Tube/Valve sound using solid state circuit

[wonderfulaudio]

Hi,

Magazines like Elektor and Siliconchip have published projects long time back that emulate the Tube/Valve sound using solid state. Tubes/Valves dont have a linear curve and when run without feedback, they produce even order harmonics that are perceived as warmth. While I believe adding or subtracting anything to audio is not truly neutral but I have heard one such tube preamp and it did have a likeable sound for lack of a better word. This is a fun project where I can bypass the circuit with a switch if needed.
I have seen 2 different implementation of the method:
1) Using JFET in non linear region, I managed to isolate this JFET part
2) Using op-amp and a soft limiter, I dont have access to this though I am sure it was published in Elektor in 90s

Below is the JFET schematic, I would like to modify and experiment with it.

Q1 is configured as voltage follower and designed to work outside its linear region and is soft-clipping.
My problem is its biased such that its transfer curve is not symmetrical for the full swing of the input signal from the opamp.
How to make it symmetrical? Probably by adding a complementary FET to Q1 but how? Can somebody please help and explain me how to modify?
And what does VR2 do, it certainly does not drive Q1 symmetrically even though the name suggests in the schematic?

Also, there are op-amp based clippers used in guitar fuzz boxes, they dont sound same as tubes/valves because they have hard clipping.
I would appreciate if we can make progress with above JFET method as well as better op-amp based circuit than fuzz boxes, diodes on feedback loop of op-amp dont work.

Thanks and Warm Regards,
WonderfulAudio

 

[Francis Vaughan]

The interesting thing about the transfer function is how this relates to odd and even order harmonics. A symmetric non-linear transfer function produces only odd order harmonics. If you want even order harmonics you must have an asymmetric transfer function. Asymmetric yields a mix of both odd and even. Purely even is hard.
This is probably the overarching reason why you see single ended topologies.

Making a circuit like this actually truly symmetric would be non-trivial. A complementary architecture is the obvious answer, but semiconductors are never available in exactly symmetric pairs. The nature of charge carrier mobility is not symmetric, and the transfer functions are not simply inverted versions. In push-pull amplifiers we avoid such issues because we wrap a feedback loop around the thing to linearise it. That isn't possible if you are explicitly generating distortion. It might get you close enough, but that depends upon what close enough is.

However, you almost certainly don't want a symmetric circuit. The usual desire for a bit of tube warmth, is even order harmonics. A symmetric circuit explicitly prevents this.
VR2 is changing the operating region of the device, so calling it symmetric/asymmetric is sort of true. In the middle operating region it is probably reasonably linear and thus symmetric. The operating region might also be set low enough that the signal hard clips on the negative side. So that might be another reason for the naming.

Then you get to things like fuzz pedals. Well the answer is in the name. They sound fuzzy and nasty. Hard clipping is the entire reason they exist. The precise sound of a given pedal comes down to a whole slew of interactions between the vagaries of the pickup response, impedance and output, choice of diode parameters, power rails, and choice of lousy quality passive components. Guitar amplifiers have sounds that range from merely poor linearity up to deliberate pathological behaviour. The canonical example is the step from the Fender 5F6-A Bassman to the Marshal JTM45. The circuit is nearly identical. However one critical change Jim Marshal made was to increase the feedback in the JTM45. This resulted in vastly more distortion when overdriven, as the increased feedback drove the output stage further into clipping. The push-pull topology means lots of lovely odd order harmonic crunch. This sounds very different to a simple diode clipper. The manner in which the stage recovers, the effect of transformer saturation, and a slew of other second order effects comes into play.

The takeaway is that the sound of electric guitar signal chains, and the manner in which they operate, is a different world to the desires of music reproduction for listening pleasure. Adding a tiny amount of low order even and odd distortion is captured with a simple JFET circuit. Very careful use might approximate the sort of warmness valued by the various glowing bottle fanatics in the audio world. There is a whole world of modelling tube amps with FETs, but mostly for guitar amps, where serious distortion in its many forms is the key goal.

 

[EERecordist]

Do you know about https://distortaudio.org/? It is software that can process an audio file and add harmonics.

There should be similar plugins for roughly real time which can be added to digital audio workstation software. Some of the plugins have presets or styles, not the ability to add the amplitude of each harmonic.

In recording or mastering, the gain staging is controlled for artistic reasons to overwhelm transformer and gain stages to deliberately cause distortion or saturation. The music content is going to have the spectrum of frequencies, and the amplitude within those frequencies. So much of the music may be in the linear range, and a small amount may enter the nonlinear range.

I think for your experiments, it would be good to have a familiar track, and if you are able, bring it into digital audio workstation software to analyze in detail the before and after.

FETs and tubes can be analyzed by a curve tracer. The device is biased for the normal operating range to be linear. That is what you are getting at with your circuit. If you reduce the drain to source voltage, you enter the left hand side of the standard drain current vertical and drain to source voltage horizontal graph. That would be VR2. VR1 looks to adjust the negative feedback around X1. I would guess that allows the gate of the FET to be overdriven.

Do you know abot Room EQ Wizard software? With an A-D and D-A converter, it can measure the device under test distortion.

I'm rusty on analog design, but ASR has up to date designers.

 

[DVDdoug]

The advantage of software is that you can quickly adjust what it's doing. And it's possible to emulate any particular tube amp.

The disadvantage is that these plug-ins are meant to be used in a DAW during audio production so they easily don't fit into a home playback setup. There are "hosts" for plug-ins that can work with everyday playback, but you still have to run the audio from (or through) a computer.

 

[wwenze]

The reason why it is hard to emulate tube sound using anything is because nobody really knows what tube sound sounds like. This is not regarding those guitar boxes where we know why they sound like that and can be emulated (using DSP) mainly because they are designed to sound different. But for speaker amps where the goal is to amplify the input signal without changing it, the only thing we know that audibly changes is the frequency response due to high output impedance shenanigans (transformers and resistors and HVDC-blocking capacitor and stuff). Even for curvy transfer function, which the result can be measured as distortion, -80dB is inaudible, -60dB just makes you feel like something is wrong with the sound that otherwise sounds identical, and -40dB just hurts the ears.

If you want a curving transfer function, you can achieve it using a compressor

Googled a bit and found this a good read on the topic, including if you want something symmetrical:

Audio Clipper and Compressor Circuits Explained – Homemade Circuit Projects

www.homemade-circuits.com

 

[Heinrich]

If you want a curving transfer function, you can achieve it using a compressor

As have been said, tube sound isn‘t well defined. A professor analysed it, and it spread out in all directions. The speakers, the usual suspects, showed decidedly chaotic behavior. Random Subharmonics and more. To design, and eventually play such gear is an art in itself.

Why would you add such sounding to a ready made recording?

 

[CuteStudio]

Tubes/Valves dont have a linear curve

Nothing does
I would say that an amplifier with as much local feedback as possible would sound more 'tubey' as fewer harmonics are multiplied.
Best perhaps would be an tube input driving some voltage follower MOSFET outputs - MOS simply as they are easy to drive..

 

[Chr1]

I have a Lazarus H-1A power amp which is like this. A tube/MOSFET hybrid.