Interesting and surprising experiments with valves
by B. Kainka
Source: Elektor Electronics 9/2003
Kainka's site: http://www.b-kainka.de/
Is it just nostalgia, or are valves really somehow better than transistors? Recently valves have been making something of a comeback in many areas. Using valves seems to involve a lot of effort, and in particular the high voltages frighten many people off. But there are dozens of valves lying about in many a cellar — so why not try something new with those old valves? Valves are usually driven using an anode voltage of 250 V or more; practically never with an anode voltage below 100 V. For power ampliﬁers, in particular for radio transmitters, several kilovolts can be used. Such inconveniently high voltages naturally put many people off, as do the special transformers and high-voltage electrolytic capacitors that are needed. But things need not be like this. A series of experiments has shown that most of those valves nostalgically kept at the back of the cupboard will work at very low voltages. Of course, we are not talking about achieving the ultimate in power or ampliﬁcation, but for simple applications — and a bit of fun — it ﬁlls the bill. We will describe in this article how to build simple circuits using valves with a minimum of fuss. Operating at anode voltages of, for example, 12 V is not recommended by manufacturers and is not covered in any data sheet. So, if we are going to learn anything, we will need to experiment and make some measurements. In order to forestall criticism from committed valve-lovers, we should say that the aim here is not to build the last word in hi-ﬁ ampliﬁers or find the optimal operating point for some particular valve. We are more interested in gaining some experience with valves in a simple and safe way. There is a special satisfaction in building a small circuit and making a simple working device. And it is not just about feeling the warm glow from the cathode: it is like going back in time to the early days of electronics, when the (relatively simple) technology was dominated by amateurs and everything was visible. Valves in their glass envelopes are certainly more ‘transparent’ than ICs in plastic packages. Of course, we could do things ‘properly’, and use anode voltages of 250 V; but that would not exactly make for simple and relaxing experimentation on the bench. A chassis would be required and everything would have to be built carefully into a case. And we would always have to watch out for those dangerous voltages. None of this is a problem if we stick to low voltages.
Types of valve
The question frequently arises: ‘are valves still being made?’ The answer is yes. Out of the enormous range of valves that used to be available, a few are still around, being made by several manufacturers. As well as valves still used in radio transmitters there are a few types specially designed for hi-fi ampliﬁers. Output stage valves such as the EL84 (6BQ5) and EL34 (6CA7), and ECC81 (12AT7), ECC82 (12AU7) and ECC83 (12AX7) double triodes are readily obtainable today, albeit at rather higher prices than a few decades ago. Other sources of unused valves are mail-order suppliers such as Chelmer Valve Co., who offer a particularly good stock of European, US and Russian valves at reasonable prices. Especially interesting are the numerous types of miniature valves, so-called ‘battery valves’, which are designed to run on low voltages. However, we are not limited to using new valves. Devices salvaged from the cellar will generally be old radio or television valves. In the good old days radios and televisions that were beyond repair were often cannibalised for their valves. In televisions we generally find P-series valves, designed to have their heaters wired in series, with a heater current of 300 mA, as well as the ECC81 and ECC82 types mentioned above, and any number of EF80s (6BX6). From radios we can obtain interesting devices such as EL84s and EL95s (6DL5) in output amplifiers. The E-series devices require a heater voltage of 6.3 V. All these devices can be brought back to life, and can even be used at low anode voltages. Indeed, there are so many different types of valve that we do not have space in this article to cover the technical details and socket pinouts for all of them. All the necessary information can, however, be readily found on the Internet. There are also sites that give example circuits and hobby projects alongside such data. Even the use of low anode voltages is mentioned here and there, and it is becoming something of a hobby in itself. So, for example, the hobby corner of the author’s website (www.b-kainka.de) includes example circuits from a 12 V headphone ampliﬁer using two transmitter valves to an Audion-type receiver. There are also many links to similar projects.
The ECC81 (12AT7)
The ECC81 is readily available both new and second-hand. From the double triode series ECC81/ECC82/ECC83, which all have the same pinout, the ECC82 is also suitable, but the ECC83 has too low an anode current at low voltages. The ECC81 was originally intended for HF applications and sweep circuits in televisions and oscilloscopes. They are therefore capable of operation at high frequencies, and best performance is obtained with anode currents of between 5 mA and 10 mA. For HF applications the valve is often found in a cascode configuration, where the two triodes are in series and therefore share the available anode voltage. This is why the ECC81 has adequate anode current and transconductance at low voltages. Once one has obtained a used valve, the first question is naturally whether it works or not. This does not require a complete circuit to be built: a couple of simple experiments can be carried out on the bench with the aid of crocodile clips. First a heater voltage must be applied. Figure 1 shows the socket pinout of the ECC81. Almost all valves with a nine-pin ‘noval’ base have heater connections on pins 4 and 5. The ECC81, ECC82, and ECC83 are a bit special, however: the heater element has a centre tap on pin 9. This means that the valve can be used with a heater voltage of 12.6 V (with a current of 150 mA) or with a heater voltage of 6.3 V (at 300 mA). This is very convenient for our purposes, since we can use 12.6 V (12.0 V will also do!) for both the heater voltage and the anode voltage. First connect the heater voltage of 12 V to pins 4 and 5. After about half a minute the cathode will start to glow. If no current ﬂows, the valve is probably burnt out. This case is relatively rare. More frequently, the valve is not burnt out but rather badly aged and will have rather poor characteristics. For simple experimental purposes, however, it will probably be perfectly usable. The second thing to test is whether the vacuum in the valve is still hard. Connect a voltmeter between cathode and grid (see Figure 2). If all is well, a voltage of approximately –0.5 V should appear on the grid (assuming the voltmeter has an input resistance of 1 MΩ). This is already showing the effect of free electrons. The hot cathode ejects electrons into the free space around it, and some land on the grid, giving it a negative charge. If, instead of measuring the open-circuit voltage, the short-circuit current is measured, a value of around 20 µA will be found. This effect is used in many circuits to automatically create a negative grid voltage, including in the headphone amplifier described below. Whether the vacuum is still hard can often be determined by inspection. The ECC81 has a silver-coloured speck at its end, called the ‘getter flash’. When the valve is manufactured the air is pumped out of it through a glass tube and then it is sealed. In the end of the valve there is a ring-shaped groove which is filled with a metal having a low melting point. This metal is heated through the glass using a powerful HF magnetic field: this evaporates the metal onto the inner surface of the glass below. The result of this whole process is to permanently trap the last remaining gas molecules in the envelope. If, after years or even decades, the getter ﬂash is still bright, then all is well; if it is white or grey, it means that air has leaked in and the getter metal has oxidised.
|Figure 1 - Pinout of the ECC81|
|Figure 2 - Measuring the negative charge on the grid|
If the cathode glows, a grid current can be measured, and the getter metal is still bright, the valve is in good order and you can use it to build this simple headphone amplifier. The circuit is presented here in two variations. The first circuit, in Figure 3, requires just two components per channel in addition to the valve and headphones. The anode current of 0.17 mA measured experimentally in a used valve is relatively low. In specially-designed low-voltage valves the current was somewhat higher: for comparison, the low-voltage ECC86 (6GM8 or CV5394)) has an anode current of 1 mA in this circuit. Note that if by chance you do have an ECC86 available, it cannot be ﬁtted directly in the same socket: it has no centre tap on the heater, and so requires a voltage of 6.3 V between pins 4 and 5. The first simple amplifier circuit works perfectly well with high-impedance (600 Ω or 2000 Ω) headphones. However, it is not good practice to maintain a DC current through headphones, although there is no danger of overloading the system, since the anode current is very low. The sound quality can suffer, however, at slightly increased currents and, furthermore, high-impedance headphones are relatively rare. If we want to use ordinary headphones from a personal stereo, having an impedance of 32 Ω, the output level will be very low. The problem is that the impedances are severely mismatched: the impedance of the valve output is of the order of kiloohms. An impedance converter will do the job: for example, we can use a small 230 V/24 V 1.8 VA mains transformer. This has a voltage ratio of about 10:1. The transformer we used in our experimental circuit (Figure 4) had a primary with a DC resistance of 2.5 kΩ and a secondary with a DC resistance of 100 Ω. Using a larger transformer has the advantages of lower loss and hence higher volume. With a voltage ratio of 10:1 the effective headphone impedance is increased by a factor of 100. If the headphones have an impedance of 32 Ω, the valve sees an impedance of 3.2 kΩ, which will give considerably better results. The theoretically optimal operating resistance for the valve is in the region of Ua/Ia, which, at an anode current of only 0.17 mA is around 70 kΩ. The exact impedance is not critical in this application, and so even headphones with an impedance of 600 Ω can equally well be used in conjunction with the same transformer. The impedance seen by the valve would then be about 60 kΩ, practically ideal for the given anode current.
|Figure 3 - A simple headphone amplifier|
|Figure 4 -Using an output matching transformer|
|Figure 5 - Input and output signal voltages displayed on an oscilloscope|
The oscilloscope traces in Figure 5 show that the valve does indeed work with a low anode voltage. The voltage gain of about 8 was obtained at high impedance using an impedance converter and 600 Ω headphones. In general, the following relation holds for voltage gain:
V = S × Ra
where V is the voltage gain, S the transconductance and Ra the output resistance. Given that the output resistance is 60 kΩ, the transconductance of the valve must be 0.13 mA/V. This broadly fits in with the general observation that the transconductance of a valve at any operating point is approximately equal to the anode current divided by 1 V. The official data sheet for the ECC81 gives, for example, an anode current Ia of 3.0 mA at Ua = 100 V and Ug = –1 V, with a transconductance of 3.75 mA/V. This comparison also shows a disproportionate fall in current and transconductance when operated at an anode voltage of only 12 V. This means that for serious applications the anode voltage should be as high as possible. An acceptable compromise between safety and gain might be around 24 V. In order to learn to understand the properties of the valve at low anode voltages, we need to study its characteristic curves. Manufacturers’ data sheets are of no help here, since they do not cover operation at such low voltages (it was apparently at that time not of any interest). For the same reason ordinary simulation programs do not give realistic results. To measure real data all that is needed is to apply a variable grid voltage and measure the anode current (Figure 6). The measured characteristic curve shown in Figure 7 indicates a rise in transconductance with anode current at negative grid voltages. When the grid voltage is positive, the transconductance stops rising, and in the region above +1 V, it starts to fall again. At the same time the grid current rises and, above about Ug = +0.5 V, can exceed the anode current. It is worth plotting the characteristic curve, in particular for used valves, in order to determine an optimal operating point.
|Figure 6 - Plotting the characteristic curve|
|Figure 7 - Characteristic curve for an ECC81 at Ua = 12V|
If it is desired to build the headphone amplifier as a permanent device, rather than merely experiment with it, it might be found that the output volume can be too low in some conditions, especially when operating from a 12 V supply. What is needed is to give the electrons a bit more energy by applying a slightly positive grid bias voltage. Old hands will now protest vehemently that this implies that a grid current will ﬂow, resulting in severe distortion. That is true in principle, but it is not a problem if the grid drive has a relatively low impedance. Previously, in the golden era of valves, the grid had to be driven from a high impedance source, since the output of the previous stage necessarily had a high impedance. Today we can use the low-impedance headphone output of a small CD player, and a little grid current no longer matters. In the circuit in Figure 8 a grid current of the same order of magnitude as the anode current is set up. The anode current and the achievable output drive are now three times higher than with a grid voltage Ug of –0.1 V. This gives almost ten times the output power, which should be adequate for many uses. The grid voltage is set at +0.5 V and the anode current is 0.5 mA. We are therefore in the region of the characteristic curve where the transconductance is constant, and so distortion should be low. The sound of this simple ampliﬁer is indeed very good, even though it might not be perfect from a purely technical point of view. The inevitable distortions introduced by a valve stage, especially when driven hard, are however generally not regarded as unpleasant. Perhaps you have noticed that in this circuit the valve can be simply replaced by two NPN transistors. Instead of a grid current we have a base current, instead of an anode current, a collector current. Of course, the transistors no not need a heater: such is the nature of progress. Which circuit sounds better is a matter of taste: try it for yourself. Most people come to the conclusion that the valve sounds better. It is worth putting up with the fact that the power consumed by the heater is orders of magnitude higher than the output power of the amplifier: in return one can enjoy the cosy glow of the cathode and the opportunity to warm ones hands (carefully!) on the valve.
|Figure 8 - Stereo amplifier with positive grid bias|
Readers with a lust for power who are tempted to try to go a step further and drive the headphone amplifier to the edge of distortion might prefer to wait for the second article in this series. We will be looking at real power valves such as the EL84, EL95, ECL80 and ECL86, as well as a PL504, which we will be using in an amplifier with a loudspeaker output, and at an anode voltage of only 27 V. We will also describe several miniature Russian ‘battery’ valves, which not only work with low anode voltages, but which also dissipate much less heater power.