQ of an amplifier and speaker, and then some sort of crude nonlinearity was bolted on. Not very different than a pair of anti-parallel clipping diodes across and op-amp feedback resistor, and just as harsh and unpleasant.
As we've seen right at the start of this thread, you can get lots of unpleasant harsh clipping in the (analogue!) solid-state input stage, before the signal ever reaches the ADC to be digitized. If that was happening in these older modelling amps, it would be a case of garbage in, garbage out. The signal would already be harsh and unpleasant-sounding even before it entered the ADC.
In the case of the early Line 6 products, there was a really nasty nasal tone to all of them. Speaker / cab emulation gone badly wrong? I don't know, but just that unpleasant nasal honk alone was enough to make the amplifiers sound nasty. And then you add the harsh overdrive on top of that.
It's been a while since I heard a modelling amp quite as nasty as early Line 6 Spyders and Pods, but there are still big-name modelling amps that sound harsh (to me). The best part of two decades after my poor experiences with Line 6 products, I bought a first-gen Katana 50. To me, the Katana has two usable channels: the Acoustic channel (for electro-acoustic guitars), and the Clean channel, which produces the same sort of subtle tubey colouration you get from a good Fender Blackface amp. Millions of happy Katana owners love the higher-gain voices, but they all sound rather harsh to me. And this is Roland/Boss we're talking about - a giant in the industry, with a solid reputation earned over decades.
Is it just me? Well, some months ago I stumbled across a Youtube Channel called "The Studio Rats". They did a video or two on Katanas, and Paul Drew, their extraordinarily talented guitarist, said the same thing I've been describing (the best channel is the clean one). Because he didn't like the sound of the Katana's overdriven tones, Drew left the Katana on its clean channel, and used effects pedals to get his distorted and overdriven sounds.
So it's not just me. Paul Drew spends his days playing through a variety of good tube amps, and he hears something he doesn't like in the Katana's higher-gain amp models, just as I do.
On the plus side, my little Fender Mustang Micro doesn't seem to suffer from the harsh input-stage analogue clipping described at the start of this thread. And it runs on a single lithium battery. The supply rail can't be more than about 4 volts DC, and may be only 3.3V (a voltage used by most contemporary microcontrollers and DSP chips).
How does the Micro manage to avoid harsh input clipping? I don't know, but my guess is that someone in the industry finally addressed that same harsh clipping that's been bothering me for roughly the last 35 years, and simply built in a voltage divider right at the input, before the guitar signal ever hits an op-amp. Bad for signal-to-noise ratio, but at least you can keep away the clipping op-amp harshness before the signal hits the ADC.
On the minus side, to my ears, the Micro amp models are adequate, but not great. My ears think the same is true for their very overpriced Tone Master amps. Fender Corp still doesn't have DSP tube amp models that are quite up to snuff. Rather ironically for a company that gained its reputation by building amplifiers with clean tone, they can't get their own in-house Fender clean tone right - their models sound too thin and too bright to me, lacking that subtle "tubey" warmth.
So there are still plenty of "meh" modelling guitar amps and multiFX pedals sold today, even at price points pushing well north of $1000 CAD. They don't sound as nasty as the old Line 6 stuff, but they don't sound particularly good, either.
And then there's the little Flamma FS06, which still sells for just under $100 CAD, and somehow manages to sound better than some $1500 products. Sure, I would love it if it had separate control knobs for the clean and dirty channels in each model, and a presence knob would be nice, and not every amp model it contains sounds like the real thing. But they all sound good, I haven't heard any nasty noises from it, there are sublimely good tubey clean tones from the Fender and Two Rock models, and very nice progressive, touch-sensitive overdrive from the AC30 and Two Rock models.
All of which points to the simple fact that it's clearly not a simple and obvious job to write DSP tube amp code that actually sounds good. Most of the industry giants still seem to be struggling with it, more than 25 years after the first Line 6 products hit American music stores.
Which is why I'm not too optimistic about getting good guitar amp sounds from the (otherwise very interesting) project to put DSP models on a Raspberry Pi. Roland/Boss hasn't got their code quite right in over two decades of trying. Ditto for Fender. Ditto for Yamaha. After a quarter of a century of trying, Line 6' still can't even match the Katana for sound quality. Will an open-source project do any better?
-Gnobuddy
You say "Bad for signal-to-noise ratio" and I understand resistor and op-amp noise and such, but even attenuated, a guitar signal is still bigger than a moving-magnet phono signal, and decent phono preamps have very good s/n ratio ...
But (here's my rant) what strikes me about this is the "noise" that usually comes along with an electric guitar is picked up before leaving the jack. I see this among "pro" guitarists on Youtube (ahem, Rhett), especially if they got a high-gain distortion setup, they play a few notes then as they're muting the strings to say something, they're also turning the guitar's volume control down all the way. I can hear the hum and buzz go down with the volume control if they're even a 1/4 second late. Do people think that trash being part of the guitar signal is okay? The only thing that saves hearing such noise is high distortion/harmonic content of the main signal.
There's two things going on that too often doesn't get controlled for. One is "social distancing" between the guitar and AC magnetic fields from most any line-powered audio equipment, especially guitar amps - six feet is often not enough, but at least in a big studio it's possible to position things so that the guitarist is 12 feet away from anything that plugs into the wall. Of course humbuckers fix much of this, and a no-magnet coil can be added with single-coil, but still, get the guitar away from things...I'm not sure it's "not possible" in smaller studios, they just don't make the effort or aren't aware and live with it.
The other is electrostatic. If you hear even a slight hum while not touching anything, and it reduces or goes away when touching the (presumably grounded to the jack) strings or the output jack, it's electrostatic hum pickup, and (electrostatically) shielding the pickup/control cavities fixes that 99+ percent. I'm surprised not every model of "big brand name" guitars don't come completely shielded like that (okay, I haven't looked inside many recent $500+ guitars). I've done cavity shielding on cheap and mid-tier Squiers to great effect. It's well worth the effort.
What you describe is exactly what I preach since decades. From a standpoint of noise, a stock Fender is unusable and must be heavily modded.
So is it the code or the hardware running and around the code that's causing the perception? It would be interesting to do, if not actually, a thought experiment as to what would happen with the same DSP code, but running it on different grades of hardware. Considering all the mods that can be done to just a rPI that effects how it sounds, one would think hardware improvement would be a shoe-in to correlate with better sound.
I cant recall how the rPI signal modeler worked; wasnt it an AI system that listened to what the effect did, then created a DSP assembly to best match what it "thought" it heard? I also have no idea how many dimensions - certainly frequency, perhaps amplitude too, but time(?) it takes into consideration when attempting to scope out and model a sound.
I mostly posted that in a "here it comes" vein. Jump on it now, if interested in these things - before it gets so far along that like chatGPT - you dont really know who / what you're even talking to anymore. Or in the case of amp simulation, how they did it and can so easily do anything now, no problem.
Oh, but there was. They focused very thoroughly to such things. You obviously never bothered to, for example, study what tube amp emulation pioneering companies like Line 6 or Yamaha patented.
It's not the first time your comments in this thread are totally out of valid base.
Next time don't just think, instead research. Because again you were wrong in your thinking.
Also, guys like Ty Tabor, Steve Howe, Allan Holdsworth or Richie Kotzen also upgraded from their top tier Mesa, Lab Series or boutique Pearce and Hartley-Thompsons to these "harsh and unplesant" thingies. I pretty much take that as a record that even the earliest modeling amps were actually pretty darn good.
But have you actually studied typical DSP input stages, their "coarse" and "fine" input architectures? How they work? Because they don't generally have a lot of gain, in fact some even attenuate, and while headroom of the codec is on the low side it's still pretty good for the low input signals. It's still at least three times higher than grid clipping threshold of typical tube input stage.
But it's not happening under normal operating conditions so that's it about that.
IME, they had a lot of different tones and there really wasn't a common denomimator to how they sounded. Some Line 6 amps even had an analog tube power amp emulation circuit with sag and crossover simulation, kinda like the one that "modern" Quilters feature about 20 years later. As I said, the stuff was remarkably advanced even back then, "subtle complexities" included.
Tubes are monopolar, whereas solid state is Bipolar.
If you powerup a tube with just the filaments supply and use your multimeter to measure the ohms between the grid/plate and cathode, you can see each of the grids and plates conduct in one direction, acting as diodes. (grids show 200 / inf ohms, plates about 20k/inf ohms).
[All tubes will work with a/c plate supply, with halfwave, a technique used in some HV transmission/oscillators equipment. They do not employ separate rectifier tubes. (ignoring the filter capacitor)]
This cannot be simulated in a bipolar device.
I do not think, a solid state device can ever replicate the characteristics of a tube. A digital modelling can do it, not a discrete SS device circuit.
Those who like tube sound, take the tube way. Those who like the SS sound, go the SS way. Those who like both, go the digital modelling way.
Each has it's own coloring of sound.
Player>Guitar > pickup>preamp>speakers, this is just one of these to define the sound.
Most of the amps sound clean until certain volume, afterwards sound distorted.
Regards.
If you powerup a tube with just the filaments supply and use your multimeter to measure the ohms between the grid/plate and cathode, you can see each of the grids and plates conduct in one direction, acting as diodes. (grids show 200 / inf ohms, plates about 20k/inf ohms).
[All tubes will work with a/c plate supply, with halfwave, a technique used in some HV transmission/oscillators equipment. They do not employ separate rectifier tubes. (ignoring the filter capacitor)]
This cannot be simulated in a bipolar device.
I do not think, a solid state device can ever replicate the characteristics of a tube. A digital modelling can do it, not a discrete SS device circuit.
Those who like tube sound, take the tube way. Those who like the SS sound, go the SS way. Those who like both, go the digital modelling way.
Each has it's own coloring of sound.
Player>Guitar > pickup>preamp>speakers, this is just one of these to define the sound.
Most of the amps sound clean until certain volume, afterwards sound distorted.
Regards.
In what degree should the characteristics be replicated? Ordinary BJTs and FETs already have astoundingly similar characteristic curves to pentode and beam tetrode tubes, depletion mode FETs are basically driven very similarly to tubes, and the old, quickly obsoleted VFETs even mimicked triode characteristics.
In my experience, the issue is not that we couldn't manufacture devices that operate exactly like vacuum tubes, it is that such application is a niche and generally industry calls for devices that don't. Look what characteristics industry wanted the Fetrons to have (not those of a triode they replaced), look what happened to VFETs (no commercial success in a very niche application), look at to which direction nuvistors headed (they wanted it to be more transistor like), how efficient is a vacuum tube motor driver, computer or a class-D amp (so efficient that none are made or used seriously)? Would you trust a vacuum tube life support system? I wouldn't.
Most tube characteristics overall just stopped being wonderful already in the 1950's. That train passed.
So, it can be done to detail but it's not commercially viable, and to large extent the characteristics are and already were strikingly similar, or could be "bended" if will. With BJT and FET you have "pentode" style characteristics but look at what happens when you apply feedback: The characteristics begin to resemble those of triodes. And this is not even surprising because a pentode tube is basically similar to triode tube with suppressed internal feedback, (which can by bypassed in "triode mode"). So, how much does a difference even matter when it can usually be overcome with circuit tweaks.
And there we get to the bottom line: The circuits. We are not even comparing on device level but circuit level, and these tend to be totally apples-to-oranges types of comparisons. No. The output stage of 2203 Marshall, tweaked extensively to distort nicely, isn't going to be identical to a LM3886 integrated chip designed for hi-fi sound reproduction with very different circuit architecture and design goals.
If we want to make sensible comparisons of devices we should at least evaluate them in similar circuit architectures, and when we get to that then, IME, the differences actually start to be very negligible.
If we want to compare technologies we can pick units with similar performance and characteristics in general (e.g. tube amp and a digital or analog signal processing based system emulating it) and then again the differences start to be very negligible.
In my experience, the issue is not that we couldn't manufacture devices that operate exactly like vacuum tubes, it is that such application is a niche and generally industry calls for devices that don't. Look what characteristics industry wanted the Fetrons to have (not those of a triode they replaced), look what happened to VFETs (no commercial success in a very niche application), look at to which direction nuvistors headed (they wanted it to be more transistor like), how efficient is a vacuum tube motor driver, computer or a class-D amp (so efficient that none are made or used seriously)? Would you trust a vacuum tube life support system? I wouldn't.
Most tube characteristics overall just stopped being wonderful already in the 1950's. That train passed.
So, it can be done to detail but it's not commercially viable, and to large extent the characteristics are and already were strikingly similar, or could be "bended" if will. With BJT and FET you have "pentode" style characteristics but look at what happens when you apply feedback: The characteristics begin to resemble those of triodes. And this is not even surprising because a pentode tube is basically similar to triode tube with suppressed internal feedback, (which can by bypassed in "triode mode"). So, how much does a difference even matter when it can usually be overcome with circuit tweaks.
And there we get to the bottom line: The circuits. We are not even comparing on device level but circuit level, and these tend to be totally apples-to-oranges types of comparisons. No. The output stage of 2203 Marshall, tweaked extensively to distort nicely, isn't going to be identical to a LM3886 integrated chip designed for hi-fi sound reproduction with very different circuit architecture and design goals.
If we want to make sensible comparisons of devices we should at least evaluate them in similar circuit architectures, and when we get to that then, IME, the differences actually start to be very negligible.
If we want to compare technologies we can pick units with similar performance and characteristics in general (e.g. tube amp and a digital or analog signal processing based system emulating it) and then again the differences start to be very negligible.
I saw an interesting "mixed mode" amplifier schematic a while ago, which did almost exactly the same thing as one of my experiments, which was to run 2 feedback lines to the inverting input: 1) the usual voltage sensing + 2) the less usual current sensing.
Crucially, a capacitor in series with the current-sensing blocks low frequencies from the current feedback, and forms a low-pass RC filter blocking high frequencies from the voltage feedback.
Such a system behaves much like a voltage amplifier with a large output inductor (+ active pre-emphasis, boosting high frequencies), but without the bulky inductor. At low frequencies, damping factor is fairly normal, but at high frequencies, where nearly all speakers suffer from non-linear inductance, the added series impedance linearises the current supplied by the amplifier.
AFAICT, (good) tube amplifiers achieve much the same thing, ie: low damping factor to overcome "speaker problems", just using different means.
So with that in mind it doesn't surprise me that DSP techniques have generally failed to make ordinary solid state amps sound tubey. It's a false, and fatal assumption, to assume that the distortion mechanisms of amplifiers and speakers simply add together, and never subtract.
Last edited:
Exactly, these two technologies cannot replace one another. Each one is different and have their application specialty.
But when new technologies develop, If there are other technologies developing at the same time, the expensive ones will fail.
Ex : Sony Beta-max by Panasonic VHS, then DVD to MP4. Vinyl was replaced by Audio cassette and cassette was replaced by CD and then by MP3. This is life.
Like some people, love to keep grandpa's old wooden furniture, there are many people love to keep the earlier tech products and live with it.
But the Transistors did not win the war when they arrived and valves stayed on for further 30 years.
They tried to replicate the tube design on to the transistors, with interstage and output transformers etc. Poor thermal designs.
When we thought the transistor design is good, the Class D with SMPS comes in and will end all transistor power amps and standard transformer power supplies. This will create another group of people who love to keep their transistor devices and claim Class D is bad.
The servicing industry has become high tech with cameras and special SMD soldering equipment or more expensive to just change the boards as in a computer..
I always want to keep the knowledge of the past, though that technology has become obsolete by many. It is the foundation to knowledge over which I grasp the new technologies. (otherwise, you would not have published that free book)
Like many, I have great respects for your wide knowledge,
Whatever the newest audio technology that arrives, valves have their place and cannot be replaced.
Best Regards.