I just found out that the Sencore Z Meter uses a patented circuit using DC instead of AC to measure ESR. They use DC to charge the capacitor and have found that when there is a series resistance, the charging voltage instantly rises to a voltage level (proportional to the resistance in the cap and the output impedance of the power supply) before beginning the normal R-C curve. A normal R-C curve with no ESR starts at zero voltage since the cap acts as a short for an instance but if there is ESR there is a voltage jump because of the ESR. The Z Meter measures the voltage rise in the first microsecond after applying the voltage and then calculates the resistance using Ohms law and they say that true ESR does not depend on frequency since it is a pure resistive loss and has the same impedance at any frequency. This is new to me and it's still soaking in. Sure changes my perspective on the relevance of the frequency used to measure ESR. If anyone finds the patent, I'd sure like the link.
Thanks
Hank

I don't have access to the patent, but do have a couple of comments.

First, the concept of ESR is very old. I've found it mentioned in wireless textbooks that date back to WW-1 and the early 1920s. The concept of a surge impedance test, which is what you describe as the basis of the Sencore patent, goes back even further in connection with transmission lines and antennas. Presumably their patent was for the methods they used to obtain the measurements, not the basic concepts themselves.

Second, a rigorous definition of ESR includes the impedance due to the inductance of the leads, foils, and construction of the capacitor, in addition to dielectric hysteresis losses. These are absolutely AC related and frequency dependent. If this were not so, we could use "orange drops" (for an example) at microwave frequencies just as easily as we do at audio. Apparently what Sencore is measuring is not real ESR at all, but the increase in surge impedance that occurs due to evaporation of electrolyte, internal corrosion, or chemical breakdown in faulty electrolytic capacitors. It would make sense, since they were building service instruments that were meant to identify bad components, not laboratory analyzers, but it would have been better if they had not referred to that as "true ESR."

Hank, good post...

I've felt most ESR readings were misleading, unless the test frequency
was specified. It was interesting to read about work arounds for that issue.

But, I think most of us get by with the service grade ESR meters since
we're only worried about caps that operating in power supplies and
at worst, caps operating at switching power supply frequencies.

Pete

Look at the esr versus frequency curve for an electrolytic capacitor and you will see that esr is frequency dependent. One component of esr is a dc resistance, and that is important, but the ac component is just as improtant in ac applications like switching power supplies.

Impedance includes reactive components that are frequency sensitive but DC resistance is not. DC resistance results in I squared R losses which shows up as heat but the other reactive components do not. AC losses in a pure DC resistance will also show up as heat.
Hank

I don't know how this fits into the discussion, but: Even though resistance is not a function of frequency, ESR IS..........the subtlety is that the "implied resistor" does not physically exist in the device.
In general, one would want to assume that any component in an equivalent circuit would be frequency-dependent. There are of course exceptions.

If Sencore can identify bad capacitors with their circuit, that's the object of the game. And ESR is the curent buzzword that everyone recognizes. But I think that strictly, that's not how ESR is defined. A microsecond pulse is a megahertz and I wouldn't expect all electrolytics to be rated for that.

ESR is complicated by a capacitor's tendency to rebuild a charge some time after terminal
voltage is zero.

So the first pulse ESR might be true in the case of an inert dielectric. But in a chemical dielectric
if the capacitor fails to discharge completely, the usefulness as a filter element diminishes because
the AC impedance rises.

From what I can tell from the info I have read Sencore doesn't send a pulse but measures what the applied voltage rises to in the first microsecond. During that time frame the cap acts a short and applied voltage is divided by internal resistance of the supply and the caps ESR. Also from what I've read so far I think the ESR causes a cap with high ESR to get hot (I^2R) and if that's the case it's real DC resistance. I'm all eyes and ears.
Hank

We maybe are picking flyspecks out of pepper here.... The power dissipation occurs whenever the voltage and current are anything but 90 degrees out of phase. Thus, at the test frequency, if there is significant ESR, it **appears** that there is a DC resistance. The only way to know if there is physical DC resistance is to measure it using a DC source. (tough to do when dealing with a capacitor.... )

Not if it has sufficient inductance.

The only way to know if there is physical DC resistance is to measure it using a DC source. (tough to do when dealing with a capacitor.... )[/quote]

From the Sencore info that I read that is what they do when they measure the the DC voltage rise in the first microsecond because a discharged cap is basically a DC short for that instant. But I don't know what effect dielectric memory has..
BTW I have some experience picking fly specks out of pepper with boxing gloves on! (why do they call them gloves?) But this more fun and educational, I hope!

Good point but I don't have any idea how much L a high ESR elec cap might have. Could di/dt be a significant factor?

If you are going to use resistance and heat generation in a capacitor, simple call it
what is is. Power Factor. This is the measure of the ratio the Volt Amps to the Watts
a capacitor downstream of a wattmeter reads on AC only.

On DC, it is simply, as wd5jfr stated, I squared R losses.

There's a Quadtech brochure on ESR and DF here:
http://www.low-esr.com/QT_LowESR.pdf

Hank

I think Sencore is mostly blowing smoke on this one.

You can't measure ESR by applying DC. By definition, a capacitor can't pass DC.

What they're doing is applying an impulse, which is anything but DC, a 1 microsecond pulse is made up of hundreds of AC harmonics.

And theres nothing sacred about a pulse. They could just as easily apply a 500KHz sine or square wave, and see how much of it the capacitor passes. A 10uF capacitor looks like a 15 ohms at 1MHz, so if you see more ohms than that, the amount over 15 ohms is ESR, more or less.

And all of this is mostly off-topic for old radios.

On the contrary, I think it is important that those who restore or service electronic devices, such as vintage radios, understand the basic elements of electricity and electronics. Discussions like this are amongst the best the ARF forums have to offer.

The reactance of a 10uF condenser at 1mHz is actually 15.9 milli-Ohms, or 0.0159 Ohm. At the popular and practical testing frequency of 100kHz, at which the inductance of a condenser is not yet a significantly complicating factor, the capacitive reactance of the condenser is still very low, at 0.159 Ohm. As Alan Douglas pointed out in the original thread (where the harmonic content of pulses was also discussed in this context)—

http://www.antiqueradios.com/forums/viewtopic.php?f=8&t=194460&start=20

—at a high testing frequency, the capacitive reactance is considerably lower than the ESR, especially in the case of a bad condenser. Thus, the reading of an ESR meter, which is essentially a high-frequency A.C. ohm-meter, consists predominantly of ESR.

I was only off by a factor of a thousand!

15 milliohms sounds better, that will let the ESR predominate as the source of any measured voltage.

One really cheap ESR tester is to get any old phone-charger, the light ones, with a switching power supply-- apply the AC voltage from that through a 1 ohm resistor to any capacitor and use a scope or AC meter to see how much AC voltage appears. Should be very little.