Higher voltage is not more efficient. However, you are safe to use a higher voltage rating. Safer, actually.

As to the MFD ratings, you should stay pretty close to the old one.

makes no difference as long as the voltage is not under the
rated specs. but the MFD had better be almost dead on.

I have a follow up question. What are the most causes of a dead capacitor? In other words, why can a capacitor be dead?

Without getting into detailed specifics (technical mumbo-jumbo and throwing a lot of terminology) of how a capacitor is constructed, since start and run capacitors are constructed differently,I'll answer your question as directly as possible. If you'd like more detailed information, I will be glad to go into detail here or, you are welcome to email me.
The capacitor momentarily stores a quantity of electrons causing one side to have a more positive side and the other side to have a more negative side, similar to a battery. This is done by the way the capacitor is constructed and the materials used. Unlike a battery, a capacitor in an alternating current circuit does not constantly maintain the same polarity. You are probably familiar with the sine wave associated alternating current. The first half of the wave, the terminals on the capacitor assume a positive and negative charge. On the second half of the wave, the charges are reversed. So in a 60 cycle circuit, this is happening 60 times. This eventually, over time, takes it's toll on the internal materials used to make the capacitor. Starting surges also take their toll on the longevity of a capacitor. The more the unit goes through a start cycle, the more taxing it is on the capacitor since there is some minute arching inside the capacitor. Some run capacitors have an internal fuse in them and once that blows, the capacitor has to be replaced since it is non-serviceable. A charged capacitor can hold it's charge for a long time so handling them should be done with a knowledge of how to discharge them. Some capacitors have a "bleed" resistor connected between the terminals to help to discharge a capacitor. From experience..you still want to go through the discharge process to make sure the capacitor is fully discharged!!!! =:0 External events also have an impact, basically rust, corrosion or vibrations thinning or making a hole in the capacitor causing it to leak. Older capacitors, in my experience and opinion, are much hardier than newer capacitors. The problem with them, they were filled with liquid PCB. That would leak out and since it is labeled as a carcinogen, it is considered unsafe to humans and the environment. That material is also used in large transformers. I know some guys who have frequently stood almost waist high in the stuff and never had any issues. I know others who have suffered very serious side effects.
As mentioned, this is a very high level explanation and I will surely give much more detail to you or anyone else who would like to know more. Just remember, a simple or "quick" question doesn't necessarily mean a simple and short answer.

A capacitor is composed of two plates separated a piece of insulation material. The thinner the material, the higher the capacitance value created between the two plates. The trick is to make the insulator as thin as possible in order to make the capacitor small. The problem is the insulator must withstand the voltage applied across the plates.

Over time the insulation begins to break down due the electic field caused by the voltage potential across the plates. When the insulation breaks down, a gas is released. You can spot a bad capacitor if the can looks swollen.

A capcitor doesn't "die". What happens is the capacitance changes as the insulation begins to break down. If the plates touch each other, the capacitor becomes a short. This will not allow the compressor motor to start or run.

Dual capacitors are two capacitors inside one can. They share one common terminal.

Using a capacitor with a high voltage rating means the insulator has been reated to withstand a high voltage. Using it at a lower rated voltage is not a problem. In theory the capacitor should have a longer life at a lower operating voltage.

The PCB filled capacitors have 2 aluminum electrodes suspended into the liquid PCB, the length and thickness determines the rating. The PCB allows electrons to flow from one electrode to another and aluminum chloride gas would immediately form around the electrodes acting as an electrical insulator, stopping electron flow through the liquid causing the capacitor electrodes to maintain one positive side and one negative side. The negative side would flow (discharge)to the positive side through a closed circuit such as a fan or compressor. If it is a dual capacitor, a third electrode is suspended, sized according to the rating needed and a common terminal is used for both sides. Momentary arching and the constant chemical changes eventually break down those capacitors. Many capacitors also have a marked terminal and attention should be paid as to where the marked terminal is connected. This is usually done with a red paint or ink dot. Some capacitors, like those from GE, the writing that is stamped into the casing on one side, identifies that as the marked terminal. If not connected according to the marked terminal, the capacitor would still work. Over time, the electrodes would deteriorate chemically and electrically or the capacitor would leak. These type have a tendency to last much longer than the newer designs. The newer designs are made similar to electrolytics' but not exactly and, the chemical makeup has different properties so it can stay in a closed circuit indefinitely without consequences.

A start capacitor or, also called an electrolytic capacitor is constructed a bit differently. To visualize their make up, think of a sheet of waxed paper then a sheet of tin and another layer of waxed paper, another sheet of tin and another sheet of waxed paper. If you were to roll that up so it will fit into the round black casing, that is basically what is inside, with the terminals affixed to the tin. Of course there are better things than waxed paper to insulate them but that gives you an idea of their make up. As mike home explained, the thickness of the tin and the length of the tin determine the rating. That is why you will see larger MFD and voltage rated capacitors of a larger casing size.
A run capacitor can stay in the circuit indefinitely and therefore is constructed differently than the electrolytic (start) capacitor. The start capacitor can only stay in the circuit a very short time, around 7 seconds, or it will blow. The job of the start relay is to control that. There is a relief on the start capacitor which will weaken and the black goo will come out. Otherwise the capacitor would explode from the pressures. The reason for and the effect of their failure is exactly as mike home describes.

Thanks for all your explanations. Can you give some tips to select a capacitor such as brand name, reliability, efficiency etc. or all the capacitors are the same as long as the specifications meet the original one?

To answer your question previously...Heat.

Packard / Aerovox / Genteq - a few of the most popular.

Most Manufacturers , Carrier , Trane , Lennox feature a Genteq capacitor in their systems.

Those cutting costs , Rheem/Ruud/Goodman sacrifice quality with price.

"causing the capacitor electrodes to maintain one positive side and one negative side. "

Not very useful when the voltage is reversing 120 times a second.

The capacitors used for run and start are normally NOT polarized capacitors.

PCB is an insulator.
It does not play any part in current flow, and is designed to PREVENT current flow.

The main pirpuse in capacitors was to provide additional dielectric insulation.

Many of the PCB capacitors are actually often paper capacitors.
Two layers of metal foil separated by a paper insulator.
To improve the dielectric performance the paper was saturated with PCBs.

PCBs are also good heat conductors, so they can help remove heat from the core of the capacitor (caused by a thing called the 'loss tangent' of the dielectric.

In single phase induction motors capacitors can be used for just starting, or both starting and running.

The capacitor produces a phase shift in the AC waveform that allows single phase motors to start, or in the case of a run capacitor increases their ability to deliver power when there is not enough rotational inertia available.

Since rotating mas takes power, there can be advantages to using a run winding and capacitor over more rotating mass.
It all depends on the losses for each type, and the extra expense of adding run windings and capacitors.

Three phase moors do not need any capacitors to start or run.
the 120 degrees between the phases is more than enough shift to allow starting (the most shift a capacitor can produce is only 90 degrees, and you never get that much in a motor application).

I believe whay brickeyee is saying is the run and start capacitors cannot be the electrolytic type. This is becaue the potential on the positive terminal must always be higher (more positive) than the negative terminal. This is not possible in an AC (alternating current) circuit unless you add a DC bias. This is not practical in a condenser application.

I have always been told an electolytic capacitor will explode if the polarity is not observed.

Electrolytic capacitors are typically used in DC power supplies to filter the AC ripple. This type of capacitor yields the highest amount of capacitance in the smallest volume.

Look, no need to take anyone's word about capacitors here. Visit the site I've included below. You will read how the various capacitors are made, where they're used, how they function and what causes them to die. Many here have asked various questions concerning properly sized and rated capacitors through the years. The site below does an excellent job of explaining that.

http://en.wikipedia.org/wiki/Capacitor

There is much more to know about capacitors than just that though. Whole semesters are devoted to them and in a good course, you build them and scientifically monitor every aspect of them, including electron leakage through a dielectric substance, caused by ripple currents and how to make that do work for you, such as in a run capacitor :) But that is typically used by designers of circuits or those who want to know the intricate details and usually not the end users or technicians.

Maryland,

Please don't be offened by my posts. I don't claim to be a capacitor expert.

The capacitor discussion has gone much further than the question by the OP. It is interesting reading, but way beyond what the typical homeowner needs to know.

What I find a little disturbing are HVAC techs who determine a capacitor is getting "weak" and needs to be changed at the cost of hundreds of dollars. Some techs will pull the capacitor out and measure it, others will measure the voltage and current and calculate the capacitance. The general rule is to change the capacitor if the measured capacitance is more than 10% beyond the nominal value. I don't think the capacitance for the start and run capacitors need a 10% tolerance. If they did, then in my opinion this is a poorly designed circuit.

Here is a good place to get some basic informaiton on electrolytic capacitors.

http://en.wikipedia.org/wiki/Electrolytic_capacitor

"The polarity thus changes 120 times per second. I'm confused about why you say this isn't very useful since they are used for both starting and running..quite useful in everyday applications."

If you put a polarized capacitor in an AC circuit half the time it is going to be backwards.

They usually fail rather spectacularly.

Paper capacitors loaded with PCBs are not polarized, so they do not have this problem.

You can put two polarized capacitors in series to make a non-polarized capacitor (both negative ends together).

The problem is that it cuts the value in half, and if they are not matched tightly one side will still see some reverse polarity.

'Non-polarized' electrolytic caps are made this way.
They can be matched well enough to work, but at a large volume penalty in most cases (twice as big for half the capacitance).

Just because it comes in a metal can does NOT mean it is polarized or electrolytic (in any of the available flavors).

Polarized electrolytic capacitors (aluminum or tantalum) are manufactured by applying a reverse voltage with current limiting to 'grow' the dielectric (an oxide layer).

If you apply power to them backwards again it eats the dielectric layer and since current limiting is unlikely to be present the cap quickly overheats and explodes when the current flow becomes excessive.

mike home
by no means am I offended and thank you for that concern. Yes, the discussion did get a little off topic, that's sometimes the nature of the beast here. I am in your corner and also get disturbed when some of the posters here relate to their tech claiming a weak capacitor that needs to be changed out. Many times I bypass those posts because it does lend itself to ongoing critiquing. There's a lot of information about what a capacitor does, how it is constructed and so on. But you'd be very hard pressed to find much on the web about the exact science of what is happening but more so, a high level overview with a few technical aspects thrown in.

The start capacitor is an electrolytic capacitor, period. However, an electrolytic capacitor designed specifically for DC circuits does have design changes as you mention. The start capacitor is considered expendable and it has a known life span, similar to an incandescent light bulb having so many energized times before it is expected to expire. It's encased in a Bakelite type plastic shell with safety designs to minimize it exploding in an attempt to keep them inexpensive. The run capacitor by design should last indefinitely however the materials used obviously give it a shorter life cycle, they eventually go but the anticipated life span is not projected. Staying with the PCB filled capacitor, the PCB does act as an insulator but not totally because of ripple currents expected in an AC circuit by nature, there is anticipated leakage through the substance. That leakage is known to happen and utilized in the design of the capacitor. It actually is what initiates the aluminum chloride gas around the electrodes eliminating electron flow for that half cycle. As the capacitor discharges and again builds for the second half of the wave commences, the process repeats itself with opposite polarities for that split second. You will be hard pressed to find very detailed instruction on that aspect in most technical or application type instruction, such as HVAC. You'll get a brief description of how they are constructed, their characteristics and how they work then more time on their function and troubleshooting. I did an extensive search of the web in hopes of finding some of the detailed design information about capacitors, as would be found in a textbook on them. I basically only saw the same high level descriptions as have been given here but nothing real in depth. It may have been a mistake to introduce a more drilled down version but I hope someone got something from it.

"That leakage is known to happen and utilized in the design of the capacitor. It actually is what initiates the aluminum chloride gas around the electrodes eliminating electron flow for that half cycle."

just plain old wrong.

Leakage in ANY capacitor is a DEFECT.
It directly decreases the effectiveness of the capacitor and lead to heating of the capacitor.

Leakage current is NOT what a capacitor uses to function or appear to conduct current.

You do not need (or really want) leakage in any capacitor.

PCB capacitors have not been made for MANY years.
They are far more of a historical artifact.

I respect information you offer and consider you knowledgeable, even though it is often critiquing others posts instead of originality but in this case, leakage in the examples I gave is NOT plain wrong and if you believe that, you need to go wherever you need to go, to find out more about it. You are correct as far as an electrolytic capacitor, you don't WANT leakage. The problem is, you HAVE leakage and after so many starts, the capacitor dies, it is predictable. In the example of the run capacitor I gave, that leakage is put to work to enhance the capacitor's function. And as I explained and the time of expiration is not predictable.

You're right, they haven't been made in a FEW years. But there are many millions of units in use and air handlers and a myriad of other equipment that have them. There are still many of them being used as replacements and it is perfectly legal to do so. It is a common complaint here and on other sites all over, the casing is swollen and a substantial amount of liquid leaking out. The example was much easier to explain, to give a novice an idea of what goes on inside the capacitor and why they go up.

Newer run capacitors are made more along the lines of an electrolytic capacitor but the dielectric is not PCB. Their life span is predictable albeit, theoretically longer than a start capacitor. I've seen many of the earlier ones die long before the start capacitor but they are improving them.