Chapter 6

Preamplifier Architectures

Some sort of preamplifier or control unit is required in all hi-fi systems, even if its only function is to select the source and set the volume. You could even argue that the source-selection switch could be done away with, if you are prepared to plug and unplug those rhodium-plated connectors, leaving a ‘preamplifier’ that basically consists solely of a volume-control potentiometer in a box. Mind you don’t catch your sackcloth on those heat-sink corners, and try to keep the ashes away from the turntable.

I am assuming here that a selector switch will be required, and that gives us the ‘passive preamplifier’ (oxymoron alert!) in Figure 6.1(a).

Figure 6.1: Preamplifier evolution. (a) Passive preamplifier. (b) Input buffer and phono amplifier added. (c) Amplification after volume control added. (d) Amplification split into two stages, before and after volume control

Passive Preamplifiers

A device may have only one component but it does not follow that it is easy to design, even though the only parameter to decide is the resistance of the volume pot. Any piece of equipment that embodies its internal contradictions in its very name needs to be treated with caution. The pot resistance of a ‘passive preamplifier’ cannot be too high because the output impedance, maximal at one-quarter the track resistance when volume is set to –6 dB, will cause an HF roll-off in conjunction with the connecting cable capacitance. It also makes life difficult for those designing RF filters on the inputs of the equipment being driven, as described in Chapter 14.

On the other hand, if the volume pot resistance is too low the source equipment will suffer excessive loading. If the source is valve equipment, which does not respond well to even moderate loading, the problem starts to look insoluble.

If, however, we can assume that our source equipment has a reasonable drive capability, we can use a 10 kΩ pot. Its maximum output impedance (at –6 dB) will then be 2.5 kΩ. The capacitance of most audio cable is 50–150 pF/meter, so with a 2.5 kΩ source impedance and 100 pF/meter cable, a maximum length of 5 meters is permissible before the HF loss hits the magic figure of –0.1 dB at 20 kHz. A very rapid survey of current (2009) ‘passive preamplifiers’ confirmed that 10 kΩ seems to be the most popular value. One model had a 20 kΩ potentiometer, and another had a 100 kΩ pot, which with its 25 kΩ source impedance would hardly allow any cable at all. I suspect that the only reason such a pot can appear acceptable is because in normal use it is set well below –6 dB. For example, if a 100 kΩ pot is set to –15 dB, the output impedance is reduced to 14 kΩ, which would allow just under a meter of 100 pF/meter cable to be used with HF loss still limited to –0.1 dB at 20 kHz. A high output impedance also makes an interconnection more susceptible to interference unless it is totally screened at every point on its length.

Consider also that many power amplifiers have RC filters at the input, not so much for EMC immunity, but more as a gesture against what used to be called ‘transient intermodulation distortion’ (TID); but this is actually just old-fashioned slew limiting and highly unlikely in practice. These can add extra shunt capacitance to the input ranging from 100 to 1000 pF, apparently having been designed on assumption of near-zero source impedance, and this can cause serious HF roll-offs.

There is at least one passive preamplifier on the market that controls volume by changing the taps on the secondary of a transformer; this should give much lower output impedances. There is more on that approach in Chapter 9 on volume controls.

Active Preamplifiers

Once we permit ourselves active electronics, things get much easier. If a unity-gain buffer stage is added after the selector switch, as in Figure 6.1(b), the volume pot resistance can be reduced to much less than 10 kΩ, while presenting a high impedance to the sources. If a 5532 is used there is no technical reason why the pot could not be as low as 1 kΩ, which will give a much more usable maximum output impedance of only 250 Ω, and also reduce Johnson noise by 10 dB. A phono preamplifier has also been added. Now we’ve paid for a power supply, it might as well supply something else.

This still leaves us with an ‘amplifier’ that has only unity line gain. Normally only CD players, which have an output of 2 Vrms, can fully drive a power amplifier without additional gain, and there are some high-power amplifiers that require more than this for full output. iPods appear to have a maximum output of 1.2 Vrms. Output levels for tuners, phono amps, and so on vary but may be as low as 150 mVrms, while power amplifiers rarely have sensitivities lower than 500 mV. Clearly some gain would be a good thing, so one option is adding a gain stage after the volume control, as in Figure 6.1(c). The output level can be increased and the output impedance kept down to 100 Ω or lower.

This amplifier stage introduces its own difficulties. If its nominal output level with the volume control fully up is taken as 1 Vrms for 150 mV in, which will let us drive most power amps to full output from most sources most of the time, we will need a gain of 6.7 times or 16.5 dB. If we decide to increase the nominal output level to 2 Vrms, to be sure of driving most if not all exotica to its limits, we need 22.5 dB. The problem is that the gain stage is amplifying its own noise at all volume settings, and amplifying a proportion of the Johnson noise of the pot whenever the wiper is off the zero stop. The noise performance will therefore deteriorate markedly at low volume levels, which are the ones most used.

Amplification and the Gain-Distribution Problem

One answer to this difficulty is to take the total gain and split it so there is some before and some after the volume control, so there is less gain amplifying the noise at low volume settings. One version of this is shown in Figure 6.1(d). The question is: how much gain before and how much after? This is inevitably a compromise, and it might be called the gain-distribution problem. Putting more of the total gain before the volume control reduces the headroom as there is no way to reduce the signal level, while putting more after increases the noise output at low volume settings.

If you are exclusively using sources with a predictable output, of which the 2 Vrms from a CD player will be the maximum, the overload situation is well defined, and if we assume that the pre-volume gain stage is capable of at least 8 Vrms out, so long as the pre-volume control gain is less than 4 times there will never be a clipping problem. However, phono cartridges, particularly moving-coil ones, which have a very wide range of sensitivities, produce much less predictable outputs after fixed-gain preamplification, and it is a judgement call as to how much safety margin is desirable.

As an aside, it’s worth bearing in mind that even putting a unity-gain buffer before the volume control, which we did as the first step in preamp evolution, does place a constraint on the signal levels that can be handled, albeit at rather a high level of 8–10 Vrms depending on the supply rails in use. The only source likely to be capable of putting out such levels is a mixing console with the group faders fully advanced. There is also the ultimate constraint that a volume control pot can only handle so much power, and the manufacturers’ ratings are surprisingly low, sometimes only 50 mW. This means that a 10 kΩ pot would be limited to 22 Vrms across it, and if you are planning to use lower resistance pots than this to reduce noise, their power rating needs to be kept very much in mind.

Whenever a compromise appears in engineering, you can bet that someone will try to find a way round it and get the best of both worlds. What can be done about the gain-distribution dilemma?

One possibility is the use of a special low-noise amplifier after the volume control, combined with a low-resistance volume pot as suggested above. This could be done either by a discrete device and op-amp hybrid stage, or by using a multiple op-amp array, as described in Chapter 1. It is doubtful if it is possible to obtain more than a 10 dB noise improvement by these means, but it would be an interesting project.

Another possible solution is the use of double gain controls. There is an input gain control before any amplification stage, which is used to set the internal level appropriately, thus avoiding overload, and after the active stages there is an output volume control, which gives the much-desired silence at zero volume (see Figure 6.2(a)). The input gain controls can be separate for each channel, so they double as a balance facility; this approach was used on the Radford HD250 amplifier, and also in one of my early preamplifier designs [1]. This helps to offset the cost of the extra pot. However, having two gain controls is operationally rather awkward, and however attenuation and fixed amplification are arranged, there are always going to be some trade-offs between noise and headroom. It could also be argued that this scheme does not make a lot of sense unless some means of metering the signal level after the input gain control is provided.

Figure 6.2: More preamp architectures. (a) With input gain control and output volume control. (b) With recording output and return input, and an active gain control

If the input and output gain controls are ganged together, to improve ease of operation at the expense of flexibility, this is sometimes called a distributed gain control.

Active Gain Controls

The noise/headroom compromise is completely avoided by replacing the combination of volume control and amplifier with an active gain control, i.e. an amplifier stage whose gain is variable from near-zero to the required maximum (see Figure 6.2(b)). We get lower noise at gain settings below maximum, and we can increase that maximum gain so even the least sensitive power amplifiers can be fully driven, without impairing the noise performance at lower settings. We also get the ability to generate a quasi-logarithmic law from a linear pot, which gives excellent channel balance as it depends only on mechanical alignment. The only snag is that most active gain controls phase-invert. The technology is dealt with fully in Chapter 9.

Recording Facilities

If a preamplifier is going to be used for recording, the minimum requirement is an output taken from before the volume control, and there is usually also a dedicated input for a signal coming back from the recorder that can be switched to for checking purposes. Back when recording was done on tape, the return signal could be taken from the replay heads and gave assurance that recording had actually happened. Now that recording is done on hard-disk machines or PCs the return signal normally only assures you that the signal has actually got there and back.

Much ingenuity used to be expended in designing switching systems so you could listen to one source while recording another, though it is rather doubtful how many people actually wanted to do this; it demands very high standards of crosstalk inside the preamplifier to make sure that the signal being recorded is not contaminated by another source.

Tone Controls

Let us now consider adding tone controls. While it is currently unfashionable to have tone controls on a preamplifier, this seems to be starting to change. I think they are absolutely necessary, and it is a startling situation when, as frequently happens, anxious inquirers to hi-fi advice columns are advised to change their loudspeakers to correct excess or lack of bass or treble. This is an extremely expensive alternative to tone controls.

There are many possible types, as described in Chapter 10, but one thing they have in common is that they must be fed from a low-impedance source to give the correct boost/cut figures and predictable EQ curves. Another vital point is that most types, including the famous Baxandall configuration, phase-invert. Since there is now pretty much a consensus that all audio equipment should maintain absolute phase polarity for all input and outputs, this can be highly inconvenient.

However, this phase inversion can very conveniently be undone by the use of an active gain control, which also uses shunt feedback and so also phase-inverts. The tone control can be placed before or after the active gain control, but if placed afterwards it generates noise that cannot be turned down. Putting it before the active gain control reduces headroom if boost is in use, but if we assume the maximum boost used is +10 dB, the preamp inputs will not overload before 3 Vrms is applied, and domestic equipment can rarely generate such levels. It therefore seems best to put the tone control before the active gain control. This is what I did in my most recent preamplifier designs [2, 3].

References

[1]  D. Self, An advanced preamplifier, Wireless World (November 1976).

[2]  D. Self, A precision preamplifier, Wireless World (October 1983).

[3]  D. Self, Precision Preamplifier 96, Electronics World (July/August and September 1996).

Small Signal Audio Design; ISBN: 9780240521770

Copyright © 2010 Elsevier Ltd; All rights of reproduction, in any form, reserved.

..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset
44.222.87.38