Makes you wonder if any one else has thought of doing voltage and current feedback into an ADC, along with the guitar's signal output. 3 independent channels that is, all discrete-time-math on sample data from there. How many hundreds of kHz can we do at 24 bits these days?
Quite many modern modeling amps employ current feedback input to DSP as part of processing. Refer to, for example, Blackstar's patent. Their ID series came out in 2007 so it's not exactly a new thing either.
Formerly quite a many modeling amps had all analog power amp stages, with current feedback included like is usual with guitar amps, and a fully analog tube emulation implemented to it: Vox Valvereactor series, Roland Cube series, Peavey Vypyr series, and some of the more expensive Line 6 amps at least would be in this category. Whoops, that's lion's share of early modelling amps.
I fail to accept the claims that these things were simple amps with crude modeling of e.g. just EQ and some waveshaping function, lacking dynamics or low damping factor, because when you look at them and learn about the DSP processing involved they just weren't.
Formerly quite a many modeling amps had all analog power amp stages, with current feedback included like is usual with guitar amps, and a fully analog tube emulation implemented to it: Vox Valvereactor series, Roland Cube series, Peavey Vypyr series, and some of the more expensive Line 6 amps at least would be in this category. Whoops, that's lion's share of early modelling amps.
I fail to accept the claims that these things were simple amps with crude modeling of e.g. just EQ and some waveshaping function, lacking dynamics or low damping factor, because when you look at them and learn about the DSP processing involved they just weren't.
Last edited:
Yet they still sounded (to quote the late Slipperman) "like **** and balls". Harsh, nasty stuff. Like early CDs or straight diode clipping into cheap piezos.
Yet we still have products coming out every year claiming to have finally solved the emulation problem. A bit like autonomous cars, really.
And I'm 100% with Gno (a rare thing I know) on the "nasal" thing on Line 6. Now, if that's what you're looking for in guitar sound, lucky you. But it's not the sound of any traditional tube amp.
Brining my work and hi-fi experience biases, it's likely that there's a pile of unplanned clipping events (digital and analog) and a raft of non-musically related distortion/signal-generated-noise, plus some heroic assumptions about filter effectiveness vs the variety and vagaries of human hearing.
The unwanted clipping aka spikes in waveform is actually the signal overloading the amp.
More evident at higher volumes. The only solution is to get a higher power amplifier and run at the lower watts.
The signal is not any sine wave, but complicated waveform.
If the amp has maximum input level of 800 millivolts, The peak spike signal must be limited to 800 millivolts.
Means, you will have to use a power amp which can put more power and control it at the power level as you require.
Generally, an IC type preamp will not overload if the supply is +/- 12 or above.
For transistor based preamps, the emitter voltages must be increased as the gain increases, which also require higher supply voltages well above that of an IC preamp.
Signal overloading occurs to the digital amp also from the first analog preamp stage feeding the DSP. Usually they operate at 5 volts DC
Regards.
More evident at higher volumes. The only solution is to get a higher power amplifier and run at the lower watts.
The signal is not any sine wave, but complicated waveform.
If the amp has maximum input level of 800 millivolts, The peak spike signal must be limited to 800 millivolts.
Means, you will have to use a power amp which can put more power and control it at the power level as you require.
Generally, an IC type preamp will not overload if the supply is +/- 12 or above.
For transistor based preamps, the emitter voltages must be increased as the gain increases, which also require higher supply voltages well above that of an IC preamp.
Signal overloading occurs to the digital amp also from the first analog preamp stage feeding the DSP. Usually they operate at 5 volts DC
Regards.
Ok, seems like a time for a practical "reality check": I'm looking at circuit diagram of basic 20-year old 15W Line 6 Spider and the CODEC actually operates at, not even 5V, but at 3V.
Yes, seems low.
Yet, if we go by those Rod Elliot measurements that gnobuddy quotes at the start of this thread and assume a 500 mV initial transient then the extremely low gain of the input stage (x1.25) still keeps the input signal amplitude well below the overdrive threshold of the CODEC. We could push the thing with a lot higher transients before DAC or input stage would clip. The input would easily handle even a 2-volt transient, so four times as much as measured by Rod Elliot! Line 6 Service manual actually quotes 2.5V (peak) maximum input signal amplitude level before clipping, so that's five times as much as measured by Rod Elliot and certainly much higher than grid clipping level of any typical triode input stage of tube amps.
Yet these things actually operate at so low input signal amplitudes in general that Line 6 follows the usual procedure of "DAC stacking" and adds a sidechain with 16x gain just to feed the usual millivolt range of input signals to the other channel of the input DAC at higher amplitude. This is a very common setup to artificially increase dynamic range of a DAC and reduce noise.
In practice, this is a good example how CODEC input stage design actually does acknowledge the transient nature of the input signals and are therefore made accordingly; to tolerate a signal with high dynamic range, noiselessly and without overloading.
Note that this is a rudimentary scheme of even cheapest "practice amps" from 20 years ago, not some newest high tech circuit. And they still have ample headroom to tolerate at least four times higher transients than measured by Rod Elliot or assumed in this thread discussion.
Yes, seems low.
Yet, if we go by those Rod Elliot measurements that gnobuddy quotes at the start of this thread and assume a 500 mV initial transient then the extremely low gain of the input stage (x1.25) still keeps the input signal amplitude well below the overdrive threshold of the CODEC. We could push the thing with a lot higher transients before DAC or input stage would clip. The input would easily handle even a 2-volt transient, so four times as much as measured by Rod Elliot! Line 6 Service manual actually quotes 2.5V (peak) maximum input signal amplitude level before clipping, so that's five times as much as measured by Rod Elliot and certainly much higher than grid clipping level of any typical triode input stage of tube amps.
Yet these things actually operate at so low input signal amplitudes in general that Line 6 follows the usual procedure of "DAC stacking" and adds a sidechain with 16x gain just to feed the usual millivolt range of input signals to the other channel of the input DAC at higher amplitude. This is a very common setup to artificially increase dynamic range of a DAC and reduce noise.
In practice, this is a good example how CODEC input stage design actually does acknowledge the transient nature of the input signals and are therefore made accordingly; to tolerate a signal with high dynamic range, noiselessly and without overloading.
Note that this is a rudimentary scheme of even cheapest "practice amps" from 20 years ago, not some newest high tech circuit. And they still have ample headroom to tolerate at least four times higher transients than measured by Rod Elliot or assumed in this thread discussion.
Last edited:
Second "reality check":
I'm looking at US patent #5,841,875 from 1998 assigned to Yamaha Corp. This discusses tube amplifier emulation and "harmonics modification" featured in their DG series modeling amps.
Along with plenty of other topics it describes a waveshaping function, which is made non-static by clever implementation of "virtual" DC offset estimation superposed to audio signal. In practice this works like an emulation of a gain stage with a dynamically shifting input offset; which in correspondence alters overall (a)symmetry and therefore harmonic characteristics of "clipping".
What the patent describes is a circuit emulation with several such stages cascaded (just like in tube preamps) or operated in push-push (just like in tube power amps). So we have a preamp emulation estimating the interstage bias shifting and its dynamic effects, and a power stage emulation doing the same.
So definitely not just a single filter and static waveshaper like pair of antiparallel diodes in the feedback loop, and definitely modeling subtle complexities of tube amps back in 1998 when pretty much the earliest digital amp modeling technology even debuted.
I still remember few years past remarks about "digital fizz" in a certain modeling unit, which turned out to be dynamic crossover emulation of similar effects appearing in real tube amps. The manufacturer kindly pointed that out and also instructed how to turn it off and boy, did that shut up the mouths. That was nearly about as much fun as somewhat earlier popular comments badmouthing digital distortion in units that actually were analog as far as signal processing goes.
I'm old enough to remember the times when looking like digital was enough to also make units sound digital so needless to day that today I'm sceptic towards all anecdotes about characteristics people claim to hear. The bias is usually evident when you read between the lines.
I'm not going to bother to dig out further examples. I understand that digital modeling amp bashing is wonderful sport for people but could you please at least try to get the facts straight when engaging to such sport. 1. Their input does not clip under normal operating conditions and 2. The emulations were extremely realistic from day one.
I'm looking at US patent #5,841,875 from 1998 assigned to Yamaha Corp. This discusses tube amplifier emulation and "harmonics modification" featured in their DG series modeling amps.
Along with plenty of other topics it describes a waveshaping function, which is made non-static by clever implementation of "virtual" DC offset estimation superposed to audio signal. In practice this works like an emulation of a gain stage with a dynamically shifting input offset; which in correspondence alters overall (a)symmetry and therefore harmonic characteristics of "clipping".
What the patent describes is a circuit emulation with several such stages cascaded (just like in tube preamps) or operated in push-push (just like in tube power amps). So we have a preamp emulation estimating the interstage bias shifting and its dynamic effects, and a power stage emulation doing the same.
So definitely not just a single filter and static waveshaper like pair of antiparallel diodes in the feedback loop, and definitely modeling subtle complexities of tube amps back in 1998 when pretty much the earliest digital amp modeling technology even debuted.
I still remember few years past remarks about "digital fizz" in a certain modeling unit, which turned out to be dynamic crossover emulation of similar effects appearing in real tube amps. The manufacturer kindly pointed that out and also instructed how to turn it off and boy, did that shut up the mouths. That was nearly about as much fun as somewhat earlier popular comments badmouthing digital distortion in units that actually were analog as far as signal processing goes.
I'm old enough to remember the times when looking like digital was enough to also make units sound digital so needless to day that today I'm sceptic towards all anecdotes about characteristics people claim to hear. The bias is usually evident when you read between the lines.
I'm not going to bother to dig out further examples. I understand that digital modeling amp bashing is wonderful sport for people but could you please at least try to get the facts straight when engaging to such sport. 1. Their input does not clip under normal operating conditions and 2. The emulations were extremely realistic from day one.
So which products did the Yamaha patent end up in?
And a loud bass guitar plucked hard will generate 2V p-p. Even if you don't clip the input stage, it takes only two additions to clip it digitally. Then you've got to reconstruct it at the output, which is it's own can of worms. Ran into all of those problems and more as an EE in my previous life
As for anecdotes, I've been playing guitar(badlly) for decades, including the recent Yamaha digital amps, and I remain disappointed. I've spent much of my spare time in the last decade trying to work out why. If you can't hear the difference, you're very lucky.
And a loud bass guitar plucked hard will generate 2V p-p. Even if you don't clip the input stage, it takes only two additions to clip it digitally. Then you've got to reconstruct it at the output, which is it's own can of worms. Ran into all of those problems and more as an EE in my previous life
As for anecdotes, I've been playing guitar(badlly) for decades, including the recent Yamaha digital amps, and I remain disappointed. I've spent much of my spare time in the last decade trying to work out why. If you can't hear the difference, you're very lucky.
Their very early DG series. The example was to show that complex dynamics of tube circuits were acknowledged and modelled very early on, practically in the very first modeling amps ever. The argument that early modeling tech was "simple" in this regard does not hold water.
And as was ascertained, even the cheapest modeling amps from 20 years ago were rated to tolerate some 5V p-p, more than half of that generated by that loud bass plucked hard.
Well, I'm sure of that but that's a different argument. My point was to debunk the common belief of input stages running out of headroom and clipping.
And as for anecdotes, they are merely reflecting people's opinions. If your opinion is not to like Yamaha products and consider them to feature a poor emulation then you are entitled to it. Yet a countering opinion may come from, say, Allan Holdsworth who ditched a lot of pro gear in favour of his DG-1000 amp and had nothing but good to say about it. Needless to remind, those were amongst the very first modeling amps to appear. There are Internet forumites whose opinion is that modeling amps are horrible and continue to be such this very day while there are famous professional musicians whose opinion is that the technology has been extremely good since the very beginning. I may not have to state whose opinion and expertise I'm more willingly listening.
Problem with opinions is that they may not neccessarily reflect everyone's experiences and preferences. I don't know what your preferences are musically nor how you even tweak your amps. I know I can dial the most expensive boutique amp to sound like **** just as easily as "user errors" are made with modeling amps, yet that doesn't imply anything particular about characteristics of either, and opinions could still be said about supposed tones of them. Opinion is not a fact.
That inputs of generic digital modeling amps seldom clip under designed operating conditions, and that early modeling amps already featured rather complex and dynamic emulations of tube amps, however, are not opinions. They are proven by practical examples of units featuring such technology, (e.g. circuit diagrams, patents that describe the technology, actual measurements etc.) and furthermore that has nothing to do with how my playing, using my equipment, sounds like.
If you want to hear tones of a very early modeling amp you can just take a listen of, say, the album "Sixteen Men of Tain" (which definitely is not my cup of tea in slightest but Holdsworth is infinitely more famous musician and guitarist than I am nor will ever be) ...or just youtube search for soundclips in general and find out that things like the Yamaha DG-1000 or Line 6 Axsys even today actually sound pretty damn good in right hands and have a lot more than a single tone to characterize them with.
Last edited:
Except they don't. Again, 2V PP plus any sort of gain and it'll clip. And If you clip an op-amp it will sound like crap.
This is not news (see Hamm or Crowhurst). Your second claim (and it is your claim) that them emulations "were extremely realistic" presumes that the right things were being compared.
Now, it is entirely possible that, with my budget approach, I've (and everyone else) has missed a digital emulation that's not merely equal to a RunOffGroove circuit but basically identical to the origin. As opposed to merely cheaper to produce and ship in volume. At the end of the day, I can put whatever crap I like in a patent, it's not truth (or practicability) that's being examined.
Finding one artist for whom a digital amp "floated their boat" (back in the day) is like finding someone with a PhD who has issues with some aspect of global warming. Inevitable, but like looking at the edge of an axe under a microscope it ignores the weight of the rest of the axe head.
For those interested in the original topic, something I stumbled over the other day which goes through some basic concepts
http://www.geofex.com/effxfaq/distn101.htm A Musical Distortion Primer 2000 R.G. Keen
Hamm, R.O., 1973. Tubes versus Transistors-is there an Audible Difference. Journal of the audio engineering society, 21(4), pp.267-273.
e.g. https://milbert.com/Files/articles/TvsT/tstxt.pdf
Crowhurst, N.H., 1957. Some Defects in Amplifier Performance not covered by Standard Specifications. Journal of the Audio Engineering Society, 5(4), pp.195-202.
e.g. https://www.audiofaidate.org/articoli/Crowth_SomeDefects.pdf
For those interested in the original topic, something I stumbled over the other day which goes through some basic concepts
http://www.geofex.com/effxfaq/distn101.htm A Musical Distortion Primer 2000 R.G. Keen
This excerpt treats just the fundamental basics of different types of distortion, nothing wrong with that, nothing new at all. I assume you did not read Teemuks book, which is far more informative.
http://www.geofex.com/effxfaq/distn101.htm A Musical Distortion Primer 2000 R.G. Keen
This excerpt treats just the fundamental basics of different types of distortion, nothing wrong with that, nothing new at all. I assume you did not read Teemuks book, which is far more informative.
Sigh. Firstly, that was 5Vpp and yes, indeed any additional gain once input sensitivity is exceeded leads to clipping. We are still way above 2 - 3 Vpp limits of typical tube amp input stages and I haven't heard much complaints that it would be impossible to get clean tones with them because of inputs overdriving.
Opamp clipping can sound just fine. Tech 21 has earned millions with it.
People still keep referring to this BS? Didn't you know that majority of that article's contents got debunked way in... drumroll... 1973. Right after the article was even released.
Pretty much
Amp / speaker emulation and classic effects.
All very well done
Especially in software applications.
More than 20 years ago.
Always some guy saying blah blah blah, it is this and that solid state, blah blah.
nothing touches blah blah tubes blah blah.
Usually the better players easily proof it.
Just like bass amps and all these magical things they need.
Then some just plug in and play.
Make a farty speaker and 2 band EQ go down in history.
9 volts 18 volts 30 volts
all work
Amp / speaker emulation and classic effects.
All very well done
Especially in software applications.
More than 20 years ago.
Always some guy saying blah blah blah, it is this and that solid state, blah blah.
nothing touches blah blah tubes blah blah.
Usually the better players easily proof it.
Just like bass amps and all these magical things they need.
Then some just plug in and play.
Make a farty speaker and 2 band EQ go down in history.
9 volts 18 volts 30 volts
all work
Yes. And the goal posts and targets of nostalgia keep on shifting whenever new features settle in: after tubes vs. solid-state it turned out to be class-D vs. class-AB debate, the latter as an older technology naturally miraculously better, as if regions of output device operation told anything about characteristics of various and specific circuits. Then the debate got to SMPS versus traditional linear supplies.
Then it got to the digital vs. analog debate, and now that digital modeling has finally been settled in people are once again longing back to the days of those old analog transistor amps, as if we would have forgotten that their deficiencies lead to generation of all the newer stuff in the first place.
For example, when Fender released the Mustangs people were all over hyping them over the old Dynatouch analog amps they used to make. And now that they introduced the Tone Masters people complain why Fender doesn't make analog amps, because they are supposedly better, because someone in the Internet thinks that way. Or something like that.
IMO the preferences shift with no evident logic and I'm not going to let that cloud my judgment. Agreeing with the latest fad today means you are disagreeing with it tomorrow when opinions again shift without no apparent reason.
Then it got to the digital vs. analog debate, and now that digital modeling has finally been settled in people are once again longing back to the days of those old analog transistor amps, as if we would have forgotten that their deficiencies lead to generation of all the newer stuff in the first place.
For example, when Fender released the Mustangs people were all over hyping them over the old Dynatouch analog amps they used to make. And now that they introduced the Tone Masters people complain why Fender doesn't make analog amps, because they are supposedly better, because someone in the Internet thinks that way. Or something like that.
IMO the preferences shift with no evident logic and I'm not going to let that cloud my judgment. Agreeing with the latest fad today means you are disagreeing with it tomorrow when opinions again shift without no apparent reason.
My guitar synths take the frequency, velocity from the divided pickups and transform it to sounds of other guitars or synths.
I believe, the modelling amps do the same. The signal can be 2.5 volts, changed to a less square wave to detect the plucked frequency and processed.
I will not compare a digital modeled amp sound to analog amp sound, since one is produced and the other processed as for this discussion.
Regards.
I believe, the modelling amps do the same. The signal can be 2.5 volts, changed to a less square wave to detect the plucked frequency and processed.
I will not compare a digital modeled amp sound to analog amp sound, since one is produced and the other processed as for this discussion.
Regards.
I dont think so.
The modelling amps / tech are in large part trying to reproduce the sounds / tones / feel / response of (mainly) classic valve amplifiers, so comparing the sounds from each is the whole point!
Guitar synths are a different thing entirely. They aren't trying to emulate any particular "amplifier" as such; they are being used as triggers for synthesisers to generate any number of unique sounds. Very different goal.
FWIW: I've not played any modelling amps yet that I can actually enjoy the tone of. (I gave away the Line 6 Spider I briefly owned.) But that said, I've never managed to get a filth tone that I ever liked out of several quality valve amps that I've tried either, so I suppose not liking the modellers I tried until now is not a stretch. I'm sure Kemper Profilers,, Axe FX Ultra, Quad Cortex and the like are very good, but right now and given their prices, I'm not in a hurry to replace my valve-based fleet yet!
Cheers, and regards,
Ant.
Not necessarily simple, but then again there have been improvements in the state of the art since mid 90s. Although for the times, some of the early processors like Line 6 POD did quite a lot of pretty amazing things and were pretty much ground-breaking. I still have my Pod, and used it live and recording for quite a while, but the amp sims on newer stuff these days seem to me much more realistic. Whether it's faster processing (lower latency), better AD resolution and sample rates, better algorithms for the amp simulations..overall it makes for a better experience and sound. TC Electronics came out with a line of amp-in-a-box DSP based pedals, similar to the UA offerings but at lower cost. As mentioned earlier there is the Flamma Preamp, and Katana series amps that are digital based.
The Katana was one of the first DSP amps that really made me realize that the DSP technology could be at the point of being indistinguishable from playing through a vacuum tube based amp.
I'm not saying digital tube amp emulation hasn't improved. My intention was to point out that the overall idea that first examples of modelling technology from the late 1990's were super simple and lacking the finer dynamic details is falsified.
The early digital modeling technology was compared to simple EQ and a static waveshaper, which is far from the truth. The bar was raised far higher than that already by ordinary analog SS guitar amps and their modeling circuit ideas. If, say, Peavey had pushed out clever analog emulator ideas like TransTube and T-Dynamics, SLM the FlexWave, Roland their analog tube logic, Pritchard his amps, and back in the 1980's we already had Rockmans, SansAmps, ADAs, H&K and all, then why would have the digital units settled down to offering performance equaling the "Tube Screamer"?
They didn't. The digital modeling was a BIG leap ahead because it enabled to introduce very complex signal processing at lesser circuit complexity. This exactly allowed more detailed focusing to all the subtle nuances of tube circuits.
And most of those designers also knew a thing or two about those tube amps they set to emulate. You can google search PS Systems EB100S or Power Tool to have some idea on what the Line 6 guys, for instance, had cooking before the Line 6 was established.
The early digital modeling technology was compared to simple EQ and a static waveshaper, which is far from the truth. The bar was raised far higher than that already by ordinary analog SS guitar amps and their modeling circuit ideas. If, say, Peavey had pushed out clever analog emulator ideas like TransTube and T-Dynamics, SLM the FlexWave, Roland their analog tube logic, Pritchard his amps, and back in the 1980's we already had Rockmans, SansAmps, ADAs, H&K and all, then why would have the digital units settled down to offering performance equaling the "Tube Screamer"?
They didn't. The digital modeling was a BIG leap ahead because it enabled to introduce very complex signal processing at lesser circuit complexity. This exactly allowed more detailed focusing to all the subtle nuances of tube circuits.
And most of those designers also knew a thing or two about those tube amps they set to emulate. You can google search PS Systems EB100S or Power Tool to have some idea on what the Line 6 guys, for instance, had cooking before the Line 6 was established.
No. As said, the synthesizers use the signal as a mere trigger for a totally individual synthesizing process. They basically generate an entirely new signal according to specific parameters like the said note frequency and velocity.
Digital signal processing in turn processes its input signal, just like analog signal processing processes it instead of synthesizing a entirely new one. Just like in analog realm the signal processing introduces effects, which most commonly alter frequency response and add harmonic distortion via compression / clipping, "waveshaping". In various combinations, steps and styles, with dynamics added in to turn "waveshaping" non-static, these two signal processing effects already get extremely far in emulating various kinds of analog audio processing circuits, like tube or transistor amps or effect pedals.
Time-based effects, like delays, and their derivites are also very common additions to arsenal, though tube (amp) circuit emulations as is seldom make any use of such. However, time-based effects were initially the only digital signal processing part of many "digital" amps like Fender Cybers, Digitech preamps, Johnson, etc. Back in the days digital was once again a superior solution to replace the old analog "delay lines" like bucket brigade delays, or electromechanic systems consisting of springs or plates connected to transducers, or just overall mechanic systems such as rotating speaker baffles.