Thermal Runaway

[dai h.]

Yeah, ties in nicely with what Steve and R.G. mentioned above. Here's a bit more I found on the (cooling) specifics:

Tube Trivia

Because the grid is next to the hot cathode, and the outside of that is the hot plate, there is a tendency for (grid) temperature to rise. However, if the grid gets hot, "emission" will take place, and electrons from the grid will jump into the plate. In other words, current will end up flowing from the plate to the grid. In order to prevent this, there are various measures taken to cool the grid. The supports for the grid stick out out the white-ish plates (these are called "mica"[Japanese term used for the mica spacers] since they are made of mica), and attached to these are small metal plates several millimeters square. Tubes such as 6AH4GT and 6BX7GTs have such parts. These tiny plates have a role in radiative cooling. The bottom portions connect to the tube base through leads, and the cooling from thermal conduction via this path through the thin leads is also important. With tubes that have a tendency to exhibit grid overheating, (the grid posts) are connected to the socket pins using two or more lead wires. This is a measure taken to increase (conductive) cooling paths as much as possible...[then some stuff about the getter]

by T.Kimura
Building My Very First Tube Amplifier

as a corollary, I suppose it points to whatchamacallit(thermal design?) being an important aspect to think about. Some pro tube preamps come into mind (some are super well ventilated with numerous holes on the chassis). Don't "box" the tubes in so heat can't escape, etc.

[jmaf]

Thanks Steve, that helps. so, any input on the issue of some tubes being more susceptible to this than others?

Depending on how they're built even same designation tubes from different manufacturers may have different traits. Some EL84's for example tend to dissipate more wattage on the screen grids than others at the same level. When changing el84 tubes some will make the screen glow during loud playing, other brands will not, same cathode resistor, same setup, it's probably the way the screen is aligned with the control grid. Some folks have reported adding a 3k3 screen resistor to el84's to limit current.

The same thing may happen with the control grid to some degree, they're both in series with the electron path, the control grid being closest to the cathode.

As an example, the KT120. I have been wanting to build an amp based on the new KT-120's(for about a year now...I'll find the time, someday...).

I asked for feedback about that tube from the more senior folks here and learned then that the problem with the KT120 is its propensity to drive current from out of the grid like Steve Conner explained. If your bias supply is "hard" and it can take the loading the low impedance a positive grid creates, this won't be a problem. The problem is to create a regulated bias voltage when you can't increase the bias supply impedance much, so you can't just create a resistor divider and just hope it works like with smaller power tubes. When you create a lower Z bias supply, that's also a path for the signal to be grounded, so you also need a powerful signal drive to make it through.

The KT-120 requires at most 50K ohms(IIRC) source bias impedance or it'll tend to go into thermal runaway, and think of the kt120(60W) as over 1.5 x the power of an almighty 6550(~ 36W), the one you see melted on the photo above. When that thing spirals down, it'll probably take a transformer and other parts with it unless you've got appropriate fuses or cathode resistors.

[Steve Conner]

Thanks for the useful information Dai! The first tube amp I ever tried modifying came with a set of Matsushita tubes.

Jmaf: new year's resolution?

Incidentally that is one reason why big power amps like the SVT and Marshall Major used cathode followers between the PI and the power tubes. The paralleled big tubes add up to lots of positive grid current, and a low grid leak resistance is needed to get reliable operation. Then the cathode followers are needed to drive the grid leak resistance.

So it might be a good idea to include some in your KT120 build, or use a low impedance tube like the 12BH7 with 33k plate resistors for your PI.

[jmaf]

Jmaf: new year's resolution?

Incidentally that is one reason why big power amps like the SVT and Marshall Major used cathode followers between the PI and the power tubes. The paralleled big tubes add up to lots of positive grid current, and a low grid leak resistance is needed to get reliable operation. Then the cathode followers are needed to drive the grid leak resistance.

So it might be a good idea to include some in your KT120 build, or use a low impedance tube like the 12BH7 with 33k plate resistors for your PI.

Thanks for the ideas, Steve. I've been looking at the SVT drive circuit for a few weeks now, I would rather not hijack this thread with questions about it, but funny you should mention it because I was thinking of using something similar to that or EL84's / 6V6's to drive the kt120's. Speaking of new year resolutions, happy new year to all!