|
 | | | | May31Written by:Rip Rowan
Sunday, May 31, 1998 6:00 PM  Which is better, tube or semiconductor circuits? And if differences exist, what are these differences?
The tubes vs. semiconductors issue is both emotional and rational. Many leaders in audio technology continue to hold up the vacuum tube as the state-of-the-art in audio amplification. Others find greater value in modern solid-state devices. Many amateurs and professional alike are left wondering what the reality is. Manufacturers have been likewise polarized: a number of market-driven manufacturers have found that their equipment sells better when it's designed around a tube - regardless of the sound - because of the current "tube hype." Reputable manufacturers of tube equipment - with more esoteric and expensive designs - find it difficult to sell into this market that assumes that "anything with a tube in it is better".
This reminds me of the "Mac vs. PC" debate - unending, unresolvable, polarizing - often vitriolic. Therefore, the aim of this article is not to decide anything, but you let you - the reader - decide for yourself what the issues mean to you.
I have enlisted the help of a couple of people to offer their opinions on this issue.
...EveAnna Manley
EveAnna Manley is the President of Manley Labs in Chino, CA. which produces some of the finest quality vacuum tube equipment made in the world. Manley products are standard items in the finest recording and mastering facilities worldwide.
...José-Maria Catena
José-Maria Catena is President of Sistemas Electrónicos, an engineering firm specializing in both robotics and DSP. He has significant experience in DSP R&D, audio equipment manufacturing, and live sound reinforcement.
I have asked each for their opinions on certain issues, and I will try to summarize as we go along.
Establishing the Criteria
The first thing we need to fix is a correct criteria to evaluate the technologies. We must keep in mind all possible factors, both subjective and objective to find a really comprehensive explanation.
Giving subjective priority to "what sounds better" without analyzing objective measurements is a source of error. Possibly, the sound we prefer has been "colored" to suit our individual tastes. "Better" needs to be properly understood. If we find value in a particular "effect" a device has on a sound, then we should understand what the effect is before we simply say the sound is "better." On the other hand, if a large preponderance of people seem to like that color or effect, then maybe our subjectivity is telling us something, and we should review our objective measurements as well.
If we ignore the subjective tests instead, we can easily lose sight of the goal: good sound. Far too often, laypeople and engineers alike choose a product based on its "specs" without EVER considering the subjective quality of the sound. As for myself, I won't buy it if it doesn't sound good TO ME. Moreover, as as been pointed out by many an audio expert, the objective measurements are simply an abstraction designed to quantify the attributes we think characterize "good sound". Paul Ierymenko, Director of R&D for QSC Audio, sums it up nicely, "the specifications we manufacturers provide for our amplifiers don't provide enough information to guide users in choosing product." (Mix, Nov. 1997).
The purpose of recording equipment is different from the purpose of audiophile hi-fi equipment. The audiophile's goal is to exactly reproduce the recording as it was intended by the musicians, engineers, and producers to be heard. The recording enthusiast often has a very different goal. It is (virtually) impossible to capture sound exactly as it is heard. Anyone who has tried to record piano knows what I'm talking about. First of all, stereo is a huge limitation. In the simplest form, live sound is composed of two sonic arrivals at the listener: the arrival of the instrument's sound at the left ear, and the arrival at the right ear. Stereo offers four sonic arrivals: the arrival of sound from each speaker at each ear. The use of multichannel technologies such as 5.1 simply complicate the issue for the engineer.
Therefore the recording engineer has the goal of artistically approximating the sounds desired by the musicians and producers. If you're recording a symphony, chances are that you're trying to capture the live sound unchanged, to the degree possible. However you're making compromises in imaging to overcome bass response issues. You're moving microphones to understate that overenthusiastic trumpet player. And you may be trying to accentuate or limit the amount of room reflections present in the live room. The bottom line is that you are not simply trying to capture exactly what you hear. If you record modern pop/rock material, then this statement goes double or triple.
Finally, we must understand specific tube or semiconductor properties. Obviously, both a tube and a solid state amplifier can be designed better or worse, and we should generally disregard design dependent parameters not linked to the type of components used.
Size
J-MC: Tubes are much larger. That means basically that we can design much more complex circuits using semiconductors. Many of the circuits that can be done with semiconductors can't be done with tubes, simply because the final product size would be an absolute constraint. There are today many design improvements that can't be implemented using tubes because of the size constraint.
EM: OK. Like the Godzilla ad, "Size Does matter!" I am very grateful for the semiconductor. I use them everyday. In my calculator, computer, coffee-machine timer... but I don't have to listen to them. Funny how those big radio stations don't have a problem dedicating an entire room to a six-foot tall 100 kilowatt transmitting tube. But back to pro gear, don't worry, there is plenty we can pull off with vacuum tube technology within 1 or 2 rack unit spaces. Complex is not necessarily better sounding. And the size thing is not only the realm of vacuum tubes. There are plenty of solid-state power amplifiers out there which take two men to lift.
RR: I have seen the ENIAC (first electronic computer) and I believe in the power of semiconductors to make computers smaller! Given the audio "straight line with gain" design principle that simpler is better, however, I can't see that size is a constraint. My Yamaha power amplifier is all transistor, takes up 4 rack spaces, and weighs in at a hefty 60 or so lbs. Most tube preamplifiers and power amplifiers are not significantly larger than their solid-state counterparts.
Wearing and Reliability
J-MC: This is definitively an important lack of tube circuits. Tubes are very fragile and their characteristics drift over time. Solid state circuits are more rugged and are less variable over time.
EM: Really? That's news to me! I don't know about you, but my grandfather still has his tube console radio from the 1940's working, and all those guitarists out there, they sure as hell aren't using vintage solid state designs from the early ‘60's. Those "fragile" Marshall, Fender, and Ampeg tube guitar amplifiers take plenty of abuse on the road. Tubes are found in Russian MIG fighters... why? They'll survive a nuclear induced electro-magnetic pulse, provided the vacuum remains intact.
The thing about tubes is that if one does fail or go noisy, you can simply plug another one in! The fact that a tube has a socket is not to imply that it is inherently unreliable, but rather it is designed as a convenience. Try replacing (read: break out your soldering iron and prepare for a few hours' work) all those matched pairs of power transistors in some big solid-state power amplifier which just blew up ‘cause you accidently touched the two speaker wires together for just a split second. Oh yeah, and replace your welded speaker drivers while you're at it...
RR: Anecdotally speaking, the most reliable piece of equipment I own is my 1950's model Bogen integrated amplifier. It has all the original tubes intact, and sounds absolutely lovely, although it probably would benefit from retubing. I also have owned a dozen or so Apple II computers which have routine semiconductor failures. And everybody knows someone who has a car whose ignition module failed.
However another way of looking at reliability is at the device level. Semiconductors (especially in computers) typically have hundreds to millions of devices on them. Analyzing the failure rate at the device level gives semiconductors a lower fail rate than tubes. But audio amplifiers are not computers. A power transistor is a single device.
Price
J-MC: Solid state circuits are typically much less expensive to design than tube circuits, because semiconductors are built using a much higher-volume process than tubes.
EM: Sad but true. And Ferraris cost more than Fords.
RR: Even sadder is the proliferation of equipment from would-be tube gear manufacturers that simply throw a tube into a low-cost design in order to get a marketing advantage.
Current vs. Voltage Gains
J-MC: A tube works with voltage gains. There are semiconductors that also work with voltage gains: the FET (field effect) transistor and related technologies, and ones that do with current gains: the bipolar transistor.
This is not really a performance issue, as currents and voltages are accurately proportional though a fixed resistor.
Low impedance, high current circuits, more often found in semiconductor systems, could more easily cause inter-modulation (IM) distortion, but adequate design makes it a non-issue. Although more care is necessary using transistors than tubes, this is a design issue and not a constraint of transistors.
EM: Voltage is a good thing. High voltage used in most tube designs gives you headroom and large peak-to-peak voltage swing.
Component Functionality Flexibility
J-MC: Tubes can only be used as voltage amplifiers: the grid/s are always the input/s, very high impedance. The output is always the same, medium to high impedance, high voltage gain.
With transistors, we can use all three possible configurations:
- Common emitter: medium to high impedance input (very high with FETs), medium impedance output, high voltage gain.
- Common collector (or emitter follower): high impedance input, very low impedance output, high current gain, unity voltage gain.
- Common base: low impedance input, high impedance output. Very rare (if not absolutely unused) in audio circuits.
EM: The most common application of vacuum tube design is to use the grid as the input, but there exists many other configurations as well such as cathode input or simultaneous cathode-grid inputs. Output topologies using tubes can be designed to give extremely low output impedance such as the totem-pole (White Follower) output stage. And are we talking about what we can do with transformers yet?
Impedance flexibility
J-MC: Tubes are always high impedance input, medium to high impedance output. This means that we need to use high power output transformers for power amplifiers. A power transformer is really a bad thing in any audio path. Even if we make silver wire transformers with state of the art materials for the nucleus, we get an important output resistance compared with the 4 or 8 ohms of the speakers, resulting in a poor voltage source. Unavoidable non-linearity of the speaker's impedance along the frequency bandwidth translates in non-linearity in the frequency and phase responses. Maybe if we discover ambient temperature superconductors we could solve some of the disadvantages of power transformers, but it's fiction for now.
Transistor circuits can achieve ultra-low impedance output, even virtually zero impedance. That is, they can deliver the ideal voltage source (zero source resistance). That is what speakers require. No transformers needed.
About input impedance: it's always very high with tubes. With semiconductors, we can get from low to extremely high input impedance (even greater that 10E12 ohms). This flexibility is a big advantage to achieve lowest noise floors, as we'll discuss in the noise point.
EM: Definitely a power transformer is a bad thing in any audio path. Most folks would choose to put a well designed AUDIO OUTPUT transformer at the end of a tube power amplifier to couple the high output impedance of the tubes to the speaker. A well-designed tube power amplifier can easily have an output impedance of 1/4 ohm. And what happens to the output of most transistor amplifiers into lower impedances? We've all seen specs like 100W into 8 ohms, 200W into 4 ohms, 400W into 2 ohms. The transistor amplifier will want to dump more current into the lower loads and we all know speaker impedances are not constant. This presents a very interesting thing to think about.
RR: Also, the impedance issue is only an issue when you're talking about power amplifiers driving speakers. Tube preamplifiers and compressors do not require transformers because they are driving a high-impedance load.
Noise
J-MC: Any amplifier adds noise. The noise at the output is proportional to the gain, and so, the meaningful measurement is what's called "Equivalent Input Noise". It's the noise we would be adding at the input to get the measured output noise.
The equivalent input noise has two components: the voltage noise and the current noise. Total noise is a function of both terms and the source impedance. For low impedance sources (below 2 Kohm), only the voltage noise matters (the term of the equation for the current noise becomes negligible compared with the voltage term). For high impedance sources (more than 30K), what's important is the current noise, and the voltage noise can be ignored.
For a transistor, we can choose the balance between input noise voltage and current, adjusting the collector current. This way, we can get ultra low voltage noises below 1 nV/sqr(Hz) (that's less than the thermal noise produced by an ideal 200 ohm resistor, to give an idea of how low it is), or ultra low currents below 10 pA/sqr(Hz) (also extremely low).
Well, with tubes, the minimum equivalent input noise voltage is many orders of magnitude larger than with semiconductors. The current noise isn't so high, but much larger than with semiconductors, anyway.
This mean that for low level signal inputs, especially with low impedance sources, tubes have a much lower Signal / Noise ratio.
EM: Again, watch the over-generalisations. But believe me, it is possible to design tube circuits which are quiet; we do it every day. Tape Head Preamplifiers, Moving Coil Preamplifiers, Microphone Amplifiers... careful and attentive design is required for best noise performance in these more difficult high-gain circuits if you're using tubes or transistors.
RR: Jose makes a good point. Generally speaking, tubes do offer more low-level hiss than transistors. The noise floor is the only kind of distortion that amateurs understand, so many people tend to assume that the gear with the lowest noise floor is better. However tube circuits tend to have much higher headroom (+30dBu as opposed to +12 for high-quality solid state gear). So if the tube gear has an 80dB noise floor and the solid state gear has an 87 dB noise floor, but the tube gear has 15 dB greater headroom, then it stands to reason that the tube gear still has more usable dynamics. In my opinion, most recording engineers don't realize how much clipping they listen to every day. A drum that reads -10 on a VU meter has peaks that are +28 dB! Only very rare and overbuilt solid-state amplifiers have that kind of headroom.
We will try to bring you more of this discussion in future issue of ProRec, so keep reading, keep rocking, and keep recording....
Tags: | | | | | | | |
|
|