A guitar amp using the distortion of JFETs in the low level stages.

[Mike Sulzer]

In this discussion http://music-electronics-forum.com/t44934/ I said that I would use JFETs in a guitar amp, making sure that the low order distortion that the device makes from its gate voltage to drain current characteristic characteristic is retained. A description of the amp, which uses JFETs in the early stages, is in the next posts. The attachment to this post uses a somewhat simplified model to show that a JFET amplifier can be linear for small signals, but has a gradually increasing second harmonic with increasing sine wave input signal level. Of course it also generates low order intermod products with more complicated signals. The attachment also shows what happens if multiple JFET stages are cascaded.


The intent of this design is to understand how useful JFETs can be in the early stages of a guitar amp if the circuit is designed to retain the low order non-linearity, rather than linearizing it as in nearly all ss designs. This circuit is not intended to copy sound of a tube amp directly; instead, it is the simplicity of tube circuits that is copied, modified as necessary for use with JFETs.

[Mike Sulzer]

The first attachment shows the square law amp circuit. The idea is to cover the whole range from cutoff (no drain current) to IDSS. (IDSS is the current that flows with zero gate source voltage and enough voltage across drayn to source to keep it in the saturation region, that is, out of the so called ohmic range where it is not an effective amplifier.) The bias voltage then would be half the pinch off voltage, and the drain current would be .25IDSS (because it is a square law device).




However, it is not quite that simple. We want the device to behave well when driven hard, without undesirable changes in the wave form, such as a decreased duty cycle. The design uses voltage feedback biasing in order to prevent the the operating point from shifting. There might be better ways, but this is the one I found after encountering problems with the usual biasing methods. Even with this, the biasing needs to be shifted a bit from the ideal point, idling at a bit more than .25IDSS. This is necessary because the gate-source junction is a diode, and so the coupling capacitor can charge when the input is overdriven. It is a better diode that between the grid and cathode diode in a tube. Thus, this is more of a problem that with tubes. At the maximum possible overdrive, there is some blocking distortion, and so this circuit is not yet fully mature, but still very useful.


The 1k ‘gate stopper’ resistor is required to prevent oscillation with the source connected to ac ground.




The next two attachments show the time domain and frequency domain response, with input signal adjusted just short of clipping the output. The circuit measured was the “bread board” of this design. The TD looks as predicted in the attachment to the previous post, but up side down since the stage inverts. The frequency domain shows that the third harmonic is more than 25db below the second, and so it is a good square law amplifier.

sqLawTD.pdf
sqLawFD.pdf

[nickb]

I always enjoy hearing what you have to say, but I have to ask, how does it sound?

As the average tube has a power of 1.5 law, I would expect that adding audible harmonics would take fewer stages. although the ratio will be different.

[J M Fahey]

We want the device to behave well when driven hard, without undesirable changes in the wave form, such as a decreased duty cycle.

An d who says that´s undesirable?
At least, in the Guitar Amp World

Murphy something from Carvin invented an excellent variable duty cycle distortion, which was stolen and used to death by Crate, because he did bot patent it.

Just overdrive tubes (both triodes and pentodes) and you´ll see.

[Mike Sulzer]

I always enjoy hearing what you have to say, but I have to ask, how does it sound?

As the average tube has a power of 1.5 law, I would expect that adding audible harmonics would take fewer stages. although the ratio will be different.

The clean sound has warmth, although not exactly like a triode. It is the same sort of sound, not at all what is usually meant ss sound. IMO, it is potentially more useful than a triode because the simpler harmonic structure of a single stage can be cascaded with additional stages to make a great variety of dirty sounds, useful and not useful from my point of view.

It does take more stages, but more stages easy to do, and they provide the flexibility. I will try to discuss this more after showing the complete design.

[Mike Sulzer]

An d who says that´s undesirable?
At least, in the Guitar Amp World

Murphy something from Carvin invented an excellent variable duty cycle distortion, which was stolen and used to death by Crate, because he did bot patent it.

Just overdrive tubes (both triodes and pentodes) and you´ll see.

I didn't. But extreme decrease in duty cycle with increasing overdrive is a form of blocking distortion. In this design the duty cycle can be modified with additional stages. An example of this, short of clipping, is in the attachment to the first post.

[Mike Sulzer]

The amplifier circuit is shown in the attachment to this post.

The first stage is low gain and linear; it is a JFET for convenience, but it need not be. Its purpose is, first, to increase the voltage from the pickup but stay in the linear region with any reasonable pickup, and, second, to provide enough current to drive the 10K pot that controls the level into the first square law stage. It is just a linear buffer. The ‘gate stopper’ resistor is not needed in the input buffer amp since the source has a resistor to ac ground.

The rest of the low-level section consists of three square law stages with a tone stack before the final one. Each square law stage is preceded by a 10K gain control so that many combinations of gain and distortion are possible. So many gain controls are not convenient to use, but they are useful in a new design since the best way to set it up with a minimum number of controls is not yet known.

The power amp is a chip. The output of the last stage of the preamp is attenuated to prevent the power amp from clipping no matter how the controls are set. The power is 24 watts.

[Mike Sulzer]

The attachment shows the output of the amplifier for a sequence of 1KHz sine wave inputs increasing by 3db (a factor of 1.414 in voltage). The gain controls were adjusted so that the second square law stage overloaded. The output for the smallest input is a bit asymmetrical. For the largest input it is clipped top and bottom. The mid control was adjusted up to get closer to flat frequency response.