8072. 8560, and other conduction cooled tubes

[Opcom]

I'm not planning on doing anything with an 8072, but have questions about how these things are cooled and used. Maybe someone has some hands on experience with them. It supposedly is rated 100W dissipation with 'simple' cooling techniques, and up to 300W with better setups, but the RCA TT-5 manual is pretty vague.

The 8072 is about the size of an 8122 less the anode radiator and it looks like that is what it is. It's supposed to be used with a heatsink and is rated 100W.

The 8560 is larger, with a solid anode about the size of a 4CX150 or 8122, having a flat on one side. It is supposed to be insulated by a SK-1920 BeO insulator (having a thermal resistance of 1.9 deg C/Watt of anode dissipation), and is rated 200W dissipation at 250  deg C anode temp.
Eimac said: 250 deg C - (200W/(1.9 deg C/W)) = 145 deg C where 145 deg C is heat sink temperature. The math is written better in the datasheet.

If I recall, Heathkit made an amp using 8560's. They touted how it was silent. The whole back side was a big heat sink.

Considering that BeO and other insulators don't really conduct heat as well as metal, wouldn't an amp cool better if the tube was clamped right to an insulated heat sink? Or, does the BeO slice in such HF amps also act as a capacitor to reduce the effect of having the tube in contact with a big hunk of aluminum?

Here's a nice picture of an amp with one.
http://www.qth.net/k0awu/Amplifiers.html

[WBear2GCR]

Sure a direct connection is probably better... but what are you going to do with the HV that would then be sitting on the heatsink, making it electrically hot? Also, there would be added capacitance that way. It is (of course) the plate connection that is being cooled.

The Heathkit was known for frying tubes, fwiw. Don't think it got rid of the heat quite well enough...

                     _-_-

[W2VW]

The Heath amp used a triode, 8873. Same cooling idea.

[WQ9E]

I have heard that the 8072 will run at 8122 ratings IF proper cooling is provided.  There are reports of some people removing the anode cooler from dead 8122 tubes and grafting them onto 8072 tubes because of the price difference between the two.  The 8072 was used by Motorola and others so it is readily available and reasonably priced.  I have a Signal One CX-7 that uses a conduction cooled 8072 final.  Signal One recommended a cooling fan for RTTY use and the heat sink does get very hot in normal operation.

The Heatkit SB-230 used a single 8873, I also have a Henry K-2000 (cosmetically matches the Kenwood TS-520 series) which uses a pair of 8873 tubes.  These triode tubes had a short production life and are even more expensive and less available than their normal and transverse cooled 8874 and 8875 siblings.  My SB-230 was working and in very nice condition for $75 and I paid $125 for the mint and working K-2000.  When the tubes die I will look into a conversion to something more available, probably to a Soviet type.

As Bear pointed out you have to have an isolating link between the anode and heat sink both to avoid HV and excessive stray capacitance.  BeO is the typical product used and it has its own significant set of cautions.  Unless you have cheap access to some desirable high power tubes I wouldn't bother with a modern conduction cooled design.  You will either need a very large heat sink for high power at high duty cycles or a moderate sized sink with forced airflow.  Alignment of the tube anode surface, insulating link, and sink are critical and issues here will destroy the tube.  Even with proper setup there is additional thermal resistance making it possible to destroy a tube, just like a transistor, with a temporary overload.