Some notes about tube noise

A long Introduction

Tubes can have mechanical noise and electrical noise. When talking about failures, mechanical noise could be microphonics, and electrical noise could be a sputtering noise.

The normal "white noise" just belongs to anything which amplifies signal, like transistors and tubes, and is not called a failure.

It is hard to determine if a particular noise is a failure, or just a normal product property. Most of the time however, noise problems result from the amplifier design. However this is difficult to understand at the design level of the amplifier, and we see the same failures made over and over.

As a tube manufacturer, the main surprise with this item is, how little interest there is about understanding, about preventing, and about curing such problems. At EML, as tube builders, we say here what some of the general design errors with some amplifier are, causing tube noise, but we whish this wouldn't be necessary.

Often, designers choose inappropriate tubes for the application, (e.g., 12AT7's and 12AU7's for input tubes, or use a Thoriated Tungsten tube below 600V. Some use an output tube as input tube, or use an input tube at high signal swing. As long as these things initially seem to work, and look nice, amplifiers will be made like this. But not all things that appeal interesting, are better sounding, as if the whole rest doesn't matter. Because it DOES matter, and you may end up with issues like noise, distortion, and reliability problems otherwise. Specially when such problems occur not with all tubes, and occur only after some use period, many know of no other explanation than "bad tubes". Being helped with this by the amplifier designers, who say which tubes have problems with their amplifiers, and which not. They confuse this with quality.

The following text can not cure a problem in an existing amplifier, but hopefully some designers will READ it.

Signal to noise ratio.

Some examples
  1. Lets take an amplifier with 6SN7 in one of the stages. You sure may have noticed some amplifiers produce hardly any audible microphonics when tapping on the 6SN7, and some other produce a loud "bang....bang..." from the loudspeakers if you do so. Now we really must ask ourself the question: WHY is this one amplifier working so well, and why is the other so overly sensitive. I mean, with the same 6SN7. The answer can be many. Like too high source impedance driving the 6SN7, but most of the time it's just a matter of too small signal, passed along a too large tube. So the larger a tube is, the larger the basic noise will be (without signal). This is unavoidable, and just normal. Then, when passing small signal only, the signal to noise ratio will become bad. The only solution, given you do want to use 6SN7 at that position, is to pass larger signal through the 6SN7.
  2. Also seen often, is an output tube, for instance like 300B, used to drive a head phone. This is a very typical example, where many (!) designers are passing too small signal along a big size tube. The 300B is able to produce up to 17 Watt output power, and will do so at a very good signal to noise ratio. (Signal to Noise Ratio is also called SNR or S/N). When driving a head phone however, we need hardly any signal at all. For a 300B it is almost nothing. Head phones need so little power, you can already overdrive them with a 6SN7. The general design error here is to pass too small signal along the 300B. For good result, so a possible high SNR, we MUST HAVE high signal from the 300B output. This is extremely important, but not looked at. Besides, it is so tempting to ignore it, because it simplifies the design considerably. Most of the time however, designers are not even aware of this, and are only happy with the simplicity of the design. Yet the output transformer needs often a much higher step down ratio. This automatically requires a much higher signal swing at the primary. When doing so, every 300B which was 'noisy' before, suddenly becomes dead quiet. That's the way to do it! However, this requires an additional gain stage, designers simply avoid this, and then "tube problems" will come their way. For the additional gain stage applies the same. It should be made with LOW SIGNAL tubes, and not with driver tubes, and sure not with another output tube. An alternative way to add more gain is using a Lundahl step up transformer. Just suppose you boost the input signal a factor 3, with an LL7901 transformer. Like this, you get this amplification free of noise, free of distortion and no microphonics. Now, you will get 3x more signal at the 300B output. By using a 3x larger step down ration of the output transformer, signal to the head phone will be the same. BUT.... at 3x less noise as before, including microphonics which is now 3x lower. That is because the noise is now attenuated with a factor 3 as well. You can do even a better job, by going for a factor 20, but that requires more than an input transformer. It requires an extra gain stage of 20x, and even the noisiest 300B will come down to only 5% of the noise level as it had before.