Chapter 4: Preamplifiers

In the past, a preamplifier was thought to be comprised of a phono and line section together in a single chassis. When vinyl playback went by the wayside in favor of digital, we stopped putting phono sections in preamplifiers, and started calling them line stages. Now with the return of vinyl playback phono sections have come back. However, we have done it a bit differently this time around where in some cases people want to have the phono section and line section in a separate chassis. Again, we can still put these two components together in a single chassis, and frankly, there is no downside to doing that. It is not like when you are integrating a preamplifier and a power amplifier together, where there might be some concerns about power supply interference. Putting two preamplifier components together in one box is actually a good idea.

We are going to talk about the line section first because it is a bit simpler than the phono section. Before digital the line section of a preamplifier typically had a gain of around 20 dB. Today, with high-level digital sources, 20 dB of gain may be too much. Some preamplifiers have a way to reduce the gain and I found that about 12 dB of gain in a line stage is about the right amount for a typical system. If you have too much gain in the line stage, you can get into a problem. The problem is that in most any line stage the volume control is the very first thing that is encountered. Of course, the selector switch sits in front of that but once you have selected your input it is going directly to the volume control. This is done because if we put the gain stage first before the volume control, we would probably overload it with a 2V signal from a digital source.

However, as the line section begins with the volume control this presents a problem in that the line stage, if the gain is too high, will have a certain amount of output noise. I found that about 20 microvolts of output noise are about all that can really be tolerated in most systems. If you have too much noise coming out of the preamplifier and you have a power amplifier of unusually high gain, or if you have a speaker of unusually high sensitivity, or both of those together, you may hear a good amount of hiss at the speaker. There is a way to solve this problem. One can simply put a resistive attenuator of about 12 dB in between the preamplifier and the power amplifier, and that will lower the hiss coming out of the speaker.

There is one problem with doing this in that now the preamplifier will have to run a little higher-level signal and depending on the characteristics of this preamplifier it may produce a bit more distortion. The ideal situation would be to lower the gain of the line section itself. That is usually done by feedback, and again, I do not think there is anything wrong with using feedback. In fact, in this case the feedback would do several good things. One is it would lower the output impedance which means it could drive cables better and another is it would lower the distortion which I think is always a good thing. In transistor line stages the distortion usually stays constant until you get up to clipping. In tube line stages, especially those with no feedback, the distortion rises linearly with frequency. As a bogey value, a tube like a 6922 will have about 0.1% distortion at 1V output.

If you did use a resistive attenuator of 12 dB this would mean you would have a 4-volt output and your distortion would rise from 0.1% to 0.4% which is still not a terribly high value to be concerned about. If you use too much attenuation you may have a line stage that runs out of signal level, but that is unlikely considering most power amplifiers only need about 1V to get to full output. If you do have hiss or hum in your speakers with the volume control all the way down, then it is clearly noise from the line stage. Now, some of this noise may be picked up in the cable, but that would typically be hum, not hiss. One other thing to remember here is that power amplifiers of higher power have higher gain than amplifiers of lower power. If you have a very sensitive speaker, you really should not have an amplifier of terribly high power because you do not need this power.

If you think about the reason for this, the industry standard for a long time was to make power amplifiers that have full power at 1V of input. For example, if you have a 5-watt amplifier that puts out about 4V that is a gain of four. On the other hand, If you have a 100-watt amplifier that puts out 28 volts that is a gain of about 28. Note that I am speaking of gain factor here, not decibels. If you have a 400-watt amplifier that would have an output voltage of about 50V and a gain of 50. As you can see, as the amplifier power goes up the gain of the amplifier gets larger and that could be a problem with a high-sensitivity speaker. Now, if you have a high-powered amplifier the reason you should have it is because you have a low-sensitivity speaker that needs a lot of power. In that case you would not have a hiss problem because the speaker has low sensitivity. It is important to think about the power level of amplifiers along with your speaker as this can affect the noise of your system.

I hear audiophiles express a lot of concern about matching input and output impedance. I think this is a bit overdone but let me speak to it now. The output impedance of most preamplifiers is low, especially if they are solid state. A solid-state preamplifier can easily have an output impedance of 50 ohms. Now that does not mean you should put a 50-ohm load on it. Also, that does not mean it has a lot of current. It means when you put a resistor equal to the output impedance of the device across that device, the output will go down by half which is 6 dB. That is the definition of output impedance, and of course, you must do this at a low enough level that you are not clipping the signal that is driving this load resistor. We are talking millivolts.

Now a tube preamplifier could have significantly higher output impedance. I have made preamplifiers with 6922 tubes where the output impedance of that tube is about 3K ohms if there is no feedback. You can pretty much count on that number. Appling 6 dB of feedback drops that output impedance in half to 1500 ohms. Applying 20 dB of feedback drops that output impedance to 300 ohms. Now these are low impedance values. If you are using a 12AX7 you will not get numbers anywhere close to that. The 12AX7 will be much higher impedance and much less current. The output impedance of a 12AX7 is about 50K ohms. Of course, if you add feedback, you might get that down to 5K ohms, but you are still higher than the 6922 which is the reason a lot of people use this tube. A 6SN7 is also a popular tube used in a preamplifier, but it also is not low impedance.

The reason people like 6SN7 and 12AU7 tubes is because they have inherently lower gain, typically around a gain of 20, but that is still high for a line section. Remember a gain of 20 would be 26 dB. We have already to some extent decided that 20 dB is too high. We have this problem now in the industry where people are making line stages without feedback because the consumer thinks feedback is a bad thing. This is something I am trying to correct. In this case without feedback, you have a line stage with too much gain and usually too much output impedance. Let us talk about where output impedance gets us into trouble. Say you are going to drive a short cable and the capacitance of the cable is low, at around 100 picofarads per foot. If you have a 1m long cable, then the total capacitance will be about 300 pF. That should be no problem to drive.

However, if you have some very long cables, and I have seen cables that get close to 1000 pF, you would best be advised to calculate where the roll off would be with that much cable capacitance. This is a very easy thing to calculate, and it really has nothing to do with what is on the other end. It is the preamplifier driving this long cable with high enough capacitance that will eventually roll off the high frequencies at some point. This information I am imparting right now is a primary reason in my opinion as to why audiophiles might hear differences in the interconnect cables they are using in their system. They are hearing the coloration the cable imparts because of frequency roll off due to the total capacitance of the cable.

In addition, if the input impedance of the power amplifier is too low and value of the output coupling capacitor in the preamplifier is small you could have bass roll off. You can be fairly certain that if you have a tube preamplifier with a small output coupling capacitor driving a 10K ohm load the bass will be rolled off due to that load. Now most solid-state power amplifiers will have lower input impedance than tube power amplifiers. A standard transistor power amplifier will have an input impedance of 22, 33, or maybe 47K ohms. Although you may see some that are 100K ohms these days. Solid state power amplifier circuits gravitated toward rather low input impedances because of internal circuit design considerations. This was also acceptable because most solid-state preamplifiers were going to be driving them in a complete solid-state chain so that was not going to be much of a problem. You certainly could get into a problem where you have a tube preamplifier of high output impedance driving a solid-state power amplifier of low impedance. One should be careful about that.

There is also the thought that the impedance ratio between the power amplifier and preamplifier should be 10-to-1. In other words, if a preamplifier has a 1K ohm output impedance you could use a power amplifier with an input impedance as low as 10K ohms. While in general this may result in the best case scenario, I do not think you have to follow this guideline. In many cases 5-to-1 or even 2-to-1 may be acceptable because there are other considerations. For example, how much current can the preamplifier put out and how much voltage does the power amplifier need to reach full power. The largest problem you may find when you are using a preamplifier with a tube line section to drive a solid-state power amplifier is the mismatch in impedance where, yes, the power amplifier may have a low impedance such as 10K ohms and the output impedance of the preamplifier may be rather high, or it may be rather low, but in either case may not able to deliver enough current.

Now some early tube preamplifiers such as the Audio Research SP-3 or a Marantz Model 7 have a 12AX7 or a 12AU7 output tube that may be running low current. For example, at one point I measured an Audio Research SP-3 and found that it only put out about 1 milliamp of output current. If you have a low impedance power amplifier there may be some instances where you  may be demanding at full power more current than the preamplifier is comfortable producing. Later preamplifiers using 6922s usually had 3, 5, or even up to 10 mA of current in the output tube which gives you more current for the preamplifier to put out. Most solid-state preamplifiers can produce quite a bit of current because the output section typically includes a few transistors or even an opamp. So, the mismatch in impedance must be understood on several levels.

However, the current issue is a little bit harder to determine because most preamplifier manufacturers do not publish the specification for the output current. However, there are ways to find this out. For example, John Atkinson often makes measurements where he does load a preamplifier with 600 ohms to see what would happen, but that is a pretty low load. It might be fairer to load it with 10K ohms, but as I mentioned earlier, if the preamplifier has an output coupling capacitor this must be taken into consideration as well. Many of the older preamplifiers will not have a 5 mF or 10 mF output coupling capacitor that you see in a modern preamplifier. They might only have 1/2 mF, or some might even have as low as 1/10 mF because these earlier tube preamplifiers were meant to drive tube power amplifiers which generally have input impedances from 100K ohms all the way up to 1 mega ohm in some cases. With solid-state amplifiers it is very hard to have a high input impedance. In most cases you must include a buffer in front of the power amplifiers input stage to get a high input impedance and many designers do not want to add this extra stage. If you have this problem of running out of current, then it is not entirely an impedance issue, although we must look at it to some extent from that perspective.

If you run out of current, you are going to have increased distortion. Even if you do not run out of current the distortion will increase especially with preamplifiers using no feedback. This again is an advantage of feedback because it tends to keep the distortion and the output impedance low even when you load it. As I mentioned before a no feedback preamplifier using a 6922 can generate about 0.1% distortion at 1V unloaded. Let us say you have a load of 10K ohms, now that distortion increases by a factor of 3 or 4. In addition, this will generally be 2nd order harmonic distortion which, although some people think is benign, I am not terribly fond of because this also means you have a lot of intermodulation distortion. Again, based on the rest of your system and the resolution of your speakers you may not hear the difference from the increase in distortion.

Now let us talk about balanced versus unbalanced. There is a great deal of confusion on this matter. Might I say a couple things just to get your interest and hopefully your attention. There are a lot of people today making phono preamplifiers with balanced inputs and even fully differential circuits from input through output. They are doing this claiming they will have lower noise. What I want you to know is that using the same components, whether it is a transistor or a tube or an opamp, the noise of a phono preamplifier that is balanced will be 3 dB higher than the noise of the same components unbalanced. Sorry to inform you of that but that is the way it is. The reason for this is when you do a balanced input you must have an extra input device. That extra input device causes 3 dB more noise. It is as simple as that. Now of course some balanced phono preamplifiers may be done with a transformer and that is a whole different situation. A transformer input always helps the noise factor because the transformer has gain. With most moving coils we use a transformer. It is very challenging to make a phono preamplifier that goes down to moving coil levels without using a transformer, but I have done this. The Music Reference RM-4 used a 6922, both sections in parallel, and achieved noise low enough to run cartridges of 0.25 to 0.5mV.

Now getting back to balanced and unbalanced. On the output of a preamplifier, running balanced to balanced, or unbalanced to unbalanced, is only going to make a very small difference in noise. Here, I am referring to noise as hiss, not hum. If you have hum you probably have hum because of a ground loop between the chassis of the preamplifier and power amplifier. While this can be cured with a balanced circuit, you could also just as easily cure the problem using unbalanced cables and lifting the ground on everything in the system except the power amplifier. Now some people are very reluctant to lift grounds on pieces of equipment out of fear of safety and electrical shock, and this is certainly possible. However, older equipment such as Marantz or early Audio Research did not have grounds. Even some modern preamplifiers today do not have grounds. If the power amplifier is grounded and the fault occurs anywhere else in the system, and we are talking about a fault where the power transformer may short to the chassis or some wire shorts to the chassis, we want to have a path to ground, so the whole system does not go hot. The power amplifier will provide that path to ground and most likely blow the circuit breaker and then you will know that you have a piece of equipment shorted to ground. So this practice we have about adding grounded power cords to our equipment is not necessarily a good thing.

Earlier we were talking about noise in balanced versus unbalanced circuits. Now, I want to talk a bit about noise with balanced versus unbalanced cables. Some people are now referring to unbalanced cables as single-ended which I think is incorrect because we have amplifier circuits that are called single-ended. The correct definition is either balanced or unbalanced. Also, some people are fond of unshielded unbalanced cables, although I really do not know why, as they are an invitation for noise. However, if you have an unshielded unbalanced cable, it should be a twisted pair if you want to increase noise rejection. This may be the one instance where a shielded balanced cable in a balanced circuit makes a significant difference in noise rejection. Again, noise in this case is going to be hum, buzz, clicks and other things coming into your system, but not hiss.  It is impossible for a cable to introduce hiss in the system. Hiss is totally caused by input devices and output noise.

Let us get on to this most interesting topic of phono preamplifiers. Traditionally, a preamplifier was a combination of a phono section and a line section. Phono sections usually had 40 dB of gain measured at 1K Hz because at other frequencies the gain will be higher or lower due to the equalization. The line section typically had 20 dB of gain for a total gain of 60 dB. Now back in those days line level sources tended to be about 0.5V at full level. In fact, when the standard on many tape recorders was 0VU on the meter the output voltage was exactly .775V which is the standard that the telephone company adopted around 1900 to set the level on their long-distance phone lines. It turns out that voltage is a rather odd number. It comes from the fact that the phone lines were 600 ohms because they are 19-gauge wires spaced at 6 inches and if you do some research, you will also find they are twisted.  Yet although a phone line may be twisted only every half mile they are still balanced. It turns out that the voltage into 600 ohms produces exactly 1 mW, and 1 mW is a nice round number to deal with. That voltage is called 0DBM, the M standing for milliwatt. There is also 0DBV which is what many engineers have adopted because that is 1V, period.

This has nothing to do with load impedance. It is just a comfortable number to work with in the laboratory to say this is 1V, so we call that 0DBV, and the difference is not very large. Now, Sony and Philips when they designed the CD player for some reason decided to have 2V be the maximum output. If we think about this the .75V standard for 0VU, which is also the standard for 0VU in all studios, may have led them to determine that the CD should have 12 dB of headroom because 12 dB would get you up to about 2V. Of course they did not know that the loudness wars were coming allowing recording engineers to make CDs as loud as possible, so now we have a lot of recorded music putting out close to 2V from a CD player.

The result was now in the digital world a standard preamplifier had too much gain and we needed to run the volume control at a very low level, and this is not ideal. One of the ways to cure this problem, and probably the best way, is to install about a 12 dB resistive attenuator at the output of the digital source before you go into a standard preamplifier and then two things will happen. One is you will be using the volume control in a higher setting which is always better. The other is you will also better match your other sources if you still have a tuner, tape deck, or even a rather traditional phono preamplifier.

Now because moving coil cartridges have about 20 dB less output signal than a moving magnet cartridge, we typically need 20 dB more gain for the phono preamplifier. This can be provided by a transformer. If you look at most transformers, they can have a step-up ratio of about 10 that provides 20 dB of gain. You might also find a phono preamplifier that has adjustable gain, such as 40, 50, and 60 dB. Again, you are getting 20 dB of gain over the standard 40 dB. Now phono preamplifiers also must have equalization. If you are familiar with the RIAA curve (as it is referenced today) it has about 20 dB of boost at low frequencies and about 20 dB of cut at high frequencies while being rather flat in the middle between 500 and 2K Hz.

While people today refer to this as the RIAA curve this is incorrect. The original RIAA curve only has a total of 12 dB of equalization, not 40 dB. The equalization is such that below 500 Hz there is a cut of 6db and above 2K Hz there is a boost of 6db making a total of 12 dB. Of course, the reason they did this is so that on playback they could cut the highs, thereby reducing surface noise. In addition, they boost the bass, because by cutting the bass on the recording end they can make the groove smaller as low-frequency grooves could get rather large if the bass is not cut. So, this is the real RIAA equalization curve. It is a total of 12 dB, not 40 dB. You are welcome to argue this with your friends. The curve that we reference today that we call the RIAA curve is the original RIAA curve plus the equalization needed for a velocity sensitive cartridge. For reference, all your magnetic cartridges, whether moving coil, moving magnet, or moving iron are velocity sensitive cartridges.

If you happen to have a displacement cartridge such as a crystal cartridge, light-sensitive cartridge, capacitive cartridge, or a Strain-Gauge cartridge, all of these are amplitude sensitive cartridges and only require 12 dB of equalization. Unfortunately, some of the makers of these cartridges have tried to build the equalization into the cartridge. WinLabs tried to build the equalization into the cartridge mechanically, but instead they should have built it in electrically. Believe me I tried to get them to do it. There are also some recent cartridges that have come on the market that are light-dependent. Manufacturers have also made some mistakes in the equalization required for these cartridges as reported in some of the audio magazines.

When we start to talk about phono cartridges, the topic of loading is certainly a hot topic. Moving magnet and moving iron cartridges which are your higher output cartridges have generally been designed to run into a 47K ohm load. Grado was an exception as they typically used a 10K ohm load. Also, capacitance and load capacitance affect moving magnet cartridges much more than moving coil cartridges. The Shure cartridges are the most notorious about this because they are very high inductance and even the cable capacitance can start to become a problem. Most moving magnet cartridges want a few hundred picofarads of cable, or total capacitance. The majority of that will be found in the cables but also to an extent in the capacitance value at the moving magnet phono input of the preamplifier as well. You can easily get in trouble having too much cable capacitance. Certainly, over 150 pF is going to start to affect things and in some unusual ways.

One way we can count on capacitance to affect the response of the cartridge is by causing early high frequency roll off. Instead of getting response out to 20K Hz you only get response out to 10K Hz. On a highly inductive cartridge you can run into another problem which is the cable capacitance along with the inductance of the cartridge causing a peak in the response. You may get a peak at 5K Hz that is several decibels high. Generally, we do not want peaks in our system. Also, when something peaks in that way it is a resonant peak which means after the peak it falls off very sharply at 12 dB per octave. That is something we want to watch out for as well. I would say that with any moving magnet cartridge the goal should be to always have the minimum cable capacitance that you can find. Though with a low output moving coil cartridge the impedance is so low, some cartridges like the Lyra are as low as 6 ohms while the Denon 103 is as high as 40 ohms, cable capacitance is hardly ever a problem. You would have to get up close to 1000 pF before cable capacitance would become an issue. It is also worth noting a moving coil phono input on a preamplifier will not have a capacitance value.

Load impedance becomes a bit of an issue as it varies with different cartridges. I have discovered that low impedance cartridges such as the Lyra are hardly affected by load at all. In fact, they really do not need a load and I am sure there will be people that disagree with me, but experimenting with loads on a very low impedance cartridge will not produce many results. On a higher impedance cartridge such as the Denon 103 I do find that there is a quite audible effect of loading and typically a cartridge like that likes a load of around 40 to 100 ohms. Do not expect to be hearing the difference between 100 and 110 ohms. You really must make some fairly large changes to have a significant change in the sound of the cartridge. The loading does two things on a moving coil cartridge. One, it tends to damp the stylus motion because the cartridge generating its voltage running into the load causes a back EMF that damps the motor. The other effect, of course, is that many moving coil cartridges have a rising frequency response, and this will tend to tame the rising response.

In summary, let us look at a few things we have talked about in this chapter. A preamplifier is the combination of a phono section and a line section in a single chassis. The two can also be separate components known as a phono preamplifier and line stage. The matching of the output impedance of a preamplifier to the input impedance of the power amplifier can sometimes be a big issue, but not necessarily one to be worried about too much. We find that the impedance alone is not the whole story, but we also must consider the voltage and current levels involved. That is a little harder to determine, but of course if you have a very low impedance preamplifier driving a very high impedance power amplifier, those issues tend to go away and things become quite a bit simpler to manage. We should be more concerned when we have a tube preamplifier driving a solid-state power amplifier. Then we want to investigate the impedance and current capabilities. Keep in mind that the cable, if it is high capacitance, can roll off the high-end, and if the power amplifier has low impedance, that can roll off the low-end. Those two things are mutually exclusive. One does not affect the other. We can also say that balanced, although being a somewhat good thing, is not necessary to achieve low noise and good sound. You can do very well with unbalanced and if you have hum with unbalanced you can generally cure that by lifting grounds on everything but the power amplifier.

Also, keep in mind that most modern systems have too much gain. It is rare to see a system with too little gain. When there is too much gain you have the possibility of having noise, mostly hiss, but also hum is possible. If you could lower the gain of the system, you could probably easily eliminate that noise and you can lower the gain very simply with a resistive attenuator, which will cause no loss of audio quality. In fact, if you put this attenuator between the preamplifier and power amplifier and you size the resistors properly you can increase the input impedance of the power amplifier to better match the preamplifier.

Lastly, in systems that have too much gain one should also consider using a passive line stage which can be something as simple as a potentiometer in a box. If you do use a potentiometer in a box, one must use very short cables on the output of that box because the output impedance is going to be rather high, generally between 10K and 50K ohms, maybe higher. That is not suitable to drive any kind of cable capacitance. Some people believe that a preamplifier somehow increases the dynamics of a system. I would have to say that the line stage of a preamplifier is virtually identical to the input stage of a power amplifier, so there is no explainable reason why it would increase the dynamics at all.

If you hear increased dynamics with a preamplifier maybe that is caused by the fact there is too much cable loading with the passive preamplifier, which could easily be resolved by a short low capacitance cable. It is not difficult to find cables that are as low as 30 pF per foot and if you have a 1/2m or 1m cable from the passive preamplifier to the power amplifier there should be no cable problems. There is a lot of interest in transformer or autoformer type passive preamplifiers as well. They can be very good, but they generally like to be driven by low impedances and if not driven by low impedances they can have low frequency roll off due to a lack of inductance. In addition to having step-down values for attenuation, transformer type passive preamplifiers can also have step-up values to provide gain, although it is rare to need the step-up value.

One should be aware though that the output impedance of a transformer type passive preamplifier is directly dependent on the input that it is being driven by. The beauty of a transformer type passive preamplifier is that as you go down in volume you get lower impedance, and you can get low output impedance especially when driven by a low input impedance. On the other hand, a high impedance source such as a phono preamplifier may not be comfortable driving a transformer passive preamplifier. We must look out for some of these impedance problems. It is usually only a problem if there is a great difference in impedance, but keep in mind that each impedance mismatch, and I do not like to use words such as match and mismatch because we are not trying to match things, we are trying to have things that do not load the source. Simply you are trying to not load a source because loading a source generally increases distortion. Distortion is something we do not want.