Measuring an aluminum electrolytic capacitor's ESR

19 Jun 2019

One of my hobbies is collecting vintage computers. Computers in my collection include the Apple IIgs, Atari 400, TI 99/4A, Commodore 64 and Sinclair ZX-81 to name a few of the popular ones. Like all electronics, these systems contain capacitors. Anyone who has collected retro game systems, computers, or audio gear knows to look for caps that have leaked their electrolyte. But, what if you cannot see visual damage, what measurements can you make to verify if an aluminum electrolytic has reached its end-of-life? (Or close to it.) In this post, I show two measurements to consider and show a couple of ways to make them.

Example construction diagram from a 35V 100uF KEMET ESH35V 100uF KEMET ESH.

Why does ESR change?

Aluminum electrolytic capacitors before the 1990s all used a liquid electrolyte in their construction. Over time that liquid dries out. Part of that dry-out is just from evaporation, but that isn't the primary contributing factor. We call it a liquid electrolyte, but it is more like a paste—an acidic paste. Its pH level has the unfortunate side-effect of breaking down the capacitor's dielectric layer. The good news is that as it breaks down, applied voltage causes the layer to re-grow. But how does it grow? Well, remember the dielectric layer is an aluminum oxide. Oxides need oxygen. That oxygen comes from the electrolyte.

Example Aluminum Electrolytic Lifetime Chart (also from KEMET ESH)

So the electrolyte both breaks down and reforms the dielectric layer. This process is the reason why aluminum electrolytic capacitors have a rated "lifetime" associated with them. Eventually, the electrolyte runs out of oxygen to contribute to oxide (re-)growth. At that point, the leakage current and its equivalent series resistance (ESR) becomes very high. After either of those parameters passes a limit, the parts are considered at parametric end-of-life.

By measuring either the leakage current or ESR of a capacitor, we can determine if it has reached its end of life or not. Also, depending on the data history available, you might be able to estimate its remaining life.

It isn't just vintage

Obviously verifying a capacitor's status is essential when repairing vintage electronics. However, that is not the only use case for these measurements. When troubleshooting any electronics device, being able to verify if a capacitor is damaged is helpful. If you're designing a new circuit, you may want to characterize components from different vendors to evaluate how well multiple sources work in your design.

Or, maybe, you are paranoid like me and want to verify components work before using them even when they are new.

Measuring Leakage vs. ESR

You might think to measure the "equivalent series resistance" you set your multimeter to ohms, touch the test leads, and see what it measures. Well, that is not the case. When doing that, you measure the insulation resistance (IR) of the dielectric layer. Remember that a capacitor's dielectric is an insulator. No current should flow through it. However, real capacitors have some leakage.

IR=V_MaxI_Leak =25V14nA =1.79GΩ

Measuring IR involves charging up the capacitor for a few minutes and then measuring the current. Do a little bit of math with Ohm's Law, and you now know the Leakage Current or Insulation resistance for the capacitor. Of course, it depends on the capacitor type, but know that this value is going to be in the Mega-, Giga-, or Tera-ohms.

The resistance component of ESR is dominated by the wires and materials that connect the capacitive element to the outside world.

Simplified ESR Measurement

So then, how do you measure equivalent series resistance? We need to measure the resistance of the component without charging up the capacitive element. Instead of a DC signal, like with measuring insulation resistance, ESR has to be measured with a low-voltage AC signal. Looking at capacitor specifications, you’ll find ESR specified at frequencies like 100 Hz or 100 kHz. As frequency goes up, the ESR of an electrolytic goes down, due to its construction. So for slow bulk decoupling, the 100 Hz number is effectively the ESR at DC. If used with a switching regulator, the higher 100 kHz gives a more suitable estimation of the capacitor's ESR.

Later in the post I show details on how to measure ESR. Before that, let's look at the simpler leakage current measurement.

Measuring Leakage Current

When I can remove a capacitor from a circuit, I tend to use leakage current as a measure of its condition. This simple measurement only requires a power supply, multimeter and some patience. Ideally, the power supply should be able to limit the current. If your supply does not, then a resistor works fine. The multimeter’s purpose is to measure current.

In this video, The Learning Circuit 42: Replacing MLCCs with Polymer Capacitors , I show how to make this measurement.

Here are the steps involved:

Limit your supply to less than 100 mA, 10 mA, or 1 mA. Pick the smallest value that your supply’s limiter can provide.
Set the voltage to the capacitor’s rated voltage. (Another option is to set the voltage to the circuit’s applied voltage.)
Connect the supply to the capacitor, with the multimeter in between to measure the current.
Turn on the supply.
Wait 5 minutes.
Look at the current measurement.

If you measure an older capacitor, like those from vintage electronics, then I highly recommend starting with 25% of the rated voltage. After measuring the leakage at that level, increase by 25% until you reach the full rated voltage. If the capacitor’s dielectric is heavily damaged, even at 100 mA, there is enough energy for a catastrophic fail.

ESH Leakage Current Limits

After 5 minutes, the capacitor is mostly charged. The current draw at this point is the dielectric healing itself. But what is a good value? Most capacitor datasheets specify the limit with some portion of their CV. For example, the ESH from KEMET says the max leakage current is 1-4% of the capacitance multiplied by the voltage. If you do not have a datasheet for a particular capacitor, 5 or 10% of the CV is a conservative guideline.

One thing to watch is what happens to that current over time. As the 5-minute mark approaches and then passes, the current should continue to go down, though at a much lower rate. A sign that the electrolyte may be near end-of-life is that the current stays relatively high. If that occurs, then the next thing you need to check is the ESR.

Measuring ESR with a scope (hard way)

While leakage current uses simple tools, measuring equivalent series resistance (ESR) is a bit trickier. As I mentioned before, you’ll need a small AC signal. You might immediately think of a sine wave when it comes to AC, but really anything that has a changing component, aka dv/dt, works. For example, you can use a pulse from a function generator.

If you have an oscilloscope and function generator, the only other piece you need is a 47 ohm resistor. Ideally, you should use a 50 ohm precision resistor, but I’m not sure how many people have one of those laying around. For this measurement, you build an AC voltage divider where you measure the voltage drop across the capacitor. The voltage divider's R1 is 100 ohms and R2 is the capacitor under test. R1 is 100 ohms because the function generator has an output impedance of 50 ohms and I added a 47 ohm resistor. Using a bit of math, you can determine the ESR. I learned of this method from Geoff Graham.

Measure ESR with Scope and Function Generator

For the AC signal, set the pulse generator to output a 0 to 10 V pulse with an on time of about 1 us. The off-time should be relatively long. You want a short pulse with a slow repetition rate. The idea is you want to quickly apply a voltage, see the instantaneous voltage drop, then let the capacitor discharge.

In my setup, depending on the capacitor, I could not always get my signals to swing from 0 V. So I used cursors to measure the peak-to-peak voltage from the starting level to the ESR voltage drop level. On the right screenshot I have zoomed in on the edge event and circled the peak-to-peak cursor points. This method introduces some inaccuracy, but it is good enough for an ESR estimation. (For slightly better accuracy, you may want to turn on waveform averaging.)

R2 = R1 × 1(V_INV_OUT −1) = 100Ω × 1(10V135mV −1) = 1.35Ω

Oh. Why did I pick 47 ohms? To make the math a little bit easier. The denominator result is 14.19, which when you multiply by 100 ohms comes out to 1.42 ohms. Thing is, because of the 100 ohm and the 10 volts, you just need to see the voltage across R2 and multiply that by 10 to get the ESR. In the end, though, it didn’t matter for me. I used a Math function on my R&S RTM3000 Oscilloscope to multiply the analog channel by a constant value, which did the math for me. That result is the 1.35 Ω value circled on the right screenshot.

The scope method to measure capacitor ESR is a qualitative measurement. It gives you an order of magnitude, but it is not a precise measurement. An LCR (inductor, capacitor, resistor) bridge uses techniques, like a bridge, to measure the ESR, so it is far more accurate. However, to validate if a capacitor is “good” or not, the scope works okay. The real downside is that it can be time-consuming to set up and do.

If you only have a few capacitors to measure, it is probably fine.

Measuring ESR with an ESR meter (easy way)

Alternatively, if you use a meter designed to measure ESR, the measurement process is laughably easy. One such meter is the Atlas ESR70 from PEAK Electronics. The element14 community generously sent me one along with a handful of caps to test with. You can see my review of the ESR70 here.

atlas ESR70 from PEAK electronic design

In the case of the ESR70, a capacitor’s ESR measurement displays first, followed by the capacitance. In my brief experiments, I found the measurement to be repeatable. I also found that the results agreed with the scope result. But the thing is, it took no effort to set up. I attached the cap, press on, and then get a number. Also, I did a quick in-circuit vs. out-of-circuit comparison. The ESR was slightly different but close enough to know the capacitor had not reached its end-of-life yet.

Overall for about $100, this meter is very handy. The scope method from above works fine in a pinch or if you need to make the measurement on occasion. In my case, I am always evaluating vintage electronics before powering them on. With the ESR70, I can quickly check all of the larger capacitors for more than just visual damage.

How does the ESR70 measure ESR then?

So now that I’ve shown you the manual method and a specialized tool, I wondered, what is the ESR70ESR70 doing to measure ESR. So, I hooked it up to my scope. In the screenshot below, I have the scope connected to the meter while measuring an axial aluminum electrolytic capacitor. I setup the scope to capture a single sweep across a couple of seconds so we see could the meter’s behavior.

ESR70's capacitor measurements

When it runs a test, there are two different modes. The first uses several 100 kHz spaced pulses to evaluate the ESR. Since I used such a long sweep, the sample rate reduced significantly. When I zoomed in on the ESR section, I just saw little spikes. I’m not sure if these are actually pulses or some other wave shape. Regardless, its the frequency that I was interested in seeing. Next, it charges up the capacitor to measure the ramp time to determine the capacitance.

One thing I haven’t done yet is comparing this waveform with capacitors of different ESR values, to see how the spikes change. I suspect the measurement is very similar to the scope and function generator method mentioned above.

FYI, I have a video review of the ESR70 coming. Follow the Workbench Wednesdays page to see when it is released.

Which is better Leakage or ESR?

Since there are two measurements to tell the condition of the capacitor, which one is best to use? Remember that Leakage Current and ESR tell different but related stories for a capacitor. The leakage current comes from the dielectric layer breaking down. In an aluminum electrolytic, the ESR indicates the electrolyte's remaining life.

If you can remove the capacitor from the circuit, you should measure both to evaluate the capacitor fully. But if you cannot remove the capacitor, then you are limited to only attempting to measure ESR. Leakage cannot be measured in-circuit because it involves applying voltage to the capacitor. That voltage is going to end up powering-up the rest of the circuit.

The method used by an in-circuit tester like the ESR70 has a good chance at making the ESR measurement in-circuit. Other elements may affect its reading, but it should still at least give you a good sense for whether or not you found your problem.

Conclusion

Measuring a capacitor’s ESR is not trivial, but it is not difficult either. For troubleshooting the methods and tools shown above work great. If you need detailed characterization data, however, you probably need to look at a dedicated bench LCR instrument which uses slightly more advanced techniques. But. If your goal is to verify whether or not a big electrolytic is good, then either the scope or ESR70 method will suit you well.

Have you had to make capacitor measurements before? What have you done? Or, do you have a story of when you SHOULD have measured a capacitor before applying power? Leave a comment with you stories.

P.S. I should point out that PEAK offers a range of these component specific meters. I asked element14 to send me the semiconductor (transistor) meter, DCA75DCA75. I’ll be reviewing that in the near future. In the mean time, I already bought the LCR40 LCR40 on my own. It measures inductance, capacitance, and resistance. It does not, however, measure a capacitor's ESR. (It can, however, measure an inductor's ESR, but that is a story for another post.)