The high end audio world is a strange place. There are “gurus”,
There are always people telling you what you need to buy to achieve audio nirvana. Unfortunately most of them have their financial interests to satisfy, not your enjoyment of music. Do not make your decisions based on the opinions of others, you and you alone should decide what is right for you. There are several myths that have been propagated by the "experts" that have no engineering science behind them. We explore some of them below.
I do not consider myself an "expert". I am an engineer. Most of my associates who contribute to this site are also engineers. The information presented in this web site is based upon sound engineering principles. We are attempting to present affordable high end sound. When designing a cell phone the engineer must justify the cost of every part. If you apply the same principles to amp design, and weigh the cost versus the sound improvement, these things fall off the bottom of the list. If you believe that these things improve the sound AND you have the money to spend, then by all means buy them. For those of us with a limited budget, the money would be best spent on better components.
It would be relatively easy to build an excellent vacuum tube amplifier by copying a known good design and buying the best parts that you can afford. We have done this several times for customer's amplifiers. This is probably the best route to take if you are new to tube amps (or buy a kit from a reputable company). This is the route to take that has the highest probability of success. It is not always the path to lowest cost, or to finding the right sounding amp for your ears.
Where do our designs come from?
I started building audio amplifiers in junior high school. At that time (mid 60s) they were tube amps based on known good designs. I have worked on everything from cellular phones to automotive electronics since then but I have managed to build several tube amplifiers a year for the last 10 years or so. Most of these were variations on a known working design. Some were amps that I intended to keep, but were sold because I was offered enough cash to build a better amp. Some were "could you build me one like his with a wood base, or a shiny chrome chassis" amps. I still have a push pull 300B stereo amp, and an absolutely awe inspiring 845 Single Ended amp, a Single Ended 45 amp, a few "classic" amps (Fisher, Scott, Heathkit, Dynaco etc), a few "classic" tube guitar amps (Fender Bandmaster, Guild Ultraflex, Gibson Skylark). I have not let anyone listen to these amps because I want to keep them. The 300B was based loosely on an article in Vacuum Tube Valley, modified to use tubes I had on hand. The 845 was based on the poor mans Ongaku design, but modified heavily because I wasn't satisfied with the sound from the Ongaku design. I have decided not to build any more amps based on common designs unless a paying customer wants one. Most of my tube time will be spent turning several years of ideas and unfinished circuits into great sounding audio amps and inventing new ones. The 845 SE is a new, unique design that was created in a quest for MY ideal amplifier. As funds (for transformers) become available I plan to do an 833A SE amplifier that should do at least 75 watts per channel.
Our designs begin with an idea or a circuit which can be original or from an outside source. A rough schematic of one or more stages is drawn with first guess component values. Using tube manual or measured data, calculations are performed to verify or adjust the component values. If the design lends itself to simulation, then a simulation is often done. Then we use the Tubelab to prototype the design. Once built on the Tubelab the tweaking begins. Component values, bias points, power supply voltages, and other parameters are adjusted to optimize the sound. Different tube types or different output transformers can be tried, and the tweaking repeated. It should be noted that my optimum sound is likely to be different than your optimum sound. There are many variables to consider such as musical preference, speaker type, room size and layout as well as your ears. This is the main reason why an "expert" can not tell you what you need to hear. You must decide for yourself.
Over the past few years I have collected a warehouse full of vacuum tubes, many for little or no cost. Most of these are from WWII era military surplus, or the remains of a TV repair distributorship. All of the common audio tubes were removed before I got them. I have been trying out the tubes that I have a large quantity of to find tubes that are useful in audio applications. The tubes that have no possible audio use (hydrogen thyratrons, magnetrons, nixies and rectifiers rated for 20000 volts) are being sold. Some of the useful tubes have already found their way into prototype designs. There were about 20 833A's in the military surplus lot, just had to try those out. The results were encouraging enough for me to begin collecting the parts for the biggest SE amps that I have ever built. See the 833 SE page.
Sometimes I'll just sit down at the Tubelab and "play with some parts" or try out some new tubes to see what turns up. If you try this approach, be careful, and be prepared to fry a few parts before you create a masterpiece. Never use your good speakers when experimenting or testing a new amp. If your amp oscillates the speakers could be damaged. We use a home built cabinet with a 6" X 9" car speaker in it for testing. Car audio speakers are relatively cheap, hard to blow out, and give reasonably decent sound.
Amplifier Stage Simulation
We briefly mentioned simulation. At work I am known as the build it guy. I use simulation sparingly in an environment where simulation is the norm. I am the guy who takes someone else's simulation and transforms it into a working product. What have I learned? The biggest thing to remember is that if the simulator tells you that something won't work it usually won't. If the simulator tells you that a circuit will work, it has a chance, success is not guaranteed. A successful simulation means that the circuit will work if all of the components closely match the mathematical models. This is not often the case with 50 year old parts, or new copies of 50 year old parts. You must be aware of this, and not be too trusting of any vacuum tube simulation. I have used simulation on vacuum tube circuits quite often. It can help you determine whether or not to build a design, and often accurately predict the performance.
What kind of simulators do we use. We have all of the Tubecad single stage simulators. I have just received their Push-Pull calculator. I have been experimenting with the totem pole modes. These simulators are limited to a single stage, of certain topology. They work great and are cheap. We have used Beige Bag Software's spice simulator (only the demo version). It seems to function great, and the full version has models for many popular tubes. It is also capable of simulating a circuit containing both tubes and transistors.
Some Common Audio Myths and Our Opinion About Them
1) You should use Amperex Bugle Boy 6DJ8's (or whatever particular tube that the "expert" is selling)
I am sure that these particular tubes sound great in his particular amp. Will they give the best possible sound in your amp? Not necessarily. Every tube has a set of operating conditions (voltage and current) where it performs best. If your amp provides the tube with the same set of conditions that the tube was designed for, it will work well. Every amp, and every tube is different. I have been curve tracing a lot of tubes in the past year, and I have found a large variation in tubes of the same type. It seems that the common types 6SN7, 12AX7, 6CG7 are all over the place, even among several tubes of the same brand and vintage. If you are building a new amp, put in some known good tubes of a common manufacturer, get the amp to work well, then you can swap tubes. If there is a large variation in sound among different tubes, you are probably operating the tube too far away from its ideal operating point.
My own personal amplifiers use whatever tubes that I can find for a good price (no specific brand). I have lately been designing the amplifier to fit the tubes that I already have in my collection. I tend to avoid tubes that have a cult following because the price is often highly inflated for no real reason. I have never seen a Western Electric 300B but I do believe that it is a very good tube. The only 300B's that I have experience with are the Sovtek, which is a very linear tube, but prone to failure, and the new Chinese ones. They also seem very linear and work well over a reasonably wide range of conditions. The only other "magic audio tube" that I have used is the 45. This tube is also extremely linear over a wide range of operating conditions. Certain brands seem to have more of a cult following than others, but I have found that there are only 3 different constructions (ST version) and sound quality is more dependent on how well the tube has survived for 60 years than the brand name. Tube tester results don't correlate with sound quality either. All of my 45's were obtained on Ebay (bid low and expect to lose often) or at hamfests. I have never paid more than $25 for a 45 tube. Chinese 845's are decent tubes and can be found for about $30 each.
2) Testing an amplifier with lab equipment is unnecessary since your ears are the ultimate guide
We believe in testing amplifiers with lab equipment. Can you build a great amplifier without bench testing? Absolutely! I know several people who have built excellent amps with only a voltmeter. Can you corrupt the sound of an amp by testing it? No! Some early solid state designs had a nasty habit of self destructing when tested with a 10 KHz square wave. If an amplifier can be damaged by testing then it was poorly designed. If you build or modify an amplifier and it sounds terrible, how do you fix it without testing?
We understand that an amplifier that performs supremely on the test bench can sound lifeless in listening tests. However an amplifier that shows visible crossover distortion will surely sound terrible. Engineering science has not yet progressed to the point that a lab instrument can measure sound quality. We can measure the distortion, the harmonic spectrum, the damping factor, frequency response, and other specifications, but can not measure sound quality. Does this mean that we should not bench test our amplifiers?
I personally run multiple tests on every amplifier I build. I have found problems in amplifiers using test equipment that would have been elusive with out it. Most problems can be found using an audio generator, digital voltmeter, and an oscilloscope. You should already have a voltmeter. A test CD in your CD player can substitute for the audio generator. A cheap (or used) scope can be had for the price of a set of audiophile tubes. If you intend on building more than one amp it is a good investment. I also use an HP8903 audio analyzer and an old HP spectrum analyzer but these are not absolutely necessary.
I can remember one particular amplifier that just didn't sound right but checked out ok in the usual tests. Even the distortion remained reasonably low up to the clip point. The spectrum analyzer, however revealed that the 3rd and 5th harmonics increased abruptly when the power reached 10 watts. The total distortion did not change enough to be seen on the scope but it was clearly audible, and found only with the spectrum analyzer. It turned out that the phase splitter needed a little more current. There are spectrum analyzers (FFT analyzers) now available that run an a PC using the sound card.
I recently bought one of these FFT analyzers and this thing is too cool. I have learned more about sound in the week that I have used this than I have in years of tweaking. I now know why an amp with 5% distortion can sound better than one with .5%. Tweaking the sound of a new design is now effortless. My methods of amp testing have been revolutionized. The expensive HP8903 audio analyzer, and the spectrum analyzer will no longer be needed. See the recently added amp testing section.
3) Printed Circuit boards destroy the sound of an amplifier
A poorly designed circuit board can cause a reduction in high frequency response. So can a poorly thought out point to point wiring job. We routinely use PC boards in cell phone designs at 2 GHz with out considerable loss. If the board is properly designed it will work fine. I have designed circuits on Tubelab 1 whose frequency response actually improved when implemented on a PC board. The driver board in my 300B push pull amp will happily put out 150 Vp-p at 50 KHz, the hand wired prototype could only make 30 KHz before rolling off. Most of the amplifiers that we build use PC boards. This makes it easy to build identical amps that deliver identical results.
4) You should use silver wire with Teflon insulation. If you use a PC board it should be glass Teflon.
Ok these concepts could actually improve an amp. Whether you could hear or measure the improvement is debatable. Silver wire does have a slightly better conductivity than copper. The difference over the short lengths encountered in an amplifier would be a very small fraction of an ohm. At the impedance levels found in vacuum tube circuitry, it is negligible. Teflon has a few useful properties, the most important being temperature resistance. It has a lower dielectric constant than PVC, so it could exhibit lower stray capacitance. Again, probably negligible if wires are properly spaced. If you have ever worked with glass Teflon PC board material you already know why we don't like it. It is too flexible, hard to work with, and the plating will tear loose at the tube sockets when you change the tube.
I use a good quality stranded tinned copper wire with high temperature PVC insulation. I use Teflon insulated wire when I can find it surplus, a good place to look is in surplus avionics equipment. This stuff is full of useful high quality parts (the old vacuum tube avionics by Collins and Bendix).
If you have the money to spend, silver wire makes good sense in an output transformer since the resistive losses are considerable. I have never tried one due to the cost.
5) You should use only carbon composition resistors
These things are terrible. They are prone to temperature drift, value change, and noise, why would you use them? I don't know but some of the "experts" believe that you need them. I don't put these in anything that I build.
6) Use only Holco resistors, Black Gate electrolytic capacitors, and other premium parts in your amp
Hey if you want to spend the money, go ahead. No doubt about it these are premium parts at a premium price. There are sound engineering reasons why some of these parts are better than their generic counterparts. Should you use them everywhere in your design? Only if cost is no object. We believe that you should build your amp with good quality generic components, then after is is working and you have listened to it for a while you can upgrade selected parts. That way any improvement will be obvious. All of our amps start out with Panasonic capacitors and resistors from Digi-Key. If a customer wants specific parts we will use them. I have noticed small changes in the sound of an amp by using premium capacitors and reduced noise by using premium resistors. Putting premium parts in the places where they will do the most good is the most cost effective upgrade.
The best places for premium parts are:
1) The output transformer is the most important part in a tube amplifier. This part determines the "sound" of the amplifier. Spend most of your amp budget here. There are a few bargains available, and a few high priced brands, but for the most part the quality of an output transformer can be determined by its weight, and often price.
2) Use quality metal film or tantalum resistors in the signal path especially in the first stages of an amp. This usually means any resistor connected to the grid or plate of an amplifier tube.
3) Use a quality electrolytic (low ESR, low inductance) or a film cap or a paper in oil cap for the last capacitor (usually connected from the output transformer to ground) in the power supply. We usually use a good electrolytic on the circuit board, and connect an oil cap (or a Solen fast cap) in parallel mounted external to the board. Recently it has been discovered that motor RUN caps are excellent low ESR capacitors that can be used in the power supply, or the parafeed cap in a parafeed amp. These should not be confused with motor START caps which are usually electrolytic.
4) The capacitor from the cathode to ground in any amplifier is another candidate for an upgrade.
5) Somewhere in most any tube amplifier there is at least one coupling capacitor. It is connected from the plate of one tube to the grid of the next. This is the most important small part in the amp, and a good candidate for an upgrade. The benefit of an upgrade depends upon the surrounding circuit topology. In a perfect design (which doesn't exist) the quality of the capacitor would have a small effect. If you only want to upgrade ONE component in an amplifier, it should be the coupling capacitor connected to the output tube (two in a push - pull amp). This is where you will hear the most improvement.
7) All solid state devices are evil ----- 'sandophobic'-----fear of silicon
There are a large number of vacuum tube enthusiasts who display fear and hatred for all things solid state. Most of this is unfounded. Yes, it is obvious that a modern solid state amplifier sounds quite different than a vacuum tube amplifier. Does this mean that all solid state devices are evil? No, of course not. Actually the sanitized sound of a modern solid state amplifier has more to do with the circuit design than the choice of active devices.
Most solid state amplifiers sold in the past twenty years follow a similar design strategy. They typically have an op-amp like design with large open loop gain and heavy negative feedback. They tend to sound alike. Early solid state designs (when excess gain was expensive) did not share these traits and some of them actually sounded pretty good (if they didn't blow up). A bipolar transistor in common emitter or common base configuration is inherently nonlinear. If large signals are applied, feedback is needed to improve the linearity. J-fets and mosfets are an improvement over bipolar transistors, but feedback is still needed to create a high gain stable amplifier. For those who can remember that network analysis class in college, you will recall that if the open loop gain is high enough, the characteristics of the amplifier are determined by the feedback network, not the amplifier itself. What does this mean? Simply put that given enough excess gain and negative feedback, a vacuum tube amplifier will sound like a modern solid state amp, without any silicon in it! If you have built a single ended MOSFET amplifier without feedback (like the ZEN amps from Nelson Pass) you will hear sound that is remarkably similar to a vacuum tube (pentode) single ended amp, without any tubes. Therefore I believe that we should add a few more choices to the list of available active devices and learn how to build a better amplifier.
The emitter or source follower configuration can be extremely linear if the right transistor or FET and operating current is chosen (look for constant beta in the chosen current range). The emitter follower or source follower has high current gain (no voltage gain). The vacuum tube has good voltage gain and fair current gain.
When paired in "super tube" configuration you can realize the best of both worlds. Would you like a 300B that can deliver an amp of current. It is possible! We are currently experimenting with these combinations and will present a complete amplifier design when it is complete. See the Super tube SE page. The "super tube" comes to life.
While searching for the best solution for driving a power triode, I discovered a very good solution using solid state components. This would have never happened if I did not consider using "sand state" devices. See the PowerDrive page. I have recently found new applications where a solid state device can be used to improve vacuum tube circuitry, or solve a problem. Look for details shortly.
Tube enthusiasts have known for years that solid state diodes sound differently than vacuum tube rectifiers. There are sound engineering reasons for this. A vacuum tube has much softer turn on, no stored charge effects, and a higher internal impedance than a solid state rectifier. I have built amps both ways, and I have built amps with a switch for an instant choice. There is a definite difference in sound. You should listen to both and form your own opinion. I prefer a solid state rectifier for most music in my 300B push pull amp, but a tube rectifier in my 845 SE. Most guitar amps sound best with a tube rectifier, especially if operated into heavy clipping.
8) A2 operation causes audible distortion
This is another one of those myths that is partially true and has been propagated by the "experts" who don't understand the causes of the distortion. There are experts that claim to be able to hear the onset of A2 operation. Must be a crummy amp.
If the driver circuit poses a very low DC resistance AND a low AC impedance at ALL frequencies in the audio band AND the output tube remains linear when the grid goes positive there is no distortion. A cathode follower helps but does not have a low enough DC resistance or AC impedance. The driver transformer used in many high end amps today can present a relatively high driving impedance at the higher audio frequencies, and cause ringing if the driver tube clips.
We have perfected a design that presents the driver tube with a very high DC and AC impedance (near ideal conditions) AND provides a low impedance drive at DC and all frequencies up to about 1MHz (near ideal conditions) to the output tube. Yes solid state components are used. Yes it sounds great. See the PowerDrive page for circuit details. See the Tubelab SE and the 845 SE pages for applications.
I have spent considerable time trying to find the optimum solution to the problem of driving a power triode (or most any other tube to a lesser degree) with a relatively high impedance triode. In the early days this would have been RC coupled. Crowhurst discovered that if a large enough transient came along to forward bias the grid, the capacitor's equilibrium would be upset. The resulting distortion could last for a few seconds after the transient had passed. Cathode followers were used for isolation, but they did not solve all of the problems. Most of today's high end SE amps use a driver transformer. This helps to solve the drive problem, but has several drawbacks. I have developed a circuit that uses an Integrated Circuit constant current source to present the ideal DC load on the driver tube. Then a N-Channel mosfet in source follower mode drives the grid of the power tube.
This works so well that we use it in all current designs. You can now operate the tube in A2 with no ill effects. The results of this work are an 845 amp using a 45 for a driver that makes over 35 watts with a 900 volt power supply. A larger power supply has been designed, the 845 SE now makes over 40 watts, the tube is operated well within its limits, and the amp sounds great. This amp has the lowest distortion readings of any amp that I have built. It is a single ended amp with no feedback.