ampage Tube Amps / Music Electronics	For current discussions, please visit Music Electronics Forum.

ampage archive

Vintage threads from the first ten years

More info on cascodes and diff stages

10/1/1999 6:45 PM
kg	More info on cascodes and diff stages For those who are interested in additional hard information, I have culled the net a bit for data. Here's what I've found so far. On cascode operation: From http://www.passlabs.com/cascode.htm There is a method for eliminating such nonlinearities called cascode operation, where the voltage across the transistor, tubes, or FETS is frozen at a constant value, completely eliminating voltage-induced distortions...Having essentially unity current gain, extremely wide bandwidth, and no distortion, the common base device [or common grid, as the case may be] shields the gain transistor from voltage changes in the circuit. Besides eliminating voltage caused nonlinearities, cascode operation can yield an additional benefit in increased bandwidth. Because the collector-base voltage [AKA plate-grid] is held constant there is minimal charging of the collector-base [plate-grid] junction capacitance in the transistor [tube]. Eliminating the effects of this internal lag capacitance allows higher frequency response, thus cascode circuitry is commonly found in ultra-high frequency amplifiers and wide bandwidth oscilloscopes where response is required beyond 100 megaHertz. Cascode circuitry has also found its way into preamplifier circuitry as manufactured by Dayton-Wright Paragon, DB Systems, and Audio Directions among others. From http://www.tubecad.com/march99/page2.html Review of the Cascode's Operation The Cascode is a compound amplifier. One triode stands on top of another, while sharing a common current path. The top triode strives both to shield the input grid from the top triode's Miller effect and to preserve the transconductance of the bottom triode. The result is amplifier with both high gain and extended bandwidth. Low Input Capacitance In the Grounded Cathode amplifier, the grid-to-plate capacitance is multiplied by the gain that the triode realizes working into its plate resistor. This effective increase in capacitance greatly reduces the high frequency response of the amplifier; whereas in the Cascode amplifier, the input grid-to-plate capacitance is virtually identical to its static value, as its plate voltage is held at a nearly constant value. From http://www.atma-sphere.com/twhite.htm#Cascode The advantages of cascode operation are: --High gain, as described. --Low noise for the amount of gain developed. --Good high frequency bandwidth. This varies with stray capacitance and the internal inter-electrode capacitance of the tubes used. --Low distortion as described, without the need for high current. --Low parts count for the amount of resulting gain (simple circuitry). The main disadvantage is: --High output impedance. One of the results of this is that certain tubes become unacceptable for cascode use, such as the 12AX7. --[Personally I have to add a poor PSRR to the list of disadvantgaes. This is due primarily to the high effective output impedance. However, this can be alleviated through the use of careful HV rail filtering or regulation, or in hum cancellation techniques utilizing the second (upper) grid.] Now on to differential gain stages. From http://www.atma-sphere.com/twhite.htm#Differential Differential stages advantgaes: --Greater power supply immunity. Differential amplifiers present a constant load to the power supply, resulting in less noise in the power supply. Differential amplifiers also resist input (noise) from the power supply to a much greater degree. --Lower noise. Differential amplifiers have roughly 6 dB lower noise then the same circuit executed in a single-ended manner. This can be very important in moving-coil preamp sections. --Lower distortion. Differential amplifiers tend to cancel distortions that single-ended amplifiers cannot. --Drift is reduced by the tight coupling of the two halves of the amplifier. Performance over time is improved. --Noise rejection. Common-mode rejection ratio is the measurement of a differential amplifier's ability to not amplify noise that is common to both inputs. It is typically at least 55 dB, and can approach 140 dB in some critically-tuned designs. There are also some disadvantages: --Increased cost. Differential amplification takes more parts to execute. For a given number of stages of gain, differential amplifiers have about 50% more parts. --Greater complexity. Although the number of stages of amplification remains the same for single-ended and differential amplifiers, differential amplifiers have more requirements to execute, for example, a negative-voltage power supply. More on balanced signal transmissions from http://www.tubecad.com/march99/page4.html In the balanced setup, the two input and output voltages are in anti-phase to each other, but are equal in amplitude to each other. In other words, if the two signals are summed together, the result would be silence. On the other hand, if the difference between these voltages were amplified, the sum of the two output signals would be twice the amplitude of the same signal level coming from an unbalanced source. Here is the lever behind a balanced design: amplifying the difference between anti-phase signals does more than just double the gain, it ignores what is common to both signals in phase and voltage. Balanced designs are engineered to null what is common and amplify what differs. This makes for the single great advantage that balanced designs hold over the more conventional unbalanced designs: much greater noise suppression. Noise is, for the most part, a common mode phenomenon, which is another way of saying that the noise is mostly equally shared between the two balanced voltage outputs and will dropout of the signal when processed through a balanced device, such as a transformer or preamp. Since many professional audio environments, such as recording halls, auditoriums, and nightclubs are chock full of wiring, lights, air-conditioning, and electromechanical gear, they are also filled with electromagnetic pollution. For this reason professional audio gear is usually balanced, or at the least, balanced at the beginning of the chain where noise is more problematic, i.e. the microphone and its amplifier. [End references.] In the nearest future I shall be attempting some of these tweaks to my amplifiers, and I will share what I find with the board. In short I see no reason whatsoever to be satisfied with circuits that merely suffice for the job, rather than excel at it. The fact that few or no commercially available musical instrument amps have used this kind of circuitry makes it even more promising and appealing to me. Think about it--when was the last time you laid your hands on a well engineered product? I'm not surprised they haven't gotten away from 50 year old topologies. Clearly, the benefits of a differential input stage in terms of noise performance are many. The idea of cascoding a ultra-low noise FET with a high gm tube is hardly new; the hifi guys have been doing this--very successfully--for years in phono preamps for their MM cartridges. A guitar output is much larger in signal voltage and of a lower impedance, which only means that it will offer even better performance. Good luck and success in whatever direction you're moving. Ken Gilbert

10/1/1999 6:45 PM

More info on cascodes and diff stages

For those who are interested in additional hard information, I have culled the net a bit for data. Here's what I've found so far.

On cascode operation:

From http://www.passlabs.com/cascode.htm

There is a method for eliminating such nonlinearities called cascode operation, where the voltage across the transistor, tubes, or FETS is frozen at a constant value, completely eliminating voltage-induced distortions...Having essentially unity current gain, extremely wide bandwidth, and no distortion, the common base device [or common grid, as the case may be] shields the gain transistor from voltage changes in the circuit.

Besides eliminating voltage caused nonlinearities, cascode operation can yield an additional benefit in increased bandwidth. Because the collector-base voltage [AKA plate-grid] is held constant there is minimal charging of the collector-base [plate-grid] junction capacitance in the transistor [tube]. Eliminating the effects of this internal lag capacitance allows higher frequency response, thus cascode circuitry is commonly found in ultra-high frequency amplifiers and wide bandwidth oscilloscopes where response is required beyond 100 megaHertz. Cascode circuitry has also found its way into preamplifier circuitry as manufactured by Dayton-Wright Paragon, DB Systems, and Audio Directions among others.

From http://www.tubecad.com/march99/page2.html

Review of the Cascode's Operation

The Cascode is a compound amplifier. One triode stands on top of another, while sharing a common current path. The top triode strives both to shield the input grid from the top triode's Miller effect and to preserve the transconductance of the bottom triode. The result is amplifier with both high gain and extended bandwidth.

Low Input Capacitance

In the Grounded Cathode amplifier, the grid-to-plate capacitance is multiplied by the gain that the triode realizes working into its plate resistor. This effective increase in capacitance greatly reduces the high frequency response of the amplifier; whereas in the Cascode amplifier, the input grid-to-plate capacitance is virtually identical to its static value, as its plate voltage is held at a nearly constant value.

From http://www.atma-sphere.com/twhite.htm#Cascode

The advantages of cascode operation are:
--High gain, as described.
--Low noise for the amount of gain developed.
--Good high frequency bandwidth. This varies with stray capacitance and the internal inter-electrode capacitance of the tubes used.
--Low distortion as described, without the need for high current.
--Low parts count for the amount of resulting gain (simple circuitry).
The main disadvantage is:
--High output impedance. One of the results of this is that certain tubes become unacceptable for cascode use, such as the 12AX7.
--[Personally I have to add a poor PSRR to the list of disadvantgaes. This is due primarily to the high effective output impedance. However, this can be alleviated through the use of careful HV rail filtering or regulation, or in hum cancellation techniques utilizing the second (upper) grid.]

Now on to differential gain stages.

From http://www.atma-sphere.com/twhite.htm#Differential

Differential stages advantgaes:
--Greater power supply immunity. Differential amplifiers present a constant load to the power supply, resulting in less noise in the power supply. Differential amplifiers also resist input (noise) from the power supply to a much greater degree.
--Lower noise. Differential amplifiers have roughly 6 dB lower noise then the same circuit executed in a single-ended manner. This can be very important in moving-coil preamp sections.
--Lower distortion. Differential amplifiers tend to cancel distortions that single-ended amplifiers cannot.
--Drift is reduced by the tight coupling of the two halves of the amplifier. Performance over time is improved.
--Noise rejection. Common-mode rejection ratio is the measurement of a differential amplifier's ability to not amplify noise that is common to both inputs. It is typically at least 55 dB, and can approach 140 dB in some critically-tuned designs.
There are also some disadvantages:
--Increased cost. Differential amplification takes more parts to execute. For a given number of stages of gain, differential amplifiers have about 50% more parts.
--Greater complexity. Although the number of stages of amplification remains the same for single-ended and differential amplifiers, differential amplifiers have more requirements to execute, for example, a negative-voltage power supply.

More on balanced signal transmissions from http://www.tubecad.com/march99/page4.html

In the balanced setup, the two input and output voltages are in anti-phase to each other, but are equal in amplitude to each other. In other words, if the two signals are summed together, the result would be silence. On the other hand, if the difference between these voltages were amplified, the sum of the two output signals would be twice the amplitude of the same signal level coming from an unbalanced source.

Here is the lever behind a balanced design: amplifying the difference between anti-phase signals does more than just double the gain, it ignores what is common to both signals in phase and voltage. Balanced designs are engineered to null what is common and amplify what differs. This makes for the single great advantage that balanced designs hold over the more conventional unbalanced designs: much greater noise suppression.
Noise is, for the most part, a common mode phenomenon, which is another way of saying that the noise is mostly equally shared between the two balanced voltage outputs and will dropout of the signal when processed through a balanced device, such as a transformer or preamp. Since many professional audio environments, such as recording halls, auditoriums, and nightclubs are chock full of wiring, lights, air-conditioning, and electromechanical gear, they are also filled with electromagnetic pollution. For this reason professional audio gear is usually balanced, or at the least, balanced at the beginning of the chain where noise is more problematic, i.e. the microphone and its amplifier.

[End references.]

In the nearest future I shall be attempting some of these tweaks to my amplifiers, and I will share what I find with the board.

In short I see no reason whatsoever to be satisfied with circuits that merely suffice for the job, rather than excel at it. The fact that few or no commercially available musical instrument amps have used this kind of circuitry makes it even more promising and appealing to me. Think about it--when was the last time you laid your hands on a well engineered product? I'm not surprised they haven't gotten away from 50 year old topologies.

Clearly, the benefits of a differential input stage in terms of noise performance are many. The idea of cascoding a ultra-low noise FET with a high gm tube is hardly new; the hifi guys have been doing this--very successfully--for years in phono preamps for their MM cartridges. A guitar output is much larger in signal voltage and of a lower impedance, which only means that it will offer even better performance.

Good luck and success in whatever direction you're moving.

Ken Gilbert

Replies:

Randall Aiken For what it is worth, I am a big fa... -- 10/1/1999 11:52 PM