Carbon composition resistors in crystal oven - bad idea

[TerraHertz]

Working on that GRC-106 radio some more, at the moment trying to find why the 5MHz reference doesn't run when cold.

This thing uses a crystal oven, and most resistors throughout the radio are molded carbon composition, including all the ones inside the crystal oven. I've noticed that every single one of them in the oven is high.

For eg a 3K9 is 4K03, 3K3 is 3K49, 4K7 is 4K91, 270R is 287R, etc.
They're 5%, so still borderline in tolerance.
I've seen carbon composition resistors go high out of spec before, but never *all* of them.
Has anyone else noticed this type of resistor in old (1960s) gear all failing when run warm?

It will be quite annoying to find I have to replace all the resistors in all the ovens of these radios (I have ah... a few.)

[SeanB]

Sorry Guy, they all will drift high when exposed to gentle heat for a while, as the wax evaporates from the resistor and the pressure on the carbon composition rod inside reduces from this.  Try measuring one and then simply apply pressure from your fingers on it, it will drift, or just unsolder it and solder leads on then press the rod and it will change again.

Just look for those that are most critical and change them, the ones that do not provide biasing would probably be fine to leave.

[TheMG]

Most carbon composition resistors do in fact drift over time. Exposure to high temperatures can worsen the problem. I see it all the time in the stuff I repair at work.

They are notoriously unstable, which is why metal film resistors have replaced them in most applications many decades ago.

[vk6zgo]

Early film resistors looked like inductors,due the to the "clever" way they adjusted their values------a long spiral was cut into the film.
For that reason,carbon composition resistors hung around for longer than they might otherwise have done.

The values quoted are  close enough to not worry---the design values may well have been 4k,3.5k,5k,& 300 Ohms in the first place,with the nearest standard values selected on test.

[csmithdoteu]

I had a Tektronix 453 oscilloscope from that era for a while and that was full of carbon composition resistors. I think every one had drifted up somewhat in its 50 year lifespan at the time. I had to give it up after a bit as it was eating unobtainium and expensive legacy parts that had drifted or died faster than I could get them.

[TerraHertz]

Early film resistors looked like inductors,due the to the "clever" way they adjusted their values------a long spiral was cut into the film.

What do you mean 'early'? All the axial metal film resistors I ever broke open are constructed like that. Are there ones that are not? Sources?

I'm trying to think of what I can replace carbon composition resistors with, in general in this radio. To avoid exactly that inductance effect.
I have stock of all values in 0805 surface mount resistors, but soldering little wires on and using like axials won't be reliable. Solder fractures and in some cases, insufficient dissipation.

For that reason,carbon composition resistors hung around for longer than they might otherwise have done.

Yeah, I knew relatively flat impedance vs F was their advantage. But I don't want to buy a range of carbon composition resistors!

The values quoted are  close enough to not worry---the design values may well have been 4k,3.5k,5k,& 300 Ohms in the first place,with the nearest standard values selected on test.

In this crystal oscillator, I think they actually are the problem. It's behaving like the loop gain is *just* a little bit below unity, until it warms up. There's still one other thing to check though. Not the crystal, that's definitely OK.

On the osc-buffer A1 module, I suspect R2 & R4, which are both drifted high. Still to check Q1's gain. Plus maybe T1 is mis-adjusted due to the slug being loose.

Another question. What would be a good inductor slug-locking paint, for use in a crystal oven? The one originally used has decomposed. I'd normally use nail varnish, but I don't know what it would do when hot. Silastic won't do because it's flexible.

Schematic, 2000x1300:

[Vgkid]

What about using thin film resistors.

[CatalinaWOW]

My first answer would be that the carbon composition resistors worked for almost 50 years.  If you replace them and they go another 50 years it probably won't be a problem for you.  You could even extend the likely life by picking units on the low end of the desired range.  Since carbon composition always drifts up they might last 75 or 100 years.

ARRL suggests that newer film resistors are comparable in inductance to carbon composition.  I don't know the basis for that assertion, but I have seen many film resistors that use a relatively short trim cut in a geometry that wouldn't be particularly inductive.   Ohmite TFS series might be a good choice.

Metal foil resistors are asserted to be low inductance.  I have no hard data to support this assertion.