How Do You Test Capacitors?

The instructions that follow are intended for the measurement of high-voltage power supply capacitors and interstage coupling capacitors in tube-based radios and amplifiers. Tiny electrolytic capacitors that populate circuit boards in today's modern solid state (no tubes) amplifiers, CD players, satellite receivers, etc. are typically not worth measuring or testing. If your modern audio equipment has recently stopped working and is more than 10-15 years old, your first order of business is simply to REPLACE ALL OF THE ELECTROLYTIC CAPACITORS in the power supply (first step) and potentially in the rest of the circuit (second step, after the first step didn't solve your problem). See here for additional details.

OK, so what is the big deal here? If you've poked around my web site, you might have seen my write up on how to repair some typical problems with modern electronic devices. On this page, I recommend sweeping replacement of capacitors without much effort to test them. The rationale here is simple: New caps for modern gear fit standard form factors for size, are inexpensive, readily available, and generally easy to replace. With this winning combination, there is nearly nothing to lose from simply replacing all of the caps on the circuit board, so long as you work carefully and don't make silly mistakes or damage something in the process. When you replace capacitors on a circuit board, it is often difficult to tell that you've done anything at all to the board.

But things are different with tube-based audio gear. If you are restoring an antique radio or amplifier, there is often a desire to keep things looking as original as possible. The older your gear is, the less "standard" parts become. While new "equivalent" parts most likely exist, they certainly don't look the same or they may not fit into the physical space you have because they are a different size or shape. Thus, simple and wholesale replacement may not be a viable option with antique electronics. Thus, it is useful to have a better understanding of what shape the original parts are in. Similarly, if you are building a new tube amplifier and using NOS (New Old Stock) capacitors in your power supply or as interstage coupling caps, you need to make sure these caps function properly. An interstage coupling cap that leaks will spell disaster for your (likely expensive) output tubes!

Capacitors feature a number of attributes that can be measured - nearly all of which require you to remove at least one leg of the capacitor from the circuit in order to properly measure it. Some meters will allow you to test caps "in circuit" but it is typically best to carefully unsolder at least one leg and measure a cap outside of the influence of other components it is connected to. Sometimes, removing the tubes sufficiently isolates the capacitors you are interested in testing. Sometimes not, so you'll have to check the schematic carefully. Alternatively, follow the actual wires that are connected to the cap in question and see where they lead to be sure about this. Each radio/amp will be different.

1. Measuring Capacitance:

This is a first step in checking capacitors and nearly any $30 digital multi meter (DMM) can do this. Typically, a capacitor that is more than 20% off of its specified value (either high or low) has had a hard life and should be replaced.

2. Measuring Dissipation Factor (DF):

Some DMMs will provide dissipation information while they are measuring capacitance, but this is not the norm for a $30 DMM - you are typically talking about a meter that costs closer to $100 at this point. Dissipation Factor (as indicated by "D" or "DF" on your meter) measures the loss rate of energy stored in the capacitor and is often conveyed as a percentage. DF readings of less than 0.5% are typical, but can be as high as 1% to 2% in some capacitors. Lower is better when it comes to measuring dissipation. This measure is related to ESR (below). Here is a chart that I pilfered from Conrad's web site that provides some analysis of Dissipation Factor:

"What value do you use as a cutoff to determine that a cap is bad? If you have a datasheet for the part, it should give some limits. If you can get a datasheet for a similar class of part it should serve as a useful estimate. Hopefully it will specify a maximum dissipation factor, usually measured at 120 Hz. Here's the chart for a general purpose Rubycon YK series general purpose radial that's typical of most general purpose caps:

There is a note at the bottom of the table: 'When nominal capacitance is over 1000 uF, tan θ shall be added 0.02 to the listed value with increase of every 1000 uF.' So let's say you have a 4700 uF 50 volt cap. The base dissipation factor for 50v is 0.12 from the chart above and because it's larger than 1000 uF there's an adder of 0.08, giving you 0.20 (I rounded the value to 5000 uF). Now, the end-of-life dissipation factor is 2X, so the cap can be considered bad if the dissipation factor measures over 0.40 @ 120 Hz."

When you start getting into more expensive meters, the meter should specify at what frequency specific readings are made. Better meters will allow you to change the frequency at which readings are made. My meter allows testing at 120Hz (2x mains frequency of 60Hz) and 1kHz (typical frequency for audio-related purposes). Even better meters will provide a wider variety of frequencies for testing.

3. Measuring Equivelant Series Resistance (ESR):

This measure is related to DF (above), but is a separate calcuation performed by meters and is also not typical for a $30 DMM. Some meters will display either DF or ESR, or both. An ideal capacitor is just that - a capacitor that displays no other electrical properties. In the real world, however, ideal capacitors don't exist - a capacitor always exhibits some resistance in addition to its capacitance. Determining ESR is not as simple as measuring the resistance across a capacitor with the resistance function of your DMM, though. ESR is a much more complex entity that depends on a number of other factors.

ESR typically increases with capacitance and with the frequency of the signal that passes through the capacitor being tested. This is why simply measuring the resistance across a cap with the typical DMM (that makes measurements with a single and unspecified frequency) will not provide useful information for you. ESR also increases over time (normal aging - with or without use), as a result of high temperature (tube gear gets hot), and with large ripple currents (high-voltage or high-current power supplies commonly found in Class A amplifiers). When the ESR of a capacitor increases, it causes circuits to misbehave and can lead to the failure of other components downstream in the circuit.

But here is the interesting part: a cap with tragically high ESR (lower numbers are better - see chart below) won't be revealed by simply measuring its capacitance with an inexpensive DMM. Thus, it is possible to have a bad cap whose capacitance measurement on a cheap DMM matches the label on the cap - in this case, you'd never know the cap was bad unless you made further measurements. Darn sneaky little buggers!

The discussion above concerning capacitor testing is rather high-level summary information. If you thirst for a greater understanding of capacitors and the various tests that can be performed on them, I recommend reading Samuel Goldwasser's excellent writeup on capacitors.

4. Measuring Leakage Current:

Here is where things get interesting, especially for high-voltage capacitors that exist in tube electronics. Let's presume you've made each of the three measurements that I described above and each of your caps passed with flying colors. There could still be a tragic and fatal flaw with a capacitor that mandates it be replaced: it leaks. None of the previously described tests will identify this problem. In fact, THIS CRITICAL PROBLEM CANNOT BE IDENTIFIED WITH ANY DMM! Measuring the leakage current of a capacitor means you need to apply the full working DC voltage to the capacitor and measure how much of it "leaks" across to the other side. It is important that this test be conducted at the voltage rating of the cap or the working voltage that the cap will see in the functioning circuit. Many repair people will recount experiences of troubleshooting electronics with caps that measure good at low voltage (5-15v) only to fail (by leaking current) when exposed to 50v, 100v, or more. This is the reason why a 9v battery-operated DMM cannot perform this test.

This test is important for power supply capacitors, but is ABSOLUTELY CRITICAL for signal capacitors, especially for interstage coupling capacitors in tube radios and amplifiers. Remember, capacitors are supposed to allow alternating currents (music signals) to pass through, but are supposed to COMPLETELY BLOCK direct current (DC power supplies, plate voltages, grid voltages, etc). If DC gets through a cap, it can easily lead to burned out resistors (easy repair), blown tubes (potentially expensive repair), burned out filter chokes (sometimes impossible to repair/replace), and burned out power transformers (sometimes impossible to repair/replace). This is another good reason to install a protective mains fuse in your tube gear - it prevents damage to you and your gear when you've got leaky caps and other undiagnosed problems.

You basically have two options for directly measuring leakage current in capacitors. The first is a tube-based capacitor tester. These devices are old and no longer made, so you need to hunt for one on E-Bay, at flea markets, or at Ham/radio fests. Search E-Bay for "Capacitor Checker" or "Capacitor Tester" and you'll likely turn up a number of them. Some are more expensive than others and many will need some form of maintenance (a new "magic eye" tube, new wires, new reference caps, calibration, etc, etc) due to their age. This path represents something of a mixed bag for those with already limited skills in diagnosing and repairing old electronics. This leads us to the second option:

Building Your Own Capacitor Leakage Meter:

When people on the antique radio forums first asked me if I'd checked my vintage radio's power supply caps for leakage, I had no idea what they were asking me to do. I did a few searches that didn't turn up very much at all and my troubleshooting process ground to a halt. One day, though, I stumbled across Max Robinson's incredible web page. One of the articles he's written is A Simple Way to Test Capacitors. This page includes some description, instructions, and a handy schematic for how to build your very own modern capacitor leakage meter. I started with his schematic and made a few modifications to it. First, I did not include the rotary switch and associated power resistors (used for measuring and reforming electrolytic capacitors). Second, I added an internal volt meter to indicate the voltage being applied to the capacitor under test. Finally, I used a momentary push button instead of the original SPST toggle switch. The image below shows the internals of my new capacitor leakage meter.

There are many devices / schematics / videos floating around the Internet that provide a binary "leaky" or "not leaky" indication for capacitors: some show a flickering neon bulb, others will light up a small bulb. I prefer this method where you can vary and control the test voltage and obtain an exact reading of just how much your capacitors leak, if at all. While this device costs more to build than some others do (this one cost me $65), it provides excellent control and precision. I started with Max's schematic, made a few tweaks, and ended with this:

And here is my actual implementation: 

Instead of the 140-0-140 center-tapped transformer that Max recommended, I used an Antek AS-05T320 50VA transformer which has 320V, 300V, and 6.3V secondaries. This change has two advantages: first, my toroid was less expensive than a comparable Hammond center tapped transformer, and second, it provides a higher test voltage making it more useful for 300B tube amps. Many coupling caps used in these modern tube amps need to withstand 600Vdc or so. I wired the transformer to an IE power entry module shown in the upper left of the image. The primary is wired through a 0.5A fuse and a normally open, momentary push button switch. Only two secondary wires are used (the 320v tap); I clipped the bare ends from the remaining secondary wires, covered them in heat shrink tubing, and bundled them together so they'd stay out of the way and not cause problems (lower left corner of image). There is a 1N4007 (1A 1000v) diode (though it is hard to see because it is covered in heat shrink) just before the positive leg of the caps at the top of the image. The two caps in series are each rated 22uF at 450v, resulting in a "single" 11uF cap with a capacity of 900v DC. The bleeder resistor array is made from 4 x 150k 2w resistors arranged in parallel then series configuration for a net of 150k. The maximum output of this power supply with my variac is about 550vDC, so each half of the bleeder resistor array sees a maximum of about 275v. To calculate the maximum power dissipation of the bleeder resistors, square the maximum voltage (275v) and divide by the resistance of each resistor (150k): 275^2/150,000 works out to just about 0.504w maximum dissipation for each resistor. Since each resistor is rated at 2w, we're in good shape!

Not shown in the image above: it is a good idea to connect the centerpoint of the two caps to the center point of the two adjacent bleeder resistors. This voltage divider insures that the two caps share voltage equally - a measure that prevents exceeding the voltage rating of either cap for improved safety and longevity. Without this connection, one of the caps may tend to hog the voltage due to mismatches, etc.

The 10M resistor that comes before the 50uA meter (uppermost meter in image above) is also covered in heat shrink so it is not directly visible. Finally, the 500 volt DC meter (bottom meter) is wired across the bleeder resistor array. The power output is simply a pair of speaker binding posts that were in my parts box. It is critical that the voltage output is isloated from your box if you've used a metal project box! So, for a grand total of about $65 and a little bit of assembly time, I have a brand new capacitor leak tester. To use it, I plug the IEC power cord into my variac so I have continuously variable voltage from 0 to 550vDC (instead of the original fixed 200v and 400v test points). The momentary pushbutton switch is for safety and makes sure that the box is never left unattended in a powered-up state. Once you let go of the push button, the transformer is powered down and the bleeder resistors discharge both the internal caps and the external capacitor under test within a few seconds. While the external cap is discharging, the ammeter will deflect into the negative region and will return to the zero point when the external cap is fully discharged.

Add a few labels to the outside of the box and it's ready to go.

Important Safety Precautions:

Before using your nifty new box to test old caps, BE SURE TO WEAR SAFETY GLASSES! Old caps have been known to fail in spectacular fashion - think 4th of July! Capacitor explosions are very noisy, can be bright, and can damage your hearing or eyesight. Burns may occur. Small parts and pieces may fly away at high speed. You get the idea... In general, caution is ALWAYS warranted in situations that deal with lethal voltages!

Additionally, you may want to go one better than I did - use fully insulated speaker terminals rather than the exposed ones that you see in the image above. Inadvertently brushing your hand against 400+ vDC is no fun at all... 

Using Your New Meter:

Using your new meter is simple. Just connect the cap you want to test to the screw terminals with a pair of insulated alligator clips, plug the meter into your variac (dialed down to zero first), hold the power button down, and slowly bring up the voltage. A fuse rated at 0.5A is sufficient if you always start with the variac set to minium voltage and increase from there. If the variac is set high and you push the button, the inrush of current necessary to charge the caps to high voltage will likely blow the fuse. I ate up a few fuses before I figured this out... A 1A fuse would probably eliminate this behavior, but it's not any big deal. As you increase the voltage across the cap, you'll see the ammeter rise as the cap draws current to charge. When you have reached the desired test voltage and the caps is no longer charing, the ammeter should begin to decline toward zero over time - things should become stable after several RC time constants. One RC time constant is the product of the circuit resistance in ohms (10M in this case) and the size of your capacitor in farads. With small caps (say, 0.1uF) used for interstage coupling in tube amps, one RC time constant equals about 10 seconds, thus you are looking at 20-40 seconds for things to stabalize as the cap charges. For larger tube power supply caps (3-10uF) it may take 1-2 minutes to determine the minimum amount of leakage current. You'll get the feel for it once you test a few caps.

When the test is over, release the push button. This removes power from the transformer and the capacitor you are testing (as well as the two internal caps) will begin to discharge through the bleeder resistors in your box. The ammeter will deflect into the negative region as the caps discharge. This, too, will take several RC time constants, which could be a few minutes depending on the size of the cap you are testing. You will know that all of the caps are safely discharged when the ammeter reads zero again. Go ahead and disconnect your external cap now.

Interpreting The Results:

So, what can you learn by using your new capacitor leakage meter? Here is Max's response when I asked him: "How much leakage is too much?"

The paper in oil capacitors that were used in radio power supplies in the 1930s can tolerate some leakage - maybe as much as 40 microamps. For any capacitor that is used as a coupling capacitor (for example: interstage coupling of a more modern tube amplifier), if you can measure ANY LEAKAGE CURRENT AT ALL with this meter, it is TOO MUCH. Discard the cap and look elsewhere for a replacement.

So, there you have it. You, too, can build and use your very own capacitor leakage current meter.