There's a lot of research on that. Turns out a lot of Rock types who are also geeks have a deep interest in DSP simulation of Tube Amps, mostly for guitar. Of course, a Tube Amp for music reproduction is very different, but the general ideas are in those papers - curves and non-linearities like soft-clipping. Dr. Ivan Cohen is one guy who worked on this. Actually, I believe it was his PhD thesis. There are many more and there must be papers on music playback tube amps too.
Look into the XRK Manufacturer section. He has a thread somewhere with transistor amps but with profiles similar to some tube amps, including some Pass Zen-syle ones.
So okay, this is what I'd thought I'd seen that claims emulation of any signal chain effector, including amplifiers. https://neuraldsp.com/quad-cortex
"Quad Cortex can learn and replicate the sonic characteristics of any physical amplifier, overdrive, and cabinet with unprecedented accuracy"
It's for guitar, but it's only a matter of time before this functionality leaks over into multi-channel audio systems. I'm surprised someone hasnt take a pair of these to task yet. Sure would make for an interesting series of yt videos.
I assume after learning how the particular speakers you have sounds, perhaps it could flip their response out of the way, then it could present the tonal sound character of other speaker system builds through your system. It couldnt get the dispersion characteristic of the speaker type it was modeling at this stage - but if HiFi aficionados would pay, they could probably come up with their own powered stereo sound system that would do that part too.
Basically, hear how anothers stereo sounds, through your own stereo system. Amplifier and speaker cabinets. Imagine posting about your new Pensyl build, attaching an analysis file - and within a few subsequent posts someone goes "sounds like you need to tease out your stuffing wool a little better". Perhaps we're within reach of such capability these days.
Besides professional use, I see it as a gimmick for home HiFi use. Maybe wouldnt last long as a product. How many amp and speakers could one listen to, before settling on a pair - even if emulated? In Audio the tonality of speaker cabinets is something we wish to make vanish, rather than use it as a sculpting tool for a particular sound, like a guitar player would.
"Quad Cortex can learn and replicate the sonic characteristics of any physical amplifier, overdrive, and cabinet with unprecedented accuracy"
It's for guitar, but it's only a matter of time before this functionality leaks over into multi-channel audio systems. I'm surprised someone hasnt take a pair of these to task yet. Sure would make for an interesting series of yt videos.
I assume after learning how the particular speakers you have sounds, perhaps it could flip their response out of the way, then it could present the tonal sound character of other speaker system builds through your system. It couldnt get the dispersion characteristic of the speaker type it was modeling at this stage - but if HiFi aficionados would pay, they could probably come up with their own powered stereo sound system that would do that part too.
Basically, hear how anothers stereo sounds, through your own stereo system. Amplifier and speaker cabinets. Imagine posting about your new Pensyl build, attaching an analysis file - and within a few subsequent posts someone goes "sounds like you need to tease out your stuffing wool a little better". Perhaps we're within reach of such capability these days.
Besides professional use, I see it as a gimmick for home HiFi use. Maybe wouldnt last long as a product. How many amp and speakers could one listen to, before settling on a pair - even if emulated? In Audio the tonality of speaker cabinets is something we wish to make vanish, rather than use it as a sculpting tool for a particular sound, like a guitar player would.
I agree.
That is why I did the extensive rebuilding and modifications to my console stereo cabinet.
3/4 inch pine bonded inside to the original 1/4 inch cabinet walls to prevent them from singing.
Then, it became easier to choose a set of woofers to provide the bottom-end of things without the boom often associated with consoles.
Tokin SITs are still available on ebay from Tomo who is a reliable seller.
https://ebay.com/usr/watanabetomoaki
The Schade preamp in post #3 works and sounds as good or better than my fully tricked out 10Y preamp. The FETs cost next to nothing. Just like tubes, they each have a flavor and it takes a while to find good ones. I explored that in the linked thread.
FET followers also sound very good when coupled with a tube pre or the Schade pre. I use a Sony VFET choke follower with the Schade preamp—triodes all the way.
With the Tokin SIT triodes, you can get 50W of single ended Class A. Look here:
Circuit with Tokin device minute 19.
Circuit with Schade FET minute 35.
the whole thing is quite interesting though.
https://ebay.com/usr/watanabetomoaki
The Schade preamp in post #3 works and sounds as good or better than my fully tricked out 10Y preamp. The FETs cost next to nothing. Just like tubes, they each have a flavor and it takes a while to find good ones. I explored that in the linked thread.
FET followers also sound very good when coupled with a tube pre or the Schade pre. I use a Sony VFET choke follower with the Schade preamp—triodes all the way.
With the Tokin SIT triodes, you can get 50W of single ended Class A. Look here:
Circuit with Tokin device minute 19.
Circuit with Schade FET minute 35.
the whole thing is quite interesting though.
The Ultimate Preamp by Analog Precision (user Tranquility Bass here) has DSP emulation of several tube types at varying levels.
I assume that it is a bit like adding some 2nd and 3rd according to the transfer function of a given tube. The effect to me was very subtle.
I leave it off in my application but I do play with it occasionally.
https://analog-precision.com/home/upp/
Scroll down to "Tube Emulation".
I assume that it is a bit like adding some 2nd and 3rd according to the transfer function of a given tube. The effect to me was very subtle.
I leave it off in my application but I do play with it occasionally.
https://analog-precision.com/home/upp/
Scroll down to "Tube Emulation".
You can buy a lot of plug-ins that simulate anything from phono scratches and noise to overdriven guitar amps, echo plates and everything you can imagine.
Google sound plug-ins.
Jan
adason,
Your idea of using a tube preamp and a solid state power amp is a nice idea . . .
for some of the sounds of some tube amplifiers.
But, I do not think that there is any tube preamp that has ever changed one thing . . .
1. The interaction between:
A. The power amplifier output impedance, and other factors, etc.,
and
B. The loudspeaker impedance, and other factors, etc.
Everything else about a tube preamp's attempt to do the above power amplifier / loudspeaker interaction is just a simulation.
Just my opinion
Your idea of using a tube preamp and a solid state power amp is a nice idea . . .
for some of the sounds of some tube amplifiers.
But, I do not think that there is any tube preamp that has ever changed one thing . . .
1. The interaction between:
A. The power amplifier output impedance, and other factors, etc.,
and
B. The loudspeaker impedance, and other factors, etc.
Everything else about a tube preamp's attempt to do the above power amplifier / loudspeaker interaction is just a simulation.
Just my opinion
How close do you suppose one could get to emulating the basic characteristics of a 6L6 PP stage and OPT with solid state? If you actually configured it in a similar way instead of following the usual diff pair-VAS-followers formula. Say, a pair of mosfets running in common source - with just enough drain-gate feedback to get say 20 to 30 ohms of output resistance? Not the usual large amount to make it “triode-ish” - instead keep it “pentode-ish”. Would it be possible to have just enough source degeneration to have “similar” linearity to a tube stage and have good bias stability? The OPT‘s imperfections could be emulated with “crossover“ components - it’s just passive filtering. L-C values wouldn’t be “stupid” if it’s all referred to 8-ish ohms. That would interact with the loudspeaker in the same way, wouldn’t it? I suppose one could use an actual transformer, but eliminating that expensive block of iron does have its appeal. Clipping would have to be addressed somehow, maybe ahead of the power stage so it never actually does clip.
wg_ski,
Are you good at doing straight load lines on a solid state amplifier output stage;
And then doing analysis of harmonic distortion, intermodulation distortion, and variable output impedance versus signal level?
Then . . .
Are you good at doing eliptical load lines on a solid stage amplifier output stage;
And then doing analysis of harmonic distortion, intermodulation distortion, and variable output impedance versus signal level?
If so, then repeat the straight load lines, and elliptical load lines; and analysis for a 6L6 PP output stage, with and without an output transformer.
By the way, the harmonic distortion, intermodulation distortion, and variable output impedance versus signal level are not the same for all 6L6 PP amplifiers.
Things like little or no negative feedback; different forms of negative feedback - global, local, nested; and
Beam Power mode, Ultra Linear, Triode wired mode.
Perhaps not all 6L6 PP amplifiers sound the same.
As a point of interest, there was at least one McIntosh solid state amplifier that used an output transformer.
I am not convinced that it is simple to make a particular solid state amplifier sound like some/certain particular tube amplifier(s), no matter how much you modify the solid state amplifier.
Just my opinions
Are you good at doing straight load lines on a solid state amplifier output stage;
And then doing analysis of harmonic distortion, intermodulation distortion, and variable output impedance versus signal level?
Then . . .
Are you good at doing eliptical load lines on a solid stage amplifier output stage;
And then doing analysis of harmonic distortion, intermodulation distortion, and variable output impedance versus signal level?
If so, then repeat the straight load lines, and elliptical load lines; and analysis for a 6L6 PP output stage, with and without an output transformer.
By the way, the harmonic distortion, intermodulation distortion, and variable output impedance versus signal level are not the same for all 6L6 PP amplifiers.
Things like little or no negative feedback; different forms of negative feedback - global, local, nested; and
Beam Power mode, Ultra Linear, Triode wired mode.
Perhaps not all 6L6 PP amplifiers sound the same.
As a point of interest, there was at least one McIntosh solid state amplifier that used an output transformer.
I am not convinced that it is simple to make a particular solid state amplifier sound like some/certain particular tube amplifier(s), no matter how much you modify the solid state amplifier.
Just my opinions
Yeah the MC2100 and MC2105, I was thinking the same thing. It's not a traditional output transformer but an autoformer, like a variac. Schematic http://www.tubebooks.org/file_downloads/McIntosh/MC2100.pdf
Im not convinced that it’s simple either. You’ll probably never exactly match the harmonic structure. But it seems to me the best way to GET a high but not infinite output impedance (without resorting to a 20 ohm resistor in series) is to run an output stage in common emitter/source, and not provide the usual 40 to 60 dB of NFB. Open loop, the Z is probably way too high so somewhere in the middle might provide the right slope to the curves. This should cause similar coloration in the SPEAKER. To match or emulate the right harmonic structure, compression, etc you’d have to do that in the front end - analog, DSP or whatever. You can’t emulate output impedance effects in a DSP, and a series resistor wastes more power than you put out when you’re trying to get a fractional damping factor, while still having the wrong limiting characteristics.
Using an OPT is certainly the most straightforward approach to a transformer “sound”. Plenty of old SS amps did it - not just high end McIntosh’s. But getting one custom wound these days would cost more than an off the shelf 5k to 8 50 W and a pair of 6L6’s. A suitable power transformer used as an OPT, maybe an option. It would let you get away with all N channel devices. But the three largest contributors - primary inductance, leakage reactance, and SRF are all relatively simple - when the values are referred to the 8 ohm secondary side AND if the source impedance is in the ballpark.
What I’m wondering is that if someone were to go to the trouble of trying it, could you get the speaker coloration right and have the proper limiting across frequency - without resorting to tubes and a transformer. Personally, I like tubes - but they are not my everyday use devices because of the finite lifetimes.
Currently I'm prototyping an amp based on Germanium devices. All the soviet power transistors are junk. They were producing them until the wall came down. Some have a 1990 date code and the few that were good exhibit an avalanche breakdown above a certain voltage. Their small signal transistors fare much better. Maybe 5 bad in a bag of 100. Paging through an old Motorola book and searching eBay found me some much better output transistors. Distortion is lower and the datasheets don't require a translation. It's impossible to direct couple a speaker because everything drifts at different rates. Offset at the output would swing +/- 1 volt! The only reliable method I could come up with would be a small peltier device on the VAS transistor controlled by a microprocessor. They're so temperature sensitive it's not even funny. Also funny that being PNP transistors this amp has a positive ground.
astouffer,
Thanks for the schematic.
The McIntosh output transistors were DC coupled to the autotransformer.
So a small DC offset voltage would put DC current into the winding;
That is kind of like a tube push pull output transformer with slightly un-balanced plate currents.
In both cases, there would be a small amount of quiescent magnetism bias in the core.
When is there zero offset voltage in a DC coupled solid state output stage?
When there is no feedback hysteresis, error, or drift.
Just my opinions.
In a perfect world . . .
Thanks for the schematic.
The McIntosh output transistors were DC coupled to the autotransformer.
So a small DC offset voltage would put DC current into the winding;
That is kind of like a tube push pull output transformer with slightly un-balanced plate currents.
In both cases, there would be a small amount of quiescent magnetism bias in the core.
When is there zero offset voltage in a DC coupled solid state output stage?
When there is no feedback hysteresis, error, or drift.
Just my opinions.
In a perfect world . . .
wg-ski,
There is a repair shop in town that fixes all kinds of solid state amplifiers and solid state receivers.
I have been a pest there many many times, but occasionally to purchase something that catches my eye.
However, I have never seen a tube amp on his repair bench.
Poor statistics, but . . .
Just saying.
There is a repair shop in town that fixes all kinds of solid state amplifiers and solid state receivers.
I have been a pest there many many times, but occasionally to purchase something that catches my eye.
However, I have never seen a tube amp on his repair bench.
Poor statistics, but . . .
Just saying.
About this and Kemper's. It makes a lot of sense IMO, to do that for guitar tube amps and the accompanying internal processing with the Tone Stack and more as well as the pedal chains people use with them because these are inherently colouring gear and hence you save a lot modeling the chains rather than buying various brands. Personally I have done some initial forays in DSP coding for synthesizers mostly as that's my area of interest, but when I get to it I will also integrate some Tube stage modeling because building an actual polyphonic analogue synth with tubes gets way too expensive very quickly. With DSP though, I'll be able to model that for free.
For music reproduction, to me, it makes more sense to optimise each gear in the analogue chain and the system, no matter what device is used in an amplification stage. For example, optimising a Tube stage to minimise distortions, doing the same with op amps, etc... This is because as audiophiles/musicophiles, we want the gear to be as transparent as possible. Some people want to reproduce the live feeling, others want to stay close to the studio recording.
We have to take into account here that there could also be some euphonic element in Tube playback, namely the preponderance of the 2nd harmonic being more pleasing to the ear. It's certainly more pleasing than higher odd-order harmonics, but this start moving away from pure transparency - nothing wrong with that since we listen to music for enjoyment, but some will prefer transparency to this.
If I'm not mistaken there was a Bob Carver amp doing the emulation by tailoring the harmonics, done in the Electronics domain so that it could sound like a SS amp or a Tube amp but built with SS devices. I haven't heard it.