Application Note AN15.
A way to achieve longer tube life

Last Update:

Description

Compared to indirectly tubes, Directly Heated Tubes (DHT) have disadvantages an advantages. A small disadvantage comes nowadays, while almost everybody heats the tubes with DC, which probably was not anticipated by the historic tube designers, because in those days, most tubes were heated with AC anyway. Today however, most designs use DC, to avoid hum.

A side note

Few people realize, DC heating also increases the maximum output power of the tubes, since the peak value if a 5V AC heater is 7 Volts, and with DC you have 2 Volts more headroom for grid signal. However this is not the subject if the Application note. It is about tube lifetime.

Advantages of DC heating

Disadvantages of DC heating

Start up current of a DHT

A slow start up of the heater current is a very important factor for good tube life. The cold resistance of a tube heater is about 4x lower than the hot resistance. Like when a 300B draws 1.2 Ampere when hot, it will for a short moment draw 4.8 Ampere when cold, provided the power supply can supply this. The transformer builders of the past were limiting this current, by choosing a transformer winding specified of 5V at 1.2A, and no more than 1.2A. Of course, when the tube us cold, it will be a 4x higher load, but if the winding and transformer core are both not over dimensioned, this will make the 5V drop significantly if the tube is cold. So the start current of theoretically 4.8A will not flow, because the voltage drops to 2V or so. Then, when the heater warms up, current will drop and voltage will rise. With correct transformer design, quite a nice method, providing a soft start.

What is bad about a too fast start is not the higher current by itself, but the fast temperature rise, causing re crystallization of the metal, which is the cause for heater breakage with much used tubes.

WIth DC electronics, it becomes quite another situation. If using unregulated DC voltage, the transformer AC voltage will also drop. We have to say it clearly, this is ONLY the case, if the AC winding is designed the right way. Meaning it is not over dimensioned far too much. So when generating the DC voltage from let's say, a 6.3V, 5A winding, this will provide a MUCH higher start up current as the tube likes. And yes, this matters with most tubes. Luckily, with EML tubes build after 2005, this is not an issue, since these are slow-start by it's own design. They protect themselves against too fast warm up, and it is one of the factors why EML tubes have so much more lifetime.

The problem becomes even larger when the circuit is electronically stabilized, with over dimensioned parts. Like when you use an 8 Ampere low drop regulator IC, to supply a 1.2A heater. The IC by itself, easily supplies 4 Ampere cold current peak. That is why a 50 cent 5V IC, capable of "only" 2 Ampere is a much better idea, because it will go into current limiting mode, at 2A, until the heater gets warm, and current drops to 1.2A after heating up. So for 50 cents, you have a safe and nice working slow start circuit.

None symmetric wear out, using DC

The DC voltage is distributed along all (eight) filaments of the 300B tube. Let's say on the left side is +5V, and the right is grounded. This means, the grid to filament voltage on the left is 5V less negative. So when the grid is at -60V, the filament to grid voltage is -55V at the left, and not the -60V you intended. On the right side however, it is -60V indeed. Is this a small difference? No! It is very large. The whole tube has 5mA/V transconductance, but each filament wire only contributes with 1/8 to that. Make the calculations, and you will see it is 3mA more. For the whole tube that may seem not much. But for this one filament it is a lot. At 80mA total current, it draws 10mA, and you add 3mA to that, due to none symmetric electric fields inside the tube. Note, the same problem exists from the top to the bottom, but here it is a smaller problem..

The problem with this

It causes uneven wear out of the filaments on the left side.

The solution

It is not difficult. This is why some prefer AC, even sacrificing the grid signal head room. With DC, the solution can be very elegant. For one channel just connect the filament in REVERSE with respect to the other.

Q: Will this do something, because the filaments are equal in function?

Yes it will! Lets call the filaments a "left" side and a "right" side.

So one tube has -55V on the left side. The other tube has it on the right side. This makes no difference for the function of the amplifier, but now you can occasionally swap the left and the right tube. Like every 300 hours or so. WIth auto bias there is nothing else to consider. With adjustable bias, it requires a re adjustment. This will give even wear out along the filament. So for lifetime it is better. Another reason to do this, is uneven wear out is also a source of harmonic distortion.

Most of all, this method costs you nothing, only more intelligent wiring.