Heater / Filament Efficiency - Preliminary Report
The purpose of this short report is to discuss the efficiency of various heating structures available for vacuum tubes. It covers Tungsten-Thorium filaments, coated filaments and heater-coated cathode structures. The data presented represents data contained in the RCA TT5 manual. This is a preliminary report. Gary Longrie was kind enough to review an early edition of this report. Later reports will attempt to account for variation in cathode temperature from device to device. This report lumps all H-K devices together, as it does with coated filament devices, and W-Th filament devices.
How much cathode current can you reasonably expect to draw from a particular vacuum tube? One answer is to look at the maximum "rated" current for that tube. But this information is not always available. The real current that can be drawn is related to the heater power and the type of emitting structure. The more heating power available, the higher the cathode emission will be and thus, the more current that can be drawn before becoming emission limited. Some emitting structures are more efficient than others. As we will show, the most efficient structure is the directly heated coated filament. This means that THAT structure allows the most current to be produced for a given heating power. However, we can also plot the maximum current allowed (per datasheet) vs heating power for a number of devices. The average of this curve then provides an indication of what heating power supports what cathode current as described below.
I entered data from 64 different devices from the RCA TT5 into a spreadsheet; sorting the devices into maximum current allowed, heating power required and type of emitting structure. I then graphed the results, adding a trend line (aren't spreadsheets nice for some things?). Here's the results:
Notice that there is some variability in the ratings for each class. As Gary Longrie pointed out, there may be a difference in the temperature that different devices are heated to, and the active emitting area with respect to the heated area for an individual type may be slightly different. Why? Each individual tube type may be designed for different applications. You can sometimes infer this from the description of the device's application: one may be expected to provide long life, so the rated current is lower (and the filament burns slightly cooler). Another device may be rated for shorter life but for "intermittent" applications where performance is more critical than life. In this device, the filament may be heated to a higher than "average" temperature to increase emission at the expense of life.
Other devices may be constructed such that not ALL of the heating structure contributes to the overall plate current. Gary found a few old references to substantiate these conjectures. For instance, the Proceedings of the IRE for Oct 1929 (!) contains a nice article by Yuziro Kusunose that evaluates triode characteristics by taking into account the "active" vs "inactive" portion of the tubes structures. Due to constructional details, not all of the filament is effectively used.
A later revision of this document will attempt to factor in the emitting surface temperatures, by accounting for the cold vs hot resistance of the filament (and by inference, the emitting temperature).
Let's take a particular current and look at the heating power required to obtain a specific current. For an example, let's pick 100 mA current. For a H-K structure, about 6.2 watts is required. For a coated directly heated filament, just under 5 watts is required, and for a thoriated tungsten filament about 12 watts is required. Note that the directly heated coated filament is the most efficient heating structure in this current range. (The 2A3 and 300B fall into this class).
We can also use this data for some "predictions". A 5687 is a small signal double triode that consumes about 5.6 watts of heater current. Typically about 20 mA per section (40 mA) is used. Since this tube can be "expected" to provide about 90 mA, the lifetime should be expected to be long. No one has complained about the life of these tubes. In fact, the data suggests that 45 mA per section (90 mA total) could be obtained from this device and still achieve a "normal" life. On the other hand, a 5842 (417) triode consumes about 1.9 watts of heater current. These tubes are normally run at about 25 mA. Looking at the graph above, this also slightly above the average expected from a 1.9 watt heater. Lifetime *may* be an issue, and is sometimes complained about with that device.
Example of Other Uses for this Data
An 801 is a DHT with a 7.5V, 1.25A (9 watt) Thoriated-Tungsten filament. It is "conservatively" rated for 85 mA cathode current (in Class C, FM service, 600 volts and 70 mA plate input produces about 25 watts output and 17 watts plate dissipation; under this condition the grid current is 15 mA, for a cathode current of 85 mA). The curve above also shows a "fit" for 85 mA at 9 watts filament power.
In some applications, lowering the filament voltage on DHTs considerably lowers the distortion. Will this harm the life? Using the example above, suppose you wanted to run the tube at 6.3V (and 1.1A). The filament power is reduced to about 7 watts. Extrapolating this from the W-Th curveset above predicts that the maximum "safe" current under this starved condition is 70 mA instead of 85 mA. Therefore, I should anticipate a shortened lifetime with currents approaching 70 mA.
In one such application, I'm running that particular device with 6.3V on the filament, but only at 10 to 12 mA of plate current. Since this is considerably below the estimated "safe" 70 mA current level, I do not anticipate any problem.