Don't forget that the tolerance of electrolytic caps is appalling- and that is especially true with older units. Also, caps of value above 470uF are often used as PSU decoupling. The value doesn't matter so much, as long as the circuit "sees" some capacitance. The problem with older capacitors is that they tend to go short-circuit or open-circuit. I prefer open-circuit faults, but they always seem to be s/c :wink:
In some ways, an older analogue meter is more useful when testing caps. Set to x10 or x100 Ohms range, you see the needle dip to 0R and then rise to infinity-Ohms (hopefully!) This is the capacitor charging up and then blocking DC. The speed of the needle swing is proportional to the capacitance C and the internal resistance of the meter R. Whilst it isn't an easily measurable time constant, it's easy enough to connect a brand new 1000uF cap to your meter, check out the time it takes to charge, and then compare the suspect 1000uF cap. Check the voltage polarity of your Ohms range with another voltmeter- often the red lead becomes -Ve under Ohms testing.
If the needle fails to move, the cap is o/c. If the needle doesn't reach infinity-Ohms (where's Keef when you need him- he's got that infinity-symbol shortcut :? ) it's s/c or leaky.
Another way to test is use a 9V battery, 2k2 resistor and an LED. This basic continuity tester works in a similar way to the meter- good cap: LED lights and then slowly dims. Bad cap- LED is always on, bright or dimly depending on level of s/c, or never comes on at all. Again the RC constnat holds true, so you can compare speed of dimming.
Don't try this for caps below 100uF- it all happens quite quickly! The s/c tests still stand though. And don't forget that static meter testing out of circuit doesn't test fully enough, because in the actual circuit it's possible that the capacitor will be biased by a larger voltage than from your test meter- this may be affecting the performance of the cap. So it may test okay with a small AC signal when testing on a DMM, or a small DC voltage doing the needle-swing test, but once biased by a 300V HT, the dieletric may say goodbye!
Replace any suspect devices- they're cheap enough nowadays to prevent major headaches.
Mark