When you are considering whether you want use tube amplifiers in one of your systems you will probably also need to consider whether you want to use a Phonograph in addition to a CD player. These days if you intend to enjoy any new music a CD player will be a requirement. If you already own a record collection or have found what treasures can still be found on vinyl you will also need a record player. You might also be wondering what type of speakers would be appropriate and offer the best performance. What is your expectation of your Hi Fi and what is the purpose of a Hi Fi? I don't pretend that my opinions about High Fidelity are fact but they are what I have found to be true through years of experience. Others may have drawn different conclusions from their experiences and that's OK.What Hi Fi Is
By definition "High Fidelity" means that you are trying to create the unaltered illusion of a live performance at realistic, but not necessarily loud, sound levels. If you primarily listen to background music or music at levels below 70dB you probably don't need and won't be happy with a high fidelity system as they are intended for realistic levels. Real unamplified musical performances are most often between 80 and 90 decibels with some sounds occasionally approaching 100 decibels. So, my definition of a Hi Fi and it's duty is "A system to reproduce music in a manner that might convince you, at times and if played in a dark room, at an appropriate level, that you are hearing live musicians perform real music." There are many distractions to critical listening when you can see the speakers and components so I listen in the dark. A reasonable substitute might be a dimly lit room in which all the audio equipment is behind a loose curtain. The curtain must be very light weight if it is to remain acoustically transparent and not foul the sound. The entire goal of a Hi Fi, to me, is real live music off of a recorded disc. There are some significant hinderances to this end, of course.Hinderances
The first and most important hinderance is the fact that most recorded music has been subjected to much worse abuse in the recording studio than what takes place in a modest Hi Fidelity system and so it's entirely possible (probably even likely) that the playback equipment will be better than the recording equipment. In this case, your Hi Fi serves only to show you how lousy the recording is. I maintain that once the damage has been done to the signal there is very little that can be done to restore it. Almost all recordings today are "mixed" through a mixing console rather than recorded with just a couple of really good mics. Modern recording consoles give recording engineers a false sense of security with all of the available channels and adjustments and encourage the use of many microphones. It is nearly impossible to maintain phase integrity in a recording that uses more than 3 or 4 microphones. Spatial cues and imaging are lost as a result. Digital recordings often lose some of the low level detail when only a very small percentage of the available resolution is being used. Percussion instruments sometimes do not sound natural in digital recordings. Also, the gigantic mixing consoles are virtual "op-amp farms." The signals on popular recordings today have already passed through likely hundreds of op-amps before they get to you. These problems are out of our control, however, and if we are to continue to enjoy music (at least new music) we will have to accept these things as a given. There has been some improvement in digital recordings with the advent of 24 bit rather than 16 bit mastering. I have found that there are greater differences in the quality of master recordings than what differences exist between analog and digital discs but on the very finest recordings the analog discs still sound better to me. Listen to Henry Mancini records and/or jazz recordings from the 1960s, or the Sheffield Labs Direct to Disc records from the 1970s to see what good recordings can sound like. These are recorded with a tube audio chain. Unfortunately the cost of maintaining an exceptional analog disc playback setup is getting very high.
Another hinderance is that not many of us get to have dedicated listening rooms that can be optimised for the best acoustics. It is hard to get a live performance sound in a 200 square foot living room packed full of furniture because no musicians would ever actually perform under those conditions. Room variations make an enormous difference in sound and every conceivable arrangement should be tried until one is found that you can live with. Realistic bass reproduction requires a room large enough for the long wavelength low frequencies to "unfold." Well, not really. Actually, what happens is that the reflected sound off the back wall returns in phase at some frequencies and out of phase at others, particularly at low frequencies, making some places in the room seem to have plenty of bass or even too much while others seem to have none at all. This means you need to be 15 or 20 feet away from the speakers to get good sounding bass. Often this is impossible to do so people resort to other means of bass reproduction i.e. a "subwoofer". This does not sound as natural as bass reproduced in an appropriately sized room, and it complicates the phase problems that already exist. A real subwoofer can be a good thing when used properly but most of them are not. Most of what are called "subwoofers" today are really "woofers" as a subwoofer by definition is "sub," or below the woofer. A woofer should easily be able to provide linear output to 40 or 50 cps so a real subwoofer must operate below that. If your "subwoofer" doesn't provide lots of output (>110dB) into at least the mid 20 cps range and isn't at least the size of a clothes dryer it is not a subwoofer but a woofer. No speaker smaller than 12 inches is really optimised for subwoofer duty because they cannot move enough air. A real subwoofer will actually knock pictures off of your walls, rattle all the dishes in your kitchen cabinets and even rattle the doors and windows on your house. This is great for reproducing explosions, thunder, locomotives, trucks driving by and cannons being fired, but isn't necessary for music reproduction as there aren't many musical instruments that shake the room when they are being played. One exception might be the pipe organ.
Besides the impedance relationship between the amplifier and speaker, which was mentioned on the previous page, there is another very important relationship that must be considered, that of uniform response in the entire system. This is an area where weaknesses and flaws in each component of the system can either be overcome, at least in part, by a correct match with the other components in the system, or can be made worse by compounding the flaws into a great departure from the ideal. For example, it is common practice to rate frequency response of components within plus or minus 3 decibels. Not that numbers mean anything at all with regard to sound but it at least implies that the standard maximum deviation from ideal when the unit was designed was 6dB, i.e. that no frequency should be more than 6 decibels louder than any other. If it could be guaranteed that the deviation within the entire system was no greater than 6 decibels this would be considered a fairly high standard. However, if each component is allowed to deviate by as much as 6 decibels it's pretty easy to see that in a worst case scenario if one considers only the source (phono pickup, etc) pre-amplifier, power amplifier, and speakers, deviation could easily be more than 20 decibels. It is probably not very likely that each component would deviate in exactly the same area but it should serve to show that matching components in a complementary fashion is highly desireable. By properly matching components it is possible to keep deviations to a minimum and, hopefully, produce an accurate, musical, harmonically correct music enjoyment system.Hi Fi Hierarchy
A principle that I want to be sure you understand is the hierarchy of Hi Fi. No component should ever be better than the component which precedes it (except maybe when evaluating equipment.) If the speakers outclass the rest of the system they will simply show you what a poor job the turntable or amplifier is doing. It is true that the speakers are often the weakest link of a cheap system and some improvement may be had by upgrading them but going beyond a point will make the sound worse. Getting the music off of the record is absolutely necessary if you are ever going to hear it. Be sure to use the best turntable you can afford to buy, even if it means spending less on your speakers. It's tough to try to set a standard but perhaps a reasonable rule of thumb would be to spend about the same on each component of your system, i.e. the same on your turntable, amplifiers, and each speaker. Some would suggest that you spend about 40 to 50 percent of your total budget on your speakers. Of course, if you go the DIY route on amplifiers and speakers the price for these items will probably be less.
Compact Discs have thrown a kink in the Hierarchy rule in that cheap CD players sound a lot better than cheap record players. The hierarchy rule is still very valid when using a turntable as your source because you won't be happy with a cheap turntable no matter how good your speakers are. Harsh, junky amplifiers still sound like junk even when driving very nice speakers.Speakers
Different types of speaker enclosures are better suited to different applications but it depends on what they are going to be used for and how much room you have to spare. The best sounding low frequency speakers I've ever heard for use in your home are acoustic suspension types which have a "Q" of 0.6 to 0.7. The "naturalness" of bass from these systems is unmatched. I don't like the resonant peak provided by speakers which have a Q higher than 0.8. To get a lower Qcb often requires the use of a fairly large box, sometimes twice the size you might expect for a given size speaker. I also find it necessary to make the walls of larger enclosures at least 1 1/2 inches thick using high density materials like medium density fibreboard and brace them internally to prevent any vibration or resonance from being transmitted from the enclosure. Braces from front to back and side to side are helpful. A brace from the rear panel to the magnet structure on the woofer helps prevent the frame from ringing and refrigeration insulation tape (the sticky tar type) wrapped around the legs of the woofer basket will damp it even better. Strips of lead sheets can also be used. The dust cap over the voice coil of a driver can be a source of resonance that needs to be dealt with if the driver does not have a vented pole and it is operated through the extremes of it's upper ranges. Trimming it back near the cone and replacing it with felt will help but if any of the dust cap remains it must have some type of damping material applied so it can't resonate. Removing it may lower the highest frequency at which the driver can be used but it will also clean up peaks and resonance in the upper ranges of the driver. It is acceptable to mount the MF driver and HF dome driver on the baffle near the woofer but upgrading and experimenting are easier if the medium and high range speakers are not mounted in the LF enclosure but in their own baffles. The frequency divider should be outside the enclosure, preferably near or before the amplifiers, and separate wiring should be run to the HF and LF drivers. The divider is a critical component at the heart of every good speaker system and, while there are good passive filters, an active filter will provide the best possible performance by eliminating interaction between drivers and filter components. It is necessary to ensure that each driver gets only the frequencies it was designed to reproduce accurately. The baffles should be covered with heavy felt to prevent reflective diffraction and nothing should protrude from the front of the baffle to interfere with expansion of the wave front. These type systems are extremely accurate and can show you miniscule details in the equipment ahead of them and whether or not the rest of the system is properly designed but the low frequency section is not particularly efficient when a woofer is used that makes the most of this type alignment, generally between 85 and 93dB 1M/1W (about 1% or less.) For this type system my preference is for a woofer that has a Qts between 0.40 and 0.45 with the lowest possible Fs, no higher than 30cps for an eight inch woofer and 20cps or below for a 12 inch woofer. Dynaudio, Peerless, Focal, Vifa, ScanSpeak, Seas, Audax and Tonegen make some fine drivers for this application. I've had good results with even lesser drivers. Dayton Loudspeaker from Parts Express have some very appealing drivers. The first place I would skimp if trying to save money would be in the LF speaker. The medium frequency driver is most critical and you really need to buy the very best you can afford. The enclosure is not all that expensive to build but it is time consuming. Do not try to cut corners. A poorly designed and constructed enclosure will wreck the performance of even the finest drivers but a well built enclosure will make the most of any driver. Do not use an electrolytic capacitor if you use a passive frequency divider. Use a film type designed especially for audio dividing networks. I built a very economical system of this type for some of my family. It uses inexpensive components throughout but has a very rigid cabinet and it outperforms the mass market offerings of similar size. It has a 6 inch LF speaker and a 1 inch HF dome in a 0.6 cu ft enclosure. Here is a photo .
Bass Reflex enclosures offer the possibility of moving f3 about half to one octave lower than acoustic suspension designs but the enclosures are necessarily somewhat larger and the phase shift and group delay are much greater while the power handling below the box frequency is reduced. The enclosures can work well but there is a lot to be considered in designing one of them. A CAD program such as WinISD, LEAP, or Boxplot is very helpful in reducing laborious math calculations.
For PA use even the best sounding speaker is of little use if the volume isn't sufficient to be heard. Outdoor PA equipment may require 1 meter SPLs of 130dB or more. This is where it becomes necessary to trade off a small amount of accuracy for an improvement in efficiency. Once again, the Reflex enclosure. The speakers designed for this type enclosure have higher efficiency than acoustic suspension types because they are relying on the tuning of the enclosure for the bottom octaves (which may only need to go down to 50 or 60 cps) and they generally have a lower Qts, perhaps between 0.25 and 0.40. When tuned properly in an optimum volume box they can perform very well with efficiencies of 94 to 102dB 1M/1W (up to about 10% efficiency.) They sometimes don't equal an acoustic suspension in fidelity because of the phase shift, group delay and rapid cutoff below F3 but these factors aren't as important in PA use as they are in a critical listening High Fidelity set. An improperly tuned box sounds like a high Q acoustic suspension or worse. These types are not very forgiving of design flaws. Once again, they may be nearly twice the size of an acoustic suspension. I designed an enclosure of this type for commercial use. It uses one 15" reinforcement speaker in an 11 cu ft enclosure and one compression horn. The crossover is 12 dB/octave at 1000 cycles. The system is bi-wired with the passive frequency divider located at the amplifier. The response is very uniform from 35 to 16,000 cycles and the efficiency is about 97dB 1W/1M. Factoring in a few dB of correction for baffle step the system efficiency really ends up being about 93 or 94dB. Two of them perform admirably in the 5000 square foot room where they are installed and they are able to provide at least 95dB SPLs anywhere in the room, which is more than adequate. Here is a photo of one of them.
The acoustical horn type speaker combines the best features of both of the above types but are generally enormous for long wavelength low frequencies and are not all that easy to build. They are acoustic rather than electric transformers, matching the impedance of the speaker membrane to that of the air in the room. Because of this some are capable of achieving 50% efficiency. I have attempted one small folded horn but was not pleased with the results. I have the plans to build a folded bass horn like the Speakerlab K or the Klipschorn and I may try it someday. For information on the best implemetation of horn speaker design I've ever seen you might want to visit Rene's Horns. Be sure to download the PDFs to get all the information on the design and construction of these fantastic horns. There is much information available on the Internet regarding bass horn construction so I'm not going to get into that here. I must say that the most impressive bass horns I've ever actually heard are based on the Jensen Imperial. The performance of this huge system cannot be matched by any direct radiator and when SPLs are what you need this system will deliver. It will blur your vision with only 20 or 25 watts of input and can actually cause damage to the structural integrity of buildings. Steve Deckert has probably researched this design as much as anybody and has plans available on his site at www.decware.com . On the downside these have a cabinet volume greater than 37 cubic feet and weigh over 300 pounds each. I do plan to build a pair of these someday.
The bandpass low frequency system is another type that I've experimented with but have not found to have a place in Hi Fi. They may have their place in some types of commercial systems and portables but I do not consider them Hi Fidelity. They shift phase across their operating range and they require yet another dividing network. This is the design of the Bose AcoustiMass systems which I do not recommend to anyone. If you already own one of these, I recommend replacing it with a high quality two band system. These Bose systems are not high fidelity in any way and are designed to appeal to the "First Impression" rather than long term enjoyment of music. BOSE = Buy Other Sound Equipment. Bandpass systems do have gain over a narrow band of frequencies and are favourites with car one-note-bass "booming" SPL enthusiasts.
Once again, the way to go for the highest possible fidelity is to bi-amplify using a dedicated amplifier for each driver in the system and place the frequency dividers before the amplifiers. This is the most revealing design possible as it further eliminates distortion and resonances by doing away with the interactions between drivers and crossover components. It is a somewhat expensive proposition, especially with tube amplifiers, and it may not be worth the time and expense if the source is not first class. With the right system this setup gets you closer to the music than any other way. Anyone that says otherwise does not understand the complex relationship between amplifiers, passive crossovers and speaker drivers.Tube Amplifiers
There is a great fuss about single ended triode amps these days. Yes, they are very interesting, beautiful sounding and expensive but they are not very powerful. They require very sensitive (large) speakers on the order of 94dB or higher. For more practical and less expensive power one may want to consider push/pull tetrode or pentode amplifiers preferably connected in ultra linear configuration. They are quite capable of outstanding music reproduction in spite of what the DHT clan has to say about them.
I have read many opinions about what type of power supply should be used in tube amplifiers. Some people believe tube type rectifiers are sonically superior to solid state types. This is an opinion expressed by those who prefer the sound of a sagging power supply under high demand conditions in an AB amplifier. Adding impedance to a power supply can not make it better for high fidelity applications. I suggest and use solid state rectification even for tube amplifiers. Changing from vacuum rectification to solid state will increase the power output of an amplifier with no other changes provided that the filter capacitors, tubes and associated circuitry can handle the additional voltage but most probably won't. It also gets rid of a heat source inside the amplifier. If you build with solid state rectifiers some means needs to be considered for keeping high voltage off the plates until the heaters are warm. A simple switch will work fine. The beauty of homemade equipment is that you can try whatever you want to and see what works best for you.
Adding capacitance to a power supply will improve it's performance. After a point additional capacitance adopts the law of diminishing returns, however. A thousand microfarads of capacitance in a tube amp is very large indeed in terms of joules of stored energy. I don't find the need for more than this. In fact, a couple of hundred microfarads will perform well in a tube amp. Many old amplifiers had only a few microfarads (less than 10) and worked fine this way but an improvement in sound can be had by adding more.
It is interesting to compare the capacitance in power supplies of tube and transistor amplifiers. The energy stored in a capacitor can be stated in terms of Joules or wattseconds. J=(C*E*E)/2 where J are wattseconds, C are capacitance in farads, and E are volts. If a 50 watt transistor amplifier has two 10,000uF capacitors in it's supply and the voltage on each is about 43 volts the total energy stored is 18.49 wattseconds. A 50 watt amplifier using 6L6s will have a supply with about 450 volts. The same 18.49 wattseconds of energy storage would require about 183uF. In this amplifier 1,000uF would store 101 wattseconds of energy, twice the rated output of the amplifier! I feel that energy storage of 50 to 100 percent of the rated output is perfectly adequate and is more than offered by most manufacturers, even of very expensive equipment. By comparison the 50 watt transistor amplifier mentioned above would need 54,083uF to store 50 wattseconds and 109,248uF to store 101 wattseconds. If these amplifiers are stereophonic both channels draw from the supply making only half as much available for each channel.
I agree that regulated supplies are superior to unregulated types in applications that have a great change in current demands. In class A circuits that have relatively constant average current the improvements are minimal for the increase in complexity so I don't design mine that way. In output stages slight modulation of the supply is not particularly troublesome and inductive input power supplies can reduce any such modulation if you are so inclined. There are certainly worthwhile benefits to using separate supplies for the input and driver stages and the final output stages.
There are certainly many philosophies about inverse feedback circuits and the debates will continue to rage on. In my experience amplifiers which use a limited amount of feedback generally sound the best. Feedback can easily be demonstrated to improve bandwidth, lower output impedance and minimise phase shift. All of these are good, especially with respect to output impedance. Unfortunately feedback alters the natural harmonic spectrum and too much feedback really unbalances the harmonics. Very high levels can also make the amplifier nervous. It is a trade off. One should use only as much feedback as is required to get the bandwidth and output impedance you need. Bandwidth does not need to be infinitely wide and output impedance does not need to be infinitely low for really good sound. I've found that feedback levels of about 10 to 15dB work pretty well on many amplifiers. Some amplifiers may do better with more while others can get by with less. Experimentation is the only way I know of to determine what is best. It is a good idea to install a multi position switch to select a few different measures of feedback to match the feedback level to the application for which the amp is being used. It is very important to design amplifiers to have very low distortion before the application of feedback rather than relying on feedback to reduce harmonic distortion. Feedback cannot reduce odd order distortion at all. Consider, though, that an amplifier without feedback is similar to an engine running without a governor. Can you imagine using a generator or power washer or trying to mow your lawn with equipment whos engine has no governor? Feedback systems are part of nature and even your eye to hand coordination is part of a feedback system. Anything that "learns and corrects" is a feedback system. Triode power amplifiers using 300Bs and $$$$.$$ output transformers may be able to operate acceptably without intentional negative feedback but a tetrode amplifier with affordable transformers will not. You should be aware that there is no such thing as "no negative feedback." Anything in an amplifier that improves linearity is feedback. An unbypassed cathode resistor is one example. Ultra-linear output transformers feed the screens of the output valves with an inverse signal. And, in a plate-follower connected triode, when the input signal increases, the plate current also increases. This causes the plate voltage, which is our output signal, to drop. Since plate current depends on plate voltage, current will be lower than it would have been if the plate voltage remained constant. This, by definition, is degenerative feedback! It can be eliminated by making the plate current independent of the plate voltage and this can be done within the tube by adding another grid between the control grid and plate and connecting it to a constant high voltage. To keep any electrons that bounce off the plate from being absorbed by the second grid, a third grid called a suppressor, connected to a low voltage such as the cathode, has to be added between the second grid and the plate. This is of course called a pentode and by their nature they have lower inherent feedback than a triode. An all pentode amplifier would have less built in feedback than an all triode amplifier. Presently, I don't know anybody that prefers the sound of pentodes without feedback.Surround Sound
The trend today seems to be toward more channels of audio. A properly designed two channel system sounds a lot better for music listening and if people had ever heard a really good one they would laugh "surround sound" off for they silly marketing toys that they are. I don't recommend surround systems to anyone but the mass market has done a good job of convincing people what they want them to buy. If you are interested in convincing musical reproduction steer clear of this stuff and go with two very high quality channels.Making The Most
Some hints that I can offer to improve the sound and performance of less expensive components are as follows...
Your turntable is picking up more sound from your speakers than you realise, even if it's not howling with feedback. Move it into another room away from the sound and see how much better it sounds. You might also place it on a short, lightweight stand, and make sure it is level. Cheap turntables often come with removable headshells rather than removable tonearms. This is a poor design because it can't hold the cartridge still but not much can be done about it. Make sure the pickup cartridge is very tight to the headshell and the headshell is as tight as possible to the arm. Manufacturers often say that the radius on the diamond of their stylus is good for 1000 hours. Maybe, but I figure on about 500 hours for my critical listening system. Invest in an inexpensive microscope that will allow you to examine the radius on your stylus. If you can see any wear, replace it. The "point" of a stylus is actually a very gentle curve. It should not have any flat spots on either side. Also experiment with tracking force. The sound is dependent and you may find better sound when it is set a little heavier than you first thought it should be, especially if your stylus is a conical or spherical rather than elliptical. Upgrading to a low mass elliptical is definitely worthwhile and you may never get the sound you want if you aren't using a superior elliptical stylus because lesser styli cannot follow the short radius of an inner groove at high frequencies without mistracking and making distortion. If your table is a very inexpensive model i.e. less than $300, it is likely that you won't be happy with it's sound no matter what you do to it. It was not made rigid enough to begin with and there's little you can do to improve it. The tonearms on cheap tables are not designed to be able to dial in a pickup to the proper settings. You must be able to adjust the Stylus Rake Angle, overhang and bias settings for best possible sound. If you do buy a better turntable keep the old one as it will work fine for those occasional times when you need to take a record player to another location for a special event, party, etc. On a PA system it will probably sound just fine.
Speakers manufactured for the mass market are usually bad. They are designed to sell cheap so many corners are cut. The drivers are often very inexpensive models to save cost so better drivers can be an improvement. If the box is flimsy there's not much you can do but build another better one. If the box is good you can brace it some more internally, upgrade the dividing network (there should be an entire page devoted just to doing this!), and experiment with damping and tuning. When using them, get them off the floor. The HF and MF speakers need to be located at ear level when seated. The box is usually smaller than optimum for cheap speakers which results in a high Qcb. This means bass that booms. Sometimes you can stuff the box with damping material and help this problem. If the LF speaker has a high Qts this will help that too however if it is too high there is not much that can be done. There is an alignment called an aperiodic which uses an untuned pressure relief vent on the back of the enclosure that helps in these situations also. All of this kind of points back to just building your own in the first place. For information on a small speaker which I have used that can be a good starting point for a second system, travel system, bedroom system or other applications click here.
I welcome any comments you have about the opinions and information expressed on this page. e-mail. Remove the two X's from the address.