Embedded Files
The Direct-Coupled Cathode Follower (DCCF) has appeared in guitar amps for many years. It dates back to at least the '59 Fender Tweed Bassman (5F6A) amp. It's a common circuit in Marshalls, Fenders, and many other amps over the years. Its low impedance output makes it useful to drive a tone stack. It's also known for its compression and distortion characteristics, that are often referred to as "warm" distortion. This discussion focuses on that aspect of the circuit. A common form of the circuit is shown below, as it appears in the JTM 45, JCM800 lead (2204), 5F6A, and many other amps.
The Direct-Coupled Cathode Follower (DCCF) has appeared in guitar amps for many years. It dates back to at least the '59 Fender Tweed Bassman (5F6A) amp. It's a common circuit in Marshalls, Fenders, and many other amps over the years. Its low impedance output makes it useful to drive a tone stack. It's also known for its compression and distortion characteristics, that are often referred to as "warm" distortion. This discussion focuses on that aspect of the circuit. A common form of the circuit is shown below, as it appears in the JTM 45, JCM800 lead (2204), 5F6A, and many other amps.
The power supply voltages vary among amps, but typical voltages are 250V - 400V. For this discussion, we'll use a 300V power supply. The concepts and methods for optimizing the circuit are valid for the entire typical range.
The power supply voltages vary among amps, but typical voltages are 250V - 400V. For this discussion, we'll use a 300V power supply. The concepts and methods for optimizing the circuit are valid for the entire typical range.
So, what do I mean by "optimizing" or "tuning" the circuit? The optimization is to control the harmonic generation that occurs when the circuit is in overdrive. The circuit is known for generating distortion. What isn't known (or at least not well known) is how to control the type of distortion that it generates. The approach described below controls the character of the distortion, adding significant tone flexibility and novel tone options to a DCCF stage.
So, what do I mean by "optimizing" or "tuning" the circuit? The optimization is to control the harmonic generation that occurs when the circuit is in overdrive. The circuit is known for generating distortion. What isn't known (or at least not well known) is how to control the type of distortion that it generates. The approach described below controls the character of the distortion, adding significant tone flexibility and novel tone options to a DCCF stage.
Before describing the optimizations, let's review how the circuit works and what the circuit does. Many explanations are already available, perhaps most accessible and well known is that of the Valve Wizard (Merlin Blencowe) book and website. (http://www.valvewizard.co.uk/dccf.html) These describe the basic circuit behavior, but the circuit is treated as relatively fixed with little mention of variations or optimizations for tone. The variations described below optimize the circuit behaviors to significantly alter and control the distortion tones.
Before describing the optimizations, let's review how the circuit works and what the circuit does. Many explanations are already available, perhaps most accessible and well known is that of the Valve Wizard (Merlin Blencowe) book and website. (http://www.valvewizard.co.uk/dccf.html) These describe the basic circuit behavior, but the circuit is treated as relatively fixed with little mention of variations or optimizations for tone. The variations described below optimize the circuit behaviors to significantly alter and control the distortion tones.
Firstly, the classic DCCF, as implemented in the example amps mentioned above, use an unbypassed 820 ohm resistors for establishing U1 cathode bias. Although other bias resistors could be used, let's keep that part of the circuit for now. For the circuit shown above, Spice simulations show DC currents (no-signal) of 1.25ma in U1 and 1.75ma in U2. These currents indicate that both tubes are warm-biased, so small asymmetric distortions are produced in both stage due to transfer-curve nonlinearities inherent in warm biasing. In the no-signal condition, there is essentially zero grid current flowing in either tube. It is important to note this since grid currents are important in the large-signal distortions created by the DCCF, but they are not a factor until sufficient signal voltages are applied.
Firstly, the classic DCCF, as implemented in the example amps mentioned above, use an unbypassed 820 ohm resistors for establishing U1 cathode bias. Although other bias resistors could be used, let's keep that part of the circuit for now. For the circuit shown above, Spice simulations show DC currents (no-signal) of 1.25ma in U1 and 1.75ma in U2. These currents indicate that both tubes are warm-biased, so small asymmetric distortions are produced in both stage due to transfer-curve nonlinearities inherent in warm biasing. In the no-signal condition, there is essentially zero grid current flowing in either tube. It is important to note this since grid currents are important in the large-signal distortions created by the DCCF, but they are not a factor until sufficient signal voltages are applied.
U1 is a warm-biased common-cathode gain stage, and as the input signal voltage increases, at some point grid current starts to flow, which increases distortion. The distortion due to grid current is what we wish to control and optimize to our taste. It's often noted that grid current itself is not the cause of distortion. Distortion arises since the source impedance is often high and unable to drive the grid once grid current flows. That produces a compression (or clip) of the positive signal peaks at the grid.
U1 is a warm-biased common-cathode gain stage, and as the input signal voltage increases, at some point grid current starts to flow, which increases distortion. The distortion due to grid current is what we wish to control and optimize to our taste. It's often noted that grid current itself is not the cause of distortion. Distortion arises since the source impedance is often high and unable to drive the grid once grid current flows. That produces a compression (or clip) of the positive signal peaks at the grid.
So let's apply a signal to U1 and see what happens. For input signals below 2V peak amplitudes, U1 behaves as any common cathode stage would behave and U1 grid current is low enough to be insignificant. Once a 2V peak signal is applied, U1 grid current begins to flow at the positive signal peaks. Simulations show peaks of about 7 microamps (uA) of U1 grid current. Let's consider 2V as the threshold of U1 grid current flow. Let's call this 2V signal threshold GT1. So GT1 is the U1 input signal voltage that begins to produce grid current in U1. Let's also define an analogous GT2 value, as the U1 input signal voltage that begins to produce grid current in U2. Note, however, that GT2 is the threshold of U2 grid current caused by negative U1 input signal peaks. Note that GT1 and GT2 may be different. In this example, GT2 is 0.75V, meaning that only .74V peak signals into U1 are sufficient to cause U2 grid current (~5uA).
So let's apply a signal to U1 and see what happens. For input signals below 2V peak amplitudes, U1 behaves as any common cathode stage would behave and U1 grid current is low enough to be insignificant. Once a 2V peak signal is applied, U1 grid current begins to flow at the positive signal peaks. Simulations show peaks of about 7 microamps (uA) of U1 grid current. Let's consider 2V as the threshold of U1 grid current flow. Let's call this 2V signal threshold GT1. So GT1 is the U1 input signal voltage that begins to produce grid current in U1. Let's also define an analogous GT2 value, as the U1 input signal voltage that begins to produce grid current in U2. Note, however, that GT2 is the threshold of U2 grid current caused by negative U1 input signal peaks. Note that GT1 and GT2 may be different. In this example, GT2 is 0.75V, meaning that only .74V peak signals into U1 are sufficient to cause U2 grid current (~5uA).
Once U1 grid current flows, U1 is saturating and its anode signal begins to show compression on the negative signal swing. For signals greater than GT1, U1 grid current rapidly increases and its anode signal compression and distortion also increases rapidly. Note that this compression and distortion is created on the negative swings of the U1 anode. The positive signal swing at the U1 anode is not significantly distorted by grid current in U1.
Once U1 grid current flows, U1 is saturating and its anode signal begins to show compression on the negative signal swing. For signals greater than GT1, U1 grid current rapidly increases and its anode signal compression and distortion also increases rapidly. Note that this compression and distortion is created on the negative swings of the U1 anode. The positive signal swing at the U1 anode is not significantly distorted by grid current in U1.
Since the example U1 stage has a gain of ~38, a 2V peak input produces a ~75V peak signal at the U1 anode and also at the grid of U2. Since U2 is a cathode follower, it replicates its grid signal at its cathode, which is the output of the DCCF stage. Positive signal swings at the U2 grid cause grid current in U2, which acts to compress and distort the positive signal swings at the U2 grid and at the DCCF output. In fact, with a 75V peak signal at the U2 grid, quite a bit of grid current flows in U2. A simulation measures peaks of ~220uA flowing through the U2 grid. This is over 30 times greater than the U1 grid current for the 2V peak input signal, and it causes significant distortions. An FFT (spectrum) of the output signal (below) shows a second harmonic at -25db from the main signal (in this example a 220 Hz sine wave). There is also a 3rd harmonic about 10db below the 2nd. The dominant second harmonic confirms that the waveform is asymmetrically compressed. To a first order, we can consider the higher grid current in U2 as causing an asymmetric compression of the positive signal swing of the DCCF output.
Since the example U1 stage has a gain of ~38, a 2V peak input produces a ~75V peak signal at the U1 anode and also at the grid of U2. Since U2 is a cathode follower, it replicates its grid signal at its cathode, which is the output of the DCCF stage. Positive signal swings at the U2 grid cause grid current in U2, which acts to compress and distort the positive signal swings at the U2 grid and at the DCCF output. In fact, with a 75V peak signal at the U2 grid, quite a bit of grid current flows in U2. A simulation measures peaks of ~220uA flowing through the U2 grid. This is over 30 times greater than the U1 grid current for the 2V peak input signal, and it causes significant distortions. An FFT (spectrum) of the output signal (below) shows a second harmonic at -25db from the main signal (in this example a 220 Hz sine wave). There is also a 3rd harmonic about 10db below the 2nd. The dominant second harmonic confirms that the waveform is asymmetrically compressed. To a first order, we can consider the higher grid current in U2 as causing an asymmetric compression of the positive signal swing of the DCCF output.
So, in summary, grid current in U2 acts to compress and distort the positive swings in the DCCF output, while grid current in U1 acts to compress the negative swings of the output signal. You can make an analogy of this stage to a back-to-back diode clipper, where each diode clips half the waveform. However the two tubes behave differently from each other and from diodes due to the difference in their bias and the signal levels they operate with.
So, in summary, grid current in U2 acts to compress and distort the positive swings in the DCCF output, while grid current in U1 acts to compress the negative swings of the output signal. You can make an analogy of this stage to a back-to-back diode clipper, where each diode clips half the waveform. However the two tubes behave differently from each other and from diodes due to the difference in their bias and the signal levels they operate with.
That brings us back to the optimize question of whether we can control the behavior of each tube, thereby controlling how and when the positive and negative signal swings are compressed and distorted. Not surprisingly, it can be done by adjusting the resistor values in the circuit. The commonly used 820, 100k, 100k resistor values are so prevalent, few designers even question what they actually do so we rarely see (or hear) variations. I suspect many designers simply use the circuit as a fixed-template trying to achieve a known distortion quality. While that's fine, they have (perhaps unwittingly) left plenty of opportunity for others to explore other options.
That brings us back to the optimize question of whether we can control the behavior of each tube, thereby controlling how and when the positive and negative signal swings are compressed and distorted. Not surprisingly, it can be done by adjusting the resistor values in the circuit. The commonly used 820, 100k, 100k resistor values are so prevalent, few designers even question what they actually do so we rarely see (or hear) variations. I suspect many designers simply use the circuit as a fixed-template trying to achieve a known distortion quality. While that's fine, they have (perhaps unwittingly) left plenty of opportunity for others to explore other options.
More details and methods for controlling grid currents and harmonics are found in Guitar Amplifier Design: Tubes and Semiconductors Play Together.
More details and methods for controlling grid currents and harmonics are found in Guitar Amplifier Design: Tubes and Semiconductors Play Together.
Page updated
Google Sites
Report abuse