Hopefully, you will know:
• Capacitance and tolerance that you need
• Rated voltage of the part
• Actual working voltage applied to the device
• Physical size
• Any peculiar temperature or humidity requirements in your equipment
• If it is subject to a high current discharge at any time.
There are very few standards in the capacitor industry, other than in true MIL parts, and interchangeability can be a problem. Particularly with relation to product size. Size, capacitance and voltage rating will give an idea of how far the original manufacturer was "pushing" the voltage rating of the dielectric. That is how many volts per mil of film thickness.
So, if size is settled, let's think about voltage ratings, test voltages and operating temperature. You may find out that, with (for example) a 10 KVDC, 0.1 mfd capacitor, Company A will rate its part for operation at 10 KV at a maximum temperature of 45°C (113°F) and will test the part at only 110% of rated voltage. 20 KV! And other manufacturers will rate theirs somewhere in between these parameters.
Test Voltage and Safety Margins
In the example above, the second capacitor will probably be bigger than the first, as it will be more costly, in the short term, but is likely to last much longer, in actual service. I should note that most power generating companies, such as Commonwealth Edison, are permitted to have a 10% line voltage variation. Consequently, a capacitor that is being used at its rated voltage could, under certain circumstances, be operated at the test voltage, when the test voltage is only 10% above its’ rating! This is not a good practice, if you want long life.
High Voltage AC Capacitors
High voltage AC capacitors are somewhat different in design from DC capacitors and they are rarely interchangeable. I wish that I had a dollar for every time that someone has blown up a DC capacitor and then said "I used the thing at 7KV RMS at 20 KHz. I took the peak AC (7000 x 1.4 = 9800 V) and figured that a 10 KV DC rating would still give me a little safety margin". This is very wrong for 2 reasons. It is a tossup to decide if corona (partial discharge producing substantial ionisation) would literally eat the dielectric or if the heat build up in the capacitor would cook everything first.
Without getting too technical, the inherent nature of films limits the corona inception voltage in a single capacitor section. For example, the different dielectric combinations of paper and Mylar for 3 KVDC, 4 KVDC, 5 KVDC or 6 KVDC all start to show corona at about 1500V, RMS, 60 Hz with lower voltages at higher frequencies. Other films or combinations of film and paper have different corona thresholds; but always ask if a part can be used on AC at the voltage and frequency that you require.
The voltage and frequency have a very significant impact on the internal heat loss of a capacitor, due to its "dissipation factor" (Df) (sometimes called "power factor"). The Df of a capacitor may be called the sum of all of the errors in the manufacturing process plus the Df inherent in all of the dielectrics.
For example, a Mylar film and paper capacitor has a Df of up to 0.5%; polypropylene film, perhaps 0.02%. This means that, respectively, capacitors made with these films would be 99.5% and 99.98% efficient. Those fractions that remain in the capacitor will turn into heat. While anything that is 99+% efficient would sound darned good, these 2 items would have very different AC results.
If we had a 1 mfd, capacitor operating at 5000 volts, at 60 Hz or at 1 KHz, with paper & Mylar, we would have a capacitor that generated 39°C of internal heat at 60 Hz (assuming a 10 inch cube with no forced air cooling); but at 1 KHz, the same capacitor would dissipate 785 watts and would generate a theoretical heat rise of 654°C! Of course, the capacitor would fail long, long before it reached the theoretical temperature! But on DC voltage (no Hz), Df is of less concern.
Given the same parameters, with polypropylene film, it would warm up about 2°C, at 60 Hz; at 1 KHz, it would dissipate only 47 watts and would warm up about 40°C above the room ambient temperature. Some high frequency AC capacitors must be water cooled, to remove the heat that even forced air cannot deal with.
When it comes to replacing the "nuts & bolts" items in a piece of equipment - resistors, diodes, chokes, fan blowers, etc - all are a lot easier to replace than High Voltage Capacitors. The hunt for a close or exact replacement part can lead to a frustrating maze of unknown specs, minimum orders and discontinued items.
When in doubt, call a manufacturer or a well established distributor for a replacement or substitute part. You may need to avail yourself of their library of old, old manufacturers' specifications. Some manufacturers may not want to sell direct to you, or they may not have any cross-reference data but at least you will get an idea of where to start to look for a replacement part.