I was wondering if anyone could advise me why a valve radio reduces in volume after running for a few hours. Could it be a faulty valve?

I have replaced all paper caps and electrolytics in the power supply section and radio works great apart from the volume.

Sounds like a temperature sensitivity somewhere. Valves could be checked by substitution, other components by selective cooling.

Valves could also be checked by removing individual valves alternately, leaving the rest in place and under power so they stay hot. Allow the removed valve to cool down, then put it back in and see if the volume comes back up.

Thanks for that I will give it a try and might start with the rectifier valve will put in some diodes and see if any difference.

I would be very careful about swapping the rectifier for a pair of diodes, as the drop across a silicon diode is much lower than a valve rectifier and the additional B+ voltage will stress every other component.
If you suspect the rectifier, just put a meter across the B+ line to ground and see if it changes with time and/or the drop in volume.
Harold

What model radio is it?

Replacing a rectifier valve with diodes requires some caution. For one thing, semiconductors place HT on the other valves immediately upon power up, before their heaters have warmed up and before the valves have begun conducting, whereas the rectifier valve applies the HT gradually, and this can cause issues in some circuits.

If you suspect the rectifier, I would suggest measuring the HT when the symptom occurs.

It would also help to determine if the attenuation is in the RF or AF section of the set. Injecting a signal into the AF section via the volume pot's wiper should let you know that.

To determine if it's a component overheating issue, try briefly squirting components with non-flammable aerosol freezer, such as this:

http://www.jaycar.com.au/productView.asp?ID=NA1000&form=CAT2&SUBCATID=961#1.

Other things to try include cleaning valve pins and sockets with Deoxit and checking for bad solder joints.

Replacing the capacitors is a good start, suggest replacing all of the resistors as well.

Aged carbon resistors are notorious for drifting high in value which can cause all sorts of problems.

One of the big failures I see time and again, is in methodology. If you have a set that is old and has been sitting ages, the electrolytic's will be cactus ... all of them... not just the filters.

If you wished to save time and stress, as you changed out the capacitors, you should have changed the resistors. In the Philips 2462 on the bench, there are at least 8 resistors in the tray of components removed as I changed the caps.

Whist it is hard to get an accurate read on some resistors, if they are high "in circuit" they will be out of circuit, so renew.

A second happening is that like the resistors, if a Mica cap ends up being un-soldered at one end: It will get HV tested for leakage.

Monitor B+ a resistor (possibly cathode may be getting hot? Voltage moving low is an overload, going high loss of conduction.

There would be a lot less assumption if we knew what we were speculating on?

Marc

You all make good points now that I think about it could be the resistors as they warm up, all the caps have been changed and there are only 2 electros in the set in the HV power supply section its a Radiola 429 and I have the circuit.

Thank you for your help I will have another look.

Don't overlook the volume potentiometer. In many valve radios these are positioned close to the main AC transformer and the rectifier valve, and they pick up heat from these as the set warms up in service.

I had one where one or more of the internal contacts was slightly loose and moved slightly as it heated. I didn't discover that until I had replaced every other suspect component (many of which were not far off spec). The carbon surface that contacts the moving brush can become worn as well, making the pot susceptible to uneven resistance that can show up as volume drop as the metal parts move slightly when warming up.

Maven

Thanks guys I found the problem it was the tone pot accidentally bumped it when the volume was low and volume came up louder. but one thing I have to say about using diodes to replace rectifier valves, I keep hearing people say how bad it is, but why are so many old sets using them, one I am working on is a healing golden voice came out of the factory with one rectifier diode for the high voltage and many other circuits of old radio sets as well I am currently running a little nipper set with 2 diodes and a 300 ohm resistor in series as described in silicon chip mag a while back. works fine till I get 6x4 replacement.

one thing I have to say about using diodes to replace rectifier valves, I keep hearing people say how bad it is

Conversion to solid state rectification from valve rectification is not "bad" per se; it just needs to be done intelligently.

The main issue with a simple swap from a valve to solid state diodes is that diodes only have a voltage drop of 0.6 volts. They are thus feeding a higher voltage to the radio's circuits. They also switch faster than valves so heavier filter condensers may be required to absorb the extra mains hum. You may also need a voltage-dropping resistor.

I've always thought that restored radios should emulate manufactured specifications as much as possible - they are more valuable in an un-modified state and when operated within the manufacturer's parameters they are also more reliable.

If a radio was designed to use a diode instead of a valve, the designer would have taken the points mentioned into account. An example is the Kriesler 11-99, where they used a valve rectifier in early versions and a diode in later ones, with suitable changes or ensuring that nothing is stressed.
The resistor in series is one way of doing this.
My earlier comments only refer to a substitution without any other changes, sorry if this wasn't clear.
Harold

The 11-99 is an interesting case. I've never seen the circuit for a diode set, even the circuit in the base of a diode set shows a 6V4. I've restored a few of these, both 6V4 and diode and apart from the diode I haven't spotted any difference in component values, they all match the circuit. The voltages on a diode set are significantly higher than the specs indicate but it seems Kriesler made no circuit modifications between the 6V4 and diode set. I've always been a bit uncomfortable about the voltages but have assumed the designers knew what they were doing and stayed within the upper limits of what the valves can handle.

It comes down to the rating of the caps & efficiency of the 6V4 compared to the silicon diode. I am not familiar with the circuit but if they went to half wave that would produce less voltage and they may have changed the transformer?

The electrolytics were normally 525 surge volts and they may not have changed them as that was a wide margin to allow for valves like 80.

Need a bit more info to spot the difference.

Another thing with valve sets... The Pentagrids tend to drift high and then settle once they stabilise.

Is it drift? Reset dial pointer once it warms up to confirm (by the tuning knob only)