James Tanner - Bryston
New member
- Joined
- Apr 10, 2013
- Messages
- 1,874
- Thread Author
- #1
BRYSTON PHONO STAGE PHILOSOPHY
It is evident that many of the perceived inaccuracies in LP recordings are a result of flaws in the playback electronics. The task of simultaneously equalizing and amplifying a signal an average of about 40dB, presents many interesting challenges to the designer. In fact, the task is so complex that it may be only recently that it has been done correctly.
The most immediate difficulty in deriving the RIAA phono equalization curve lies in the extremes of gain which must be encompassed within the passband. Between 20Hz and 20kHz, there is a change of close to 40dB, or about 100:1. When this is performed by a single amplification stage, as it normally is done, it means that reactive components following this impedance ratio are contained inside the feedback loop of one amplifier. This presents many complications to the amplifier stage, and causes several types of distortions and potential pitfalls. The most important of these is that those reactive impedances are likely to be at least partially reflected to the input terminals. Since a phono cartridge is a reactive device, this causes unpredictable and often rather extreme alterations to the high-frequency response of the system, due to resonances and cancellations between the inductance and capacitance involved.
There are many other problems involved in deriving a phono preamplifier in the above manner, which almost never show up in the specifications, but which have serious audible effects. For instance, the low impedance values reflected by the feedback components at high frequencies become a difficult load on the amplifier stage, and lead to premature overload at frequencies above about 5KHz. Since headroom figures are normally stated only at 1kHz, you might not suspect why many records sound wooden instead of sparkling. This is exacerbated further by the fact that these phono stages are invariably connected in the non-inverting configuration for impedance and noise reasons. A non-inverting amplifier cannot be brought below unity gain within its feedback loop, and thus departs from true RIAA above 20kHz. Since frequency response specifications normally do not extend above 20kHz, this is not apparent in the specs, but it further aggravates the problem of high frequency transient overload. With almost all standard phono stages, when this overload occurs, the amplifier becomes unstable at low frequencies. This results in a phenomenon called “motorboating”, wherein the amplifier emits low-frequency pulses. This obviously adds to the transient inaccuracy of the sound picture.
There are other, more subtle shortcomings with the normal methods of deriving a phono preamp. The low-frequency gain is usually more than 60dB, which leaves very little feedback to reduce distortion and gain errors. The high-frequency gain reaches unity, which due to non-inverting amplifier theory, imposes a common-mode voltage on the inputs equal to the entire output signal. This partially overloads the amplifier’s common-mode rejection ability, increasing harmonic and intermodulation distortions regardless of the amount of feedback employed.
If the above were not enough, consider that the many components in the standard RCAA compensation network of most phono preamps will interact with each other in such a way as to make true mathematical analysis quite tedious and difficult. In fact, without computer-design of such networks it is highly unlikely to be done correctly even in the most expensive preamps. It is still rare to find a phono preamp without “bumps and dips” in the equalized frequency-response plot. Finally, most phono preamplifiers use Integrated-Circuit operational amplifiers to derive their gain-function. Unfortunately, these ICs have a maximum voltage-sustaining ability of +/-18v, which limits headroom. They also often have insufficient current-output available to drive the capacitance in long cables, and narrow stability margins at the low gain levels phono stages employ at high frequencies. This results in poor slew-rate due to the lag compensation necessary for stability of the complex circuitry, and a host of other compromises due to the integrated-circuit being designed for the widest possible range of applications.
While the above may make it seem impossible to accurately amplify and equalize a phono signal, rest assured it can be done, and is being done. Bryston Ltd. manufactures a phono preamplifier which employs phenomenally accurate circuitry in our audiophile preamplifiers. This novel and fundamentally correct circuit divides the equalization and amplification into two stages. The first, a non-inverting amplifier for noise and impedance reasons, has flat frequency response from 500Hz to several octaves above the audio band. Below 500Hz it has the requisite bass boost for the RIAA curve in that range. This provides the equivalent of a flat “buffer” amplifier, with 20dB of mid-band gain, so the phono cartridge is fed into a linear, stable, non-reactive load. Thus, the frequency response is completely predictable. Since the low-frequency gain in this stage is only a total of 40dB, distortion remains extremely low.
The second section of amplification is an inverting stage with a relatively low input impedance. This is possible because it is driven from the high-current output of the first stage, rather than by the phono cartridge. Thus it retains quite low noise figures. Much more important, it allows the high-frequency portion of the RIAA curve (a 6dB/octave rolloff beginning at 2120Hz) to be continued accurately beyond 20kHz and to below unity gain. Also, since inverting operational-amplifier theory imposes no common-mode voltage on the inputs, this phono stage is not subject to that source of distortion.
There are many other advantages to this mode of operation; each stage has about 20dB of mid-band gain, so their total distortion is much lower than a single equivalent stage with 40dB of mid-band gain, a mathematically calculated average of over 7 times lower, all else being equal. This incidentally frees the designer to use a much less complex amplifier circuit for each stage, one which has inherent stability to below unity gain, and which requires no internal lag compensating networks. Both are discrete amplifier gain-blocks which are designed specifically for the purpose. They have enough current available to drive loads as low as 300 ohms, and long cable-runs, without increasing distortion. They are powered from +/-24 Volts for increased headroom. They can develop the same output voltage at 20kHz as at 20Hz, so they do not overload prematurely at high frequencies. (In fact it is impossible to overload this phono stage with any record/cartridge combination we know of, but if driven to clipping by a signal generator, it will not “motorboat” under any circumstances). The respective networks for high and low-frequency EQ have no contact with each other, thus they cannot interact. They can therefore be calculated quite easily and quite exactly, and are very simple to maintain within close tolerances (+/-.05dB of true RIAA) in production.
The LP recording continues to show strength in the high-end audio marketplace, and to deliver superb quality musical entertainment. Bryston’s Phono Preamplifier provides an irreplaceable tool in the quest for beautiful music reproduction.
It is evident that many of the perceived inaccuracies in LP recordings are a result of flaws in the playback electronics. The task of simultaneously equalizing and amplifying a signal an average of about 40dB, presents many interesting challenges to the designer. In fact, the task is so complex that it may be only recently that it has been done correctly.
The most immediate difficulty in deriving the RIAA phono equalization curve lies in the extremes of gain which must be encompassed within the passband. Between 20Hz and 20kHz, there is a change of close to 40dB, or about 100:1. When this is performed by a single amplification stage, as it normally is done, it means that reactive components following this impedance ratio are contained inside the feedback loop of one amplifier. This presents many complications to the amplifier stage, and causes several types of distortions and potential pitfalls. The most important of these is that those reactive impedances are likely to be at least partially reflected to the input terminals. Since a phono cartridge is a reactive device, this causes unpredictable and often rather extreme alterations to the high-frequency response of the system, due to resonances and cancellations between the inductance and capacitance involved.
There are many other problems involved in deriving a phono preamplifier in the above manner, which almost never show up in the specifications, but which have serious audible effects. For instance, the low impedance values reflected by the feedback components at high frequencies become a difficult load on the amplifier stage, and lead to premature overload at frequencies above about 5KHz. Since headroom figures are normally stated only at 1kHz, you might not suspect why many records sound wooden instead of sparkling. This is exacerbated further by the fact that these phono stages are invariably connected in the non-inverting configuration for impedance and noise reasons. A non-inverting amplifier cannot be brought below unity gain within its feedback loop, and thus departs from true RIAA above 20kHz. Since frequency response specifications normally do not extend above 20kHz, this is not apparent in the specs, but it further aggravates the problem of high frequency transient overload. With almost all standard phono stages, when this overload occurs, the amplifier becomes unstable at low frequencies. This results in a phenomenon called “motorboating”, wherein the amplifier emits low-frequency pulses. This obviously adds to the transient inaccuracy of the sound picture.
There are other, more subtle shortcomings with the normal methods of deriving a phono preamp. The low-frequency gain is usually more than 60dB, which leaves very little feedback to reduce distortion and gain errors. The high-frequency gain reaches unity, which due to non-inverting amplifier theory, imposes a common-mode voltage on the inputs equal to the entire output signal. This partially overloads the amplifier’s common-mode rejection ability, increasing harmonic and intermodulation distortions regardless of the amount of feedback employed.
If the above were not enough, consider that the many components in the standard RCAA compensation network of most phono preamps will interact with each other in such a way as to make true mathematical analysis quite tedious and difficult. In fact, without computer-design of such networks it is highly unlikely to be done correctly even in the most expensive preamps. It is still rare to find a phono preamp without “bumps and dips” in the equalized frequency-response plot. Finally, most phono preamplifiers use Integrated-Circuit operational amplifiers to derive their gain-function. Unfortunately, these ICs have a maximum voltage-sustaining ability of +/-18v, which limits headroom. They also often have insufficient current-output available to drive the capacitance in long cables, and narrow stability margins at the low gain levels phono stages employ at high frequencies. This results in poor slew-rate due to the lag compensation necessary for stability of the complex circuitry, and a host of other compromises due to the integrated-circuit being designed for the widest possible range of applications.
While the above may make it seem impossible to accurately amplify and equalize a phono signal, rest assured it can be done, and is being done. Bryston Ltd. manufactures a phono preamplifier which employs phenomenally accurate circuitry in our audiophile preamplifiers. This novel and fundamentally correct circuit divides the equalization and amplification into two stages. The first, a non-inverting amplifier for noise and impedance reasons, has flat frequency response from 500Hz to several octaves above the audio band. Below 500Hz it has the requisite bass boost for the RIAA curve in that range. This provides the equivalent of a flat “buffer” amplifier, with 20dB of mid-band gain, so the phono cartridge is fed into a linear, stable, non-reactive load. Thus, the frequency response is completely predictable. Since the low-frequency gain in this stage is only a total of 40dB, distortion remains extremely low.
The second section of amplification is an inverting stage with a relatively low input impedance. This is possible because it is driven from the high-current output of the first stage, rather than by the phono cartridge. Thus it retains quite low noise figures. Much more important, it allows the high-frequency portion of the RIAA curve (a 6dB/octave rolloff beginning at 2120Hz) to be continued accurately beyond 20kHz and to below unity gain. Also, since inverting operational-amplifier theory imposes no common-mode voltage on the inputs, this phono stage is not subject to that source of distortion.
There are many other advantages to this mode of operation; each stage has about 20dB of mid-band gain, so their total distortion is much lower than a single equivalent stage with 40dB of mid-band gain, a mathematically calculated average of over 7 times lower, all else being equal. This incidentally frees the designer to use a much less complex amplifier circuit for each stage, one which has inherent stability to below unity gain, and which requires no internal lag compensating networks. Both are discrete amplifier gain-blocks which are designed specifically for the purpose. They have enough current available to drive loads as low as 300 ohms, and long cable-runs, without increasing distortion. They are powered from +/-24 Volts for increased headroom. They can develop the same output voltage at 20kHz as at 20Hz, so they do not overload prematurely at high frequencies. (In fact it is impossible to overload this phono stage with any record/cartridge combination we know of, but if driven to clipping by a signal generator, it will not “motorboat” under any circumstances). The respective networks for high and low-frequency EQ have no contact with each other, thus they cannot interact. They can therefore be calculated quite easily and quite exactly, and are very simple to maintain within close tolerances (+/-.05dB of true RIAA) in production.
The LP recording continues to show strength in the high-end audio marketplace, and to deliver superb quality musical entertainment. Bryston’s Phono Preamplifier provides an irreplaceable tool in the quest for beautiful music reproduction.
