Heater voltage.

[PerdioPal]

That is usual for an FM detector, has you radio got FM?

[Mike Watterson]

Great, got myself a 150ohm 25w wirewound. I have noticed something on the chassis I don't understand, a 4uf cap with positive to earth?, It looks original, it's been like that for 70 years?. Am I missing something?.

No, that's totally normal for the FM detector. Those often dry out and then reception is poor. About 0 V with no station and -15V to -30V or even more with a strong station. A 3.9 uF or 4.7 uF 50V part is fine. Plus to chassis as the the DC is based on signal level and negative

[PerdioPal]

I wonder how many sets have been re-capped with this the wrong way round?

[catch29]

It's reading 2uf so I'll replace it pos to earth, learn something every day!.

[ppppenguin]

A plastic dielectric part is also fine. I have some very small (about 10mm cube) 4.7uF 63V parts in the drawer.

In just about all cases where a low value (up to bout 4.7uF) electrolytic was originally used, a plastic dielectric (polyester, polypropylene etc) is a superior replacement. The only exceptions will be where the poor characteristics of an electrolytic were used by the designer. These cases are, or should be, very rare.

[Mike Watterson]

I wonder how many sets have been re-capped with this the wrong way round?

Usually won't work for very long!

[PerdioPal]

Modern non electrolytic caps. can have a ridiculous capacity for size. I have found a 1210 (size of a grain of wheat) ceramic capacitor with 100uf inside it, how does it all fit?

[ppppenguin]

By having abysmal performance in some other respects. The capacitance is very sensitive to temperature and also varies considerably with applied voltage. Worse than electrolytics. Used in signal paths they can cause intermodulation if there's any significant signal voltage across them. Used correctly, for example as decouplers, they can be very useful.

[Mike Watterson]

Also piezoelectric!
A problem with many ceramic caps.

Charge up a 50 year old 1nF 5 kV cap and leave connected for 20 minutes Short it, isolate it. Come back 20 minutes later and measure the voltage. It will be 100s of volts. The 5kV deformed it. So later it "recovers" and voltage rises as it does. Even more "interesting" with 10 kV or 20 kV parts all in parallel on a soak testing tray. The solution is NOT to short the tray, but put a bleed resistor.

You can guess how I discovered this 50 years ago.

Smaller values work as poor microphones.

[ppppenguin]

It's worth adding that these problems don't affect low value ceramics, sometimes known as NP0 or C0G, which are accurate and stable. As you go over 1000pF you need to use dielectrics that have worse properties ushc as X7R. Not too bad up to 100nF, pretty horrible at 10uF. I don't know if z5U is used all the way up to the very high values or if there's something even nastier.

They have their place - a lot of decoupling can be done very well with nasty dielectric ceramics but be very wary of using them for many applications.