Tubes Do Something Special Letters
Editor: Congratulations on your June 2001 issue, which contains several interesting articles. With the death of Audio magazine, there are few places for the consumer to see this kind of technical information. I have been trying to understand the apparent disparity between measurements and critical reports for some time and the article by Ben Duncan is a good example of a real attempt to throw light on this.
I am a little more perplexed by Peter van Willeswaard's follow-up on "Tubes Do Something Special." A full analysis is not possible as he does not state the impedance of the speakers he used, but if the rating of the transistor amplifier is for an 8 ohm load, and the speakers are 8 ohms then at the rated output we would expect 40 volts peak-peak. (V2 = PxR gives the RMS voltage, which is 14.14V. Multiplying by 2.828 to get the peak-peak voltage gives 40V). It seems reasonable that the amplifier would overload at 42V p-p, as we know that transistor amplifiers tend to overload abruptly.
Now if the 300B SE amp is compressing, it is likely to have considerable 3rd and higher odd-harmonic components in its output and therefore have more of a squarewave output than pure sinewave, which would suggest that the RMS power output is greater than that calculated above for a sinewave. However, assuming a sinewave, at 46V p-p, the calculated sinewave power is 33W into an 8 ohm load, compared to the rating of 9W. Of course, if the speakers are actually less than 8 ohms, the power is correspondingly higher.
The same arguments apply for the 3W push-pull EL84 amplifier. Sounds as though someone is being very modest about power outputs. I do know that SE triode amps tend to be rated at relatively high harmonic distortion levels. I wonder if Peter can borrow a signal generator and spectrum analyser from Agilent and see just what is there at 46V peak-peak!—Andrew Tomlinson, ATomlinson@aol.com
Yes, that would be interesting, but it is very difficult to do a meaningful spectrum plot of one signal peak. Spectrum analyzers do not like moving objects! And, as I reported earlier, nothing special happens when driving with sinewaves. FYI, the speakers are 8 ohms and the transistor amp spec is also for 8 ohms.—Peter van Willenswaard
Editor: Over the years we have all been exposed to the basic block diagram of amplifiers: a power supply of appropriate 'stiffness' (invulnerability to any demand within the power envelope of the output devices) which is controlled by the output device(s).
Transfer factor is the ability of a device to reproduce the dynamic range of the input signal in a linear way: not less and not more. Gain is the degree to which the input signal is amplified. In the end the discussion of the various amplifiers and the scope pictures begs all these questions:
Was the gain/volume control on the device providing input to the amp and speakers adjusted for equal output for some arbitrary frequency before the experiment (in other words is the outcome merely differential gain)?
Did the various amps correctly reproduce the input dynamic range (in other words is the SET amp acting as a dynamic range enhancer or are the others either running out of power or understating the dynamic range of the incoming signal)?
Is the outcome really a measure of the stiffness or speed of the power supplies (recall the recent discussions about tube rectifiers and hexfreds)? Lets not forget dynamic headroom. Where is the scope photo of the cd player or preamp output?
Is all this discussion another way of saying that tubes have euphonious distortion? Louder is not always better. More dynamic is not always more accurate.
If the author doesn't want to do the "thesis" level work that is required then let him be silent until someone does. I keep thinking of the scene in Spinal Tap> where some musician is talking about the amp with a volume scale up to 11.
By way of full disclosure I admit that I have never heard an SET that I liked. When I heard the Havilland monoblocks I was struck by the dynamic range but that does not mean I thought my impression answered any of the questions above. I thought they sounded as though they could not control the speaker. That is not an answer to the above questions.—Pieter Williams, email@example.com
As I explained in the articles, I pushed each amplifier on an individual basis to see just how far it would go, what voltage/loudness it was able to produce; no more, no less. And I don't see why I shouldn't be allowed to report facts that haven't been explained yet on a "thesis" level. Collecting observations has always been a sound basis for scientific progress.—Peter van Willenswaard
Editor: In reaction on the "tubes sound louder" article by Peter van Willenswaard, I would like to suggest the following: The output transformer sounds louder. The output voltages measured by Mr. van Willenswaard suggests interaction between the transformer and the loudspeaker.
I'm really questioning the assumption that the difference lies in the tubes. Are the measured peak voltages of tube and transistor amplifiers equally different if one would use a (1:1 audio-) transformer after the transistor amplifier? (Maybe also an additional series resistance to account for the output impedance of the tubes?)—Paul van der Hulst, firstname.lastname@example.org
Quite to the point. I was already planning to investigate this, given time. I'll report my findings.—Peter van Willenswaard
Editor: I think I might have the answer as to why Peter Van Willensward got the measurements he did. One difference between a tube and solid-state amplifier that was not discussed is that tube amps have audio output transformers and solid-state amps do not. Transformers do have some characteristics that can influence voltage and current outputs to the load.
Now a really fair test would be to have one of the amplifier manufacturers you're buddy-buddy with to make a solid-state amp that could drive the identical transformer used in the tube amp and then make a comparison.
Have fun with this project and good luck.—Ray Moran, Austin TX, Raytx@aol.com
I have thought of that, yes. However, making a transistor amp that drives a typical tube output transformer from the same high voltage as used in tube amps is no simple feat, I can assure you. Especially not if at the same time you would want the transistors to drive that transformer in the way a tube would do. If someone had solved that, and made the amp as robust against abuse as a tube amp normally is, he could probably sit back for the rest of his life! Meanwhile, the closest thing would be to test a McIntosh: transistor, only low voltage, but with an output transformer. I am planning to look into that.—Peter van Willenswaard