The Valve Wizard

How to design valve guitar amplifiers!

Heater / Filament Supplies

Valve heaters generally require much more current than the rest of the amplifier.
Valve heaters can be run in parallel or series, or with a little juggling a combination of the two. The common dual-triode pre-amp valves such as the ECC81, ECC82, ECC83, 12AY7, E88CC, all have three heater connections, so the heater for each triode can be wired in parallel, or in series at half the current and twice the voltage.
Most of the common valve types used in guitar amps are designed to have the heaters run in parallel from a constant voltage source. That is to say, they all operate at the same voltage (usually 6.3V) but may have different current demands.
Power transformers designed specially for use with valves will usually have a secondary winding solely for the heater supply. It is important not to exceed the maximum current rating of the transformer. If the heaters are run in parallel then it is a case of adding up the current drawn by all the heaters and checking that it does not exceed the transformer's rating. If it does, a separate low voltage transformer could be added so some heaters are run from the main power transformer and some (or all) from the separate heater transformer.
A power indicator lamp can also be run from the heater supply, provided there is enough current available.

Most audio valves have indirectly-heated cathodes and can be run from an AC or DC heater supply, although AC is of course simpler to implement. Some valves rectifiers (e.g., the GZ34) have the heater internally connected to the cathode (to ensure the heater-cathode potential cannot be large) and will need their own separate heater supply. If possible, even indirectly-heated cathode rectifiers should be run from a separate supply in the same way [right]- this will ensure long life [see the sections on full-wave and bridge rectifiers for more].

Voltage considerations: The heater voltage specified in the data sheet is the optimum value (usually specified as +/- 10%). Running them at higher voltages will reduce valve lifespan and should be avoided. Running them at lower voltages, within reason, will increase their lifespan and reduce hum. However, if they are run too low then emission will suffer. Running heaters under-voltage is therefore perfectly acceptable. Heaters rated at 6.3V can be run quite happily between 5.7V and 6.9V. However, avoid runing the heaters of rectifier valves low, as they usually operate close to saturation and need as musch emission as they can get.
It is not uncommon for the mains voltage to be slightly high, resulting in a heater voltage that is also proportionately high. If this is the case where you live then the following may be useful: A pair of ordinary high-current silicon diodes can be added to the heater chain to drop the heater voltage by about 0.7V. This method can also be used for a heater standby switch.

Series and parallel: Heaters which are normally designed to be run in parallel can be run in series provided you ensure their current demands are met correctly. For example, an EL84 and ECC83 could be run in series from a 12V supply. The EL84 is rated at 0.76A while the ECC83 is rated at 0.3A, so a resistor must be placed in parallel with the ECC83 to pass the additional current. We want to pass 0.76 - 0.3 = 0.43A through the resistor, and we want the voltage across the resistor to be 6V. Use Ohm's law to calculate its value:
6 / 0.43 = 14 ohms.

The power dissipated will be:
P = I^2 * R
0.43^2 * 14 = 2.6W
So we would probably use a 15R, 5W resistor.
All sorts of heater chain combinations can be created in this way.

Because the heaters are in series it doesn't matter in which order the valves are wired. However, if one end of the (AC) heater chain is to be grounded then the most sensitive preamp valves should be closest to the grounded end of the chain where the voltage swing (i.e. hum field) is smallest.
Since the different heaters will often have different warm up times which could put stress on the other valves, a thermistor can be placed in series with the chain, or a resistor switched in and out by a standby switch [see the section on power and standby switches].

Reducing hum in AC heater supplies: AC valve heaters cause hum because the filament radiates an electro-magnetic field which can induce a hum voltage in the grid / anode via stray capacitance. This can be minimised by using the smallest heater voltage allowable (e.g. wire up ECC83 / 12AX7s to use 6V instead of 12.6V) and by using a centre tap or humdinger to balance the heater supply voltage [see below]. Hum is also caused by direct leakage between heater and cathode. This can be reduce by biasing or 'elevating' the heater supply.

Hum is also worsened by having a large cathode-heater resistance, which is the case for cathode followers and long tailed pairs. Luckily, these stages usually come after sevral gain stages, so the signal-noise ratio is good by that point.

Single ended stages are most prone to hum, whereas a (correctly wired [see below]) push-pull stage or differential pair will tend to cancel any common mode noise like heater hum.
Power valves tend to be less prone to hum since they deal with large signal voltages, so the signal-to-noise ratio is higher. In fact, in most amps, most of the audible heater hum comes from the input stage.
The following tricks can be used to reduce heater hum:

Transformer centre tap: The traditional way to reduce hum is to use a heater supply with a centre tap and connect it to ground. This creates a balanced voltage supply, e.g. +/-3.15V instead of 0V and 6.3V. This heps to induce equal-but-oppiste hum currents into the audio circuit, where they will (in theory!) cancel out.

Artificial centre tap: A better way that can also be used with non centre-tapped heater supplies is to create an artificial centre tap with resistors. The resistors should have a low resistance sto mimise susceptibility to picking up any radio noise or high-frequency noise from the transformer primary. Values of 100R (1/2W min) and 220R (1/4W min) are usual. They will of course cause a small amount of extra current draw from the transformer (32mA when using 100R resistors at 6.3V) so bare this in mind. This resistance between the heater winding and ground will also act as fuses in case of a short between the anode and heater pins of the power valves.
Another traditional method uses a potentiometer with the wiper grounded- a so-called "hum dinger". This allows minimum hum to be dialled in precisely, and is highly recommended.

DC elevation: DC elevation is often used when a valve in the circuit has a high cathode voltage. The heater voltage is elevated to a higher level to avoid exceeding the maximum heater-cathode voltage rating of the valve. This is done simply by 'adding' a DC voltage to the heater supply. The heaters still operate at 6.3V (or whatever you're using), but the AC component 'floats' on top of a DC voltage.
This method is also used to reduce audible heater hum by reducing or saturating any leakage current between heater and cathode*. Elevation voltages used are typically between 10V and a few tens of volts.

A typical way to apply the DC reference is by connecting the centre tap (real or artificial) to the cathode of a power valve, provided the power valve is cathode biased of course. The bias voltage of most power valves is usually more than 5V, and this will be 'added' to the heater voltage.
The other common method is to take the DC reference from a potential divider from the HT (useful if the amp has fixed-biased power valves). Typical voltage references are around 20V to 90V, placing the heater supply well above the potential of most cathodes in the amp.
The potential divider should have a fairly high resistance so there is no significant current drawn from the HT (it can also serve as the bleeder path for the HT smoothing capacitors).
The lower resistor in the divider (R2) should not be excessively high or the maximum heater-to-cathode resistance may be exceeded. Many data sheets do not quote this so it is advisable not to make it greater than 100k. A fairly large value capacitor (C1) can also be added to ensure a smooth DC reference. It's actual value is not critical, anything over 10uF should be fine.

DC heater supplies: Properly designed DC supplies do not cause hum since the stray current between filament and cathode is constant and therefore inaudible. However, DC supplies nearly always need to be voltage regulated or they can cause even more noise than an AC supply (although simple rectification to DC does sometimes work).
Simple, three-pin voltage regulators usually require an input voltage that is at least 2.5V above the output voltage in order to work. Most regulators cannot handle more than about 1A of current on their own, so it is quite common to operate the comparatively low current pre-amp valves from a simple regulator, and run the current-hungry power valves from an ordinary AC supply, since they are less prone to hum anyway. Higher-current regulators are available, however, at slightly higer cost. The regulator must always be fixed to a suitable heat sink.
The simplest and most popular range of fixed-voltage regulators is the 78xx series. A 5V regulator can have its output raised to ~6.3V by elevating its ground terminal by 1.3V; the voltage drop across a pair of silicon diodes is perfect for this. Higher voltages could be obtained by using zeners instead, but the input voltage must always be at least 2.5V higher than the output. 6V regulators do exist, but are not as commonly available as the 5V versions.
In the circuit below, the 7805 regulator can provide up to 1A on its own. If more current is required, the transistor can be added to provide up to 5A max (it too will need a heat sink). The 1uF capacitor must be positioned very close to the regulator. It does not have to be tantalum, a ceramic will do at a pinch. A normal 6.3V transformer winding will NOT provide sufficient voltage after rectification to power this circuit.

Layout / lead dress: The lead dress of AC heater supplies is very important for noise reduction. The AC heater wires will have significant EM radiation and should therefore be routed well away from all signal wires, and are usually tucked into the corner of the chassis. The wires should either be made from twin cable (bell-wire) or better still, should be made by twisting the wires neatly and tightly together. In this way the wires are kept perfectly parallel and close to each other, which increases opposing field density and encourages the radiated fields to cancel out. Loosely twisted wires are no use at all.
When heaters are wired in parallel; power valves should be first in the heater chain, followed by driver valves, with the input stage being last in the chain. This keeps current, and therefore radiated fields, at a minimum around the most sensitive stages of the amp. Even better is to run the pre-amp and power-amp sections from separate heater chains. If signal wires must cross the heater wires, they should do so at right angles.

Valves in push-pull or in balanced stages (such as long tailed pairs using separate valves) should have their heaters wired in phase. Any noise induced will then be common mode and rejected by the stage (mostly). Valves in parallel single-ended stages should have their heaters wires out of phase for mutual cancellation. Using two different colours for the heater wires will make this easier.
The common pre-amp valves (ECC83 / 12AX7 etc.) when run from a 6.3V supply, should be wired from one side only [see right], not by looping one heater wire all round the valve socket, which would create a hum loop and cause excessive interference noise (though many amp makers DO make this mistake and get away with it). The wire twisting must be kept very tight right up to the socket, where it matters most. Their pin arrangement is also deliberate, so that the main heater pins (4 and 5) can be orientated towards the chassis wall, allowing heater wires to be run along the wall away from any other sensitive signal wiring.

A universal heater supply?
Everyone likes tube rolling, but it is somewhat dissapointing that the only valves which are compatible with the ECC83/12AX7 pin-out are the ECC81/12AT7, ECC82/12AU7 and 12AY7. But there are many other valves which conform to the more standard pin-out such as the ECC88/6DJ8, ECC85/6AQ8, 6N1P, 6N2P and other Russian types. Although we could provide a switch at every valve socket to select between the two pin-outs, it would be nice if we could just plug in any type without any changes. The following circuits are designed to allow this, but automatically detecting which type is inserted. All these circuits operate the ECC83 types from 12.6V. If an ECC88 type is plugged in, however, a zener diode is placed in series with the heater to limit the heater voltage to about 6.3V.
Each circuit uses pin 9 on the valve socket as a control port. If an ECC83 is plugged in this pin goes positive, causing the SCR (U1 left-most circuit) or NPN transistor (Q1 middle circuit) to turn on, shorting out the zener. There is still a small drop across this device though, which is why the supply voltage is shown a little higher than 12.6V (it will probably be a little higher in this mode anyway, since an ECC83 only needs 150mA heater current).
When an ECC88 type is plugged in, pin 9 is connected to nothing, so U1 or Q1 turns off, and current is steered into the zener diode, which drops roughly half the supply voltage across itself.
For AC heater we simply use an NPN/PNP pair connected in parallel (Q1/Q2 in the right-most circuit). A triac or even an opto-triac could be used too. However, zener diodes can't be used in the same way for AC, so a resistor is used instead. Unfortunately this means that some of the Russian valve types (notably the 6N1P) can't be used, since they require up to 600mA current which causes too much drop across the resistor. Most European/American types and the 6N2P should be ok though. Another disadvantage of this design is that it is not balanced, so a humdinger pot will probably be needed to null any heater hum. None of these circuits have been tested yet, but are presented for enthusiastic builders to try out, so let me know if you do!

*See: Cooper, C. E. (1944). Valve Hum. Electronic Engineering, (July), pp72-5.