Chassis ventilation - Worth the effort?

[exclamationmark]

I'm not saying the temperature outside the tube envelope is irrelevant, it most certainly is! Just that by blowing air over a tube may not be as effective as you think it would be. Case in point is a little prototype I did with some 6c33c's (pass regulator tubes that were used in MIG aircraft, among other things). Using 4 of them, they dissipated around 350 watts at idle. Because the plates were very strongly connected to the base of the tube (as well as the heater pins), any crap left on or around the socket would begin to turn black over a period of time (including the socket itself). I actually stuck fans under the sockets to alleviate the problem, as well as blow any potential heat up and away. Inside the chassis wasn't too bad - the fans helped vent the hot air, and sucked in cool air.

The other thing I found extremely disturbing was that by bringing your hand anywhere near the tubes, it was reminiscent of a campfire... The air wasn't warm, it was just pure radiant heat. I wouldn't have even considered touching the glass! I also had a filter cap mounted on the top of the chassis a good 20cm away and it got HOT (though the fact that it was black couldn't of helped!). It got warm normally, but this was like scalding water hot. I can imagine if I put the tubes inside an enclosure, all the fans in the world wouldn't have helped keep components within a 20cm radius cool. Eventually I abandoned the project because the tubes were kind of expensive and unreliable (funnily enough for being used in fighter aircraft...). It was actually an OTL (output transformerless) amp, so it made a disappointing amount of power, but I may revisit it again with some other tubes (and maybe put some transformer coupled amps to shame!).

Definitely add a fan to chassis if you want, as I know from personal experience that it works (and I know that it accelerates dust build-up too!). But as for cooling the tubes directly, I'm still quite sceptical. Maybe if you did it in the fashion outlined above , where the air was sucked in through the chassis, and exited over the tubes, it would perhaps work. But for tubes mounted outside the chassis, with an exterior fan... that sort of seems useless to me.

[trobbins]

The 6c33c certainly looks like a beast! At 90W dissipation by heater and anode, it is certainly starting to separate from most of the typical cooling issues of the likes of KT-88 or 6L6 etc etc. As you indicate, a fair few watts are meant to be 'sunk' out through the pins, and so they should be set up to sink heat away from the contact surface and dissipate it elsewhere - I'd be aiming for thick copper cables or some similar technique at least.

The surface area of the radiating anode plays a part - to whit the 6c33c looks to be large - given it has about 2x the rating of KT88.

The datasheet certainly identifies that the glass needs to be kept cooled, and indicates the service life tradeoff with glass temperature that you may experience - with life shortening by a factor of 10 for glass temp increasing from 260 to 300C.

The glass for a valve is like any other smooth surface when it comes to convection cooling in to the air - a vertical surface is better than a horizontal surface, and any increase in air flow by forced air arrangement will increase the heat flow and hence the surface temperature - there's a lot of basic physics on this, extending out in to heatsink design.

The physics of the plate radiation, vacuum, glass absorption, air temp and velocity etc is complex. It's probably easiest to simplify by thinking that 50% gets radiated past the glass in to the surroundings - and hence a cap or transformer etc standing up nearby will get some heat added (like placing the part in the sun - so a baffle, or forced air cooling the surrounding parts will likely return their skin temp to normal. The amount of heat in the radiation 'dilutes' pretty quickly with distance - sort of like antenna radiated power, or placing you hand near an olde world incandescant light bulb (if you can find one nowadays!).