Our air conditioner is short-cycling (cooling cycle is very short, so I think the compressor is quitting early. Fan works great). After eliminating everything else that may be wrong, I decided to test the capacitors.
There are 2 huge capacitors. I think one is for the fan, and one is for the compressor. Not sure which is for which.
Capacitor A is smaller, with 2 terminals. It's labelled 7.5MFD, 370VAC, 50/60Hz. I disconnected the wires from one of the terminals and hooked up my multimeter with capacitance testing, and I got the reading 625 nF.
Capacitor B is larger, says "Motor Run Capacitor" on it. 15MFD plus-or-negative 5%, 370VAC, 50/60Hz. My multimeter reads 4.93uF.
Both capacitors are labelled CBB65, and both are in great condition (no bulging, rust, or burn marks. They actually look brand new).

Both of these readings are WTF? I tried hooking up my multimeter to a small 330uF electrolytic and it read 318uF. So what's going on? Do I need to use a different kind of capacitor tester? Remember the fan is fine, the air conditioner does turn on and starts cooling, but the cycle is short. Both capacitors look fine (like brand new).

 

Have you got a capacitor nearer to the 7.5uF or 15uF just in case your meter doesn't like smaller caps.
Otherwise it looks like both caps are way past their best.

 

What do you mean by 'short'? if the motor does not overheat it could be due to something in the control loop.
@#12 may confirm if these motors generally have a bimetal temp sensor.
Max.

 

Have you got a capacitor nearer to the 7.5uF or 15uF just in case your meter doesn't like smaller caps.

I wanted to try that, but the closest I have are 2uF and 100uF capacitors. I'll test them when I get a chance.

What do you mean by 'short'? if the motor does not overheat it could be due to something in the control loop.

An air conditioner, when "on" has 2 cycles. One is the cooling cycle (fan and compressor), and the non-cooling cycle (fan only). Usually, it's something like 10 minutes cooling, 10 minutes non-cooling (and repeat). My AC is switching to non-cooling about 1 minute after cooling starts (hence "short-cycle"). If anything, it's the compressor that would be overheating (not the fan). But I'm suspecting one of the capacitors are not operating properly.

 

Have you got a capacitor nearer to the 7.5uF or 15uF just in case your meter doesn't like smaller caps.

I just tested a 2.2uF plus-or-negative 20%, 275V capacitor with the multimeter. It read 1.867uF (reading looks good).

 

I just tested a 2.2uF plus-or-negative 20%, 275V capacitor with the multimeter. It read 1.867uF (reading looks good).

OK, I think the meter is in the clear and both capacitors are due for replacement.

 

You can't test those caps with a meter. You calculate the proper amount of current if they are connected to the power line and measure the current. The 7.5 uf should have 424.4 ohms of Xc at 50 Hz. Connect it directly across the 250 volt line and expect 0.59 amps. 15 uf will obviously expect twice as much current.

HOWEVER, short cycling is usually caused by cold air bouncing back at the thermostat bulb. Another option is the compressor having the run winding wire burnt off the terminal. I assume you haven't traced the wiring or you'd know which cap is for the fan. See if the compressor is stuck in, "start" (too much current) because the run winding has no current. That could also be caused by an open run winding in the compressor motor. The internal overheat bimetal is in the common leg so it stops both windings. If it's cycling on the bimetal, it will be so hot it practically radiates. Again, it's all down to current measurements if you don't see a cold air bounce hitting the thermostat.

 

You can't test those caps with a meter.

What is it about them which confuses a capacitance meter?

 

They are bipolar, oil filled capacitors. My best guess is that the cap meter doesn't polarize the plates enough to get them to hug each other due to a deficient difference of charge on adjacent plates.

 

You calculate the proper amount of current if they are connected to the power line and measure the current. The 7.5 uf should have 424.4 ohms of Xc at 50 Hz. Connect it directly across the 250 volt line and expect 0.59 amps.

Errrr... do I measure series or parallel? Do you (or anyone) know of a video or a guide I could use? What you describe I'm not sure how to do it.

HOWEVER, short cycling is usually caused by cold air bouncing back at the thermostat bulb.

Yeah, I can't find that *#@%@ bulb anywhere. Should be near/on/in front of the evaporator coils. Don't see it anywhere. Thinking maybe the unit uses some sort of timer/oscillator to cycle on/off instead of a thermistor (it's an ancient AC unit, from the 70's or 80's).

 

A thermostat bulb is a copper tube with a larger end. Tiny tube going to about 1/4 inch diameter bulb about 2 inches long. There is no such thing as using a timer as a thermostat. Timers can't measure temperature.

How do you measure current through a capacitor? Plug the capacitor into the wall outlet and use an amp meter. Either clamp-on or DVM. Then again, if you don't know whether one single part plugged into a wall outlet is wired in series or parallel, you shouldn't be doing this.

 

Yeah, I can't find that *#@%@ bulb anywhere.

The problem isn't the bulb, it's the air bouncing back at the intake from a short distance, like a few feet.

 

Hi,

I have to second the idea that the cold air may be getting back to the thermostat too much and therefore making it think the whole room is cool.

Alternately, bad thermostat. This is common really.

Also, check the filter. If the filter is dirty the first thing above will happen because the cold air can not 'get out' of the unit.

1. Check filter.
2. Check for cold air recirculation.
3. Check thermostat.

If the thermostat is bad you can tell my taking it out of the unit and connecting with long wires that are not near the AC unit, for one example. That way it can not detect any cold air. I suppose you could stick insulation in there somewhere too to keep cold air from getting to it.

 

Or may be clean the darn thing. Just a suggestion

 

They are bipolar, oil filled capacitors. My best guess is that the cap meter doesn't polarize the plates enough to get them to hug each other due to a deficient difference of charge on adjacent plates.

I've been thinking about this. On WWW I can find plenty of sites of people successfully measuring the value of these capacitors with a multimeter. I wonder if it is to do with the frequency that the multimeter uses to make the measurement. These capacitors, high value for non-electrolytics, and high voltage necessitating thick dielectric, have many turns and so high inductance. This would pose no problem at 50/60Hz but would at higher frequencies.

 

Quick update: I ordered and received 2 replacement capacitors. Using the same multimeter from before, these both measure fine (very close to 15 uF and 7.5 uF). So the previous capacitors are toast. I installed them. One of the caps is definitely for the fan motor; it started up instantly. But the AC is still short-cycling!

So, I have:
1) cleaned the evaporator and condenser coils.
2) checked for obstructions (inside the AC, and also made sure the air outlet and vent outlet are clear).
3) Replaced both capacitors.
4) Replaced the air filter.

It may be a thermistor issue, but I can't find it. Then again, it may be something else altogether. Busted thermostat? Pressure issues?
Time to bring in the repair guy.

 

You can easily check to see if it is the motor end, e.g. embeded thermal bi-metal detector etc, by checking to see if the power is still present at the motor.

 

You can easily check to see if it is the motor end, e.g. embeded thermal bi-metal detector etc, by checking to see if the power is still present at the motor.

Not sure what you mean. The motor (fan) always runs. No issues with the fan. Do you mean to check the compressor?

 

If the compressor motor is prematurely stopping see if there is power at the terminals, if so it points to internal/embeded bi-metal switch.
Either faulty or detecting over temp.
Max.

 

You can easily check to see if it is the motor end, e.g. embeded thermal bi-metal detector etc, by checking to see if the power is still present at the motor.

Not sure what you mean. The motor (fan) always runs. No issues with the fan. Do you mean to check the compressor?

@MaxHeadRoom That's easy enough to say from here, but jellytot can't tell whether you're telling him to check the fan motor to see why the compressor stops. I'm just going to have to agree:

Time to bring in the repair guy.