Heat from cathode resistors & sag resistors

[potatofarmer]

Hey all,

I'm currently designing a single-channel Vox-ish amp with a quad of 6V6s. Same sort of idea as a Trainwreck Rocket. I'm building this in a donor amp, a 1x12 Crate Steath 50 combo which had a fried PT. The power tubes are going to be cathode-biased, and I'm also going to add switchable sag resistors between the PT secondary and the bridge rectifier (planning on using an Antek PT).

That's a whole mess of heat though, and the chassis and cab are both essentially sealed. edit - basically the chassis is a sealed box, half wood, half metal. The combo itself is open-backed.

For the cathode bias resistors (right now I'm leaning towards individual resistors per tube, possibly zener-assisted) I'd have no problem getting the aluminum-cased resistors and mounting them on the outside of the chassis with the leads covered in heat-shrink. For the sag resistors though, that's no bueno.

So, I could try to find room to mount a fan, or I could bolt these resistors to the inside of the chassis and bolt a heat sink to the outside of the chassis. Or cut a hole in the chassis and mount the heat sink over it and attach the resistors to it somehow.

Hm.

I'm going to have to install a fan, aren't I? Maybe even cut a vent into the cabinet... With the weird layout of the chassis (all the tubes dead center!) I'm not sure how much room I can find for a fan.

Since I'm looking at a couple of watts for the power tube biasing resistors (let's go nuts and say 10W) plus the sag resistors (probably close to another 10W) plus the tubes are hanging down so the heat from the envelopes will be hitting the chassis - how much heat can I expect to be able to pump into the chassis safely? I'm primarily shooting for safety and user comfort here, but component longevity is certainly appealing as well.

This thread has been helpful, as well as confirming my suspicions are somewhat warranted: http://music-electronics-forum.com/t11813/

Anyone have experience with building cathode-biased amps into sealed chassis?

[R.G.]

A grain-of-wheat light bulb is a bit of tungsten wire that dissipates maybe half a watt or less, but it's set up so the heat can't get out easily.

The rules of thermodynamics say that if you generate heat inside a box, what's inside increases in temperature until the heat coming out of the box equals what's being put in. There is ... no ... limit... to that temperature. So if you can't get the heat out by conduction (i.e. heating the metal chassis/box) or convection (circulating air through it), the insides will increase in temperature without limit until the internal circulating air heats the metal chassis enough to get the heat out. The number of watts is not so much an issue as how hard it is for heat to get out - as witness the grain of wheat bulb where the tungsten filament its a few thousand degrees on less than a watt. The only way to get stuff to live through this is to figure out how the heat will get out and make it easy for the amp to do.

There are crude approximations to heat flow across boxes, etc. but the short answer is your amp will be a lot happier if you let some air though somehow. There are other high tech solutions, but the simplest solution is to figure out how to let hot air flow out at or near the top and let cool air flow in at or near the bottom to replace it.

[potatofarmer]

R.G., thanks for elucidating that.

I was focusing on the added wattage of the resistors because this was (presumably) at one time a working product. It was a PCB amp, and even though the tube sockets were PCB mounted, the PCB itself didn't show any obvious signs of heat stress, so I assume the chassis was handling the heat generated by the tubes, if not well at least passably.

My concern with adding passive convection would be the possibility of allowing the tubes to heat the chassis internals via convection as well. I suppose that's more an issue of carefully placing the inlet vents - I'm thinking far inside, with the outlet vent at the top rear, or even through a screen at the rear faceplate.

I did bust out my copy of Morgan Jones's "Building Valve Amplifiers" and he indicates that, with the proper use of heatsink compound, the chassis does work pretty well as a heat sink, and that the heat tends to stay localized.

Aah, over-planning. I haven't even busted out the thermocouples yet...

[R.G.]

Thermocouples were once all the rage. Today we can get infrared remote temperature readings with a $30 heat sensing gun. Point and measure. I bought one and threw away all my thermocouples and thermistors. Never looked back.

Chassis as a heat sink works. At worst, half the heat comes back off the metal inside, and you can force that to be nearly zero with a non-thermally-conductive coating inside - but the coating stuff is way over-teching it. But vents work well too.

The barest whisper of moving air makes thing immensely better compared to closed boxes.

By the way, major kudos to you for even thinking of heating. Most people do that at the end when things can't be changed very well, or when they burn their fingers.

[nickb]

A word of caution here. Many IR non-contact thermometers assume an emmisivity of the target to be around 0.95 i.e a black non reflective surface. The errors can be huge if something else. More info here http://support.fluke.com/find-sales/...50_ENG_B_W.PDF