Analog ESR meter design

[Th3_uN1Qu3]

While there are a number of analog ESR meter designs on the web, none of them are sensitive enough for low ESR caps. Also they share a common fallacy which i'll point out below.

I won't go into the digital ones because from what i've seen so far most designs that are offered for free are also pretty dodgy, and if you're not into programming micros buying a programmer just to build one piece of kit is overkill, and most of you won't justify the expense.

So, how does an analog ESR meter work? Well, it's quite simple really. It's an resistance meter (ohm meter) just that it works in AC instead of DC.

• First we have an oscillator, which should be 100kHz as that's where today's SMPS caps are spec'd at, but there are designs on the net operating at various frequencies which will be more or less accurate depending on the type of cap you are measuring...
  
• Then we have a current amplifier which makes this AC signal low impedance, so there is no voltage drop due to the source. This has been implemented in various ways - opamps, transistors, transformers, and so on. Sometimes the oscillator and amplifier stage are one and the same.
  
• Then we have a sense resistor in series with one of the probes, which drops a known voltage. When you short the probes, there's only this sense resistor in the circuit, and you use this to set the maximum indication of your meter movement. The capacitor to be tested is put in series with this sense resistor.
  
• And finally we have a rectifier and filter, which converts the AC to DC so that the meter can read it. These ESR meter designs typically use a 50 or 100 uA meter movement.
  
• Sure there's other things such as input protection and such but the four building blocks i mentioned above are the foundation of any analog ESR meter.

Now, the first thing wrong with most ESR meter designs on the web is that the sense resistor is way too high in value, so low ESR caps cannot be read accurately - when the bottom of the range is 50 ohms there is no way you're gonna spot a 0.1 ohm difference let alone a 0.01. So let's drop the value of the sense resistor. If you do this without changing anything else you'll quickly notice something funny - the meter will not go to max scale anymore! That's because by lowering the value of the sense resistor, less voltage gets dropped by it, so there's less voltage available to drive the meter. Some designs try to get around that by adding a gain stage to feed the meter, but this opens a whole new can of worms as the whole thing becomes extremely sensitive to outside interference.

So, what's the quick and easy solution? Increase the current, duh! This solves all our problems - we can use a low value sense resistor AND still have enough voltage drop to show on the meter. As to why nobody thought it before, there are two issues:

1. It eats battery. Most meters on the web are designed to be powered off a 9 volt battery. That won't work if you're trying to draw a few hundred milliamps.
2. When applied to a cap with high ESR (almost open circuit) it will develop a voltage that will turn on semiconductors, causing erroneous readings.

Point 2 can be gotten around fairly easily. As for the first one, not quite doable... So, what i'm going to do is design a bench meter, that runs off 12 volts. Or if you have a NiMH pack handy...

Everyone post their wishlist here, and i'll see what i can make of it.

[shovenose]

Re: Analog ESR meter design

Make it simple and easy to use, like an Apple product
guaranteed success.
Not really kidding.
-shovenose

[Th3_uN1Qu3]

Re: Analog ESR meter design

Simple it is... Reminds me of this particular quote:

If you've always done it like that, then it's probably wrong.

Henry Ford said it.

What all analog ESR meter circuits have in common is that they use a constant AC voltage source passed thru a shunt resistor, then thru the cap under test in series. The idea is that depending on the ESR of the cap, this circuit produces a certain AC current which makes a certain voltage drop across the sense resistor, and that is shown on the meter.

Now, to me this approach seems ass-backwards, because the source resistance of the voltage source isn't exactly zero so errors creep in, then if the shunt resistor is too high a value it won't read low ESR caps, and if it's too low the signal voltage becomes too low to be useful.

So instead of doing the same old thing i decided for a different and (at least to me) more straightforward approach, which is feeding the cap thru a constant AC current source which depending on the cap's ESR produces an AC voltage across which can be read directly across the cap itself, no shunt resistor involved!

And the true gem - the voltage source circuit produced maximum voltage output when the ESR was zero. This circuit produces maximum voltage output when the ESR is high, therefore, if you calibrate it properly, it can also be used with a DMM and display a voltage proportionally to ESR, for example 1mOhm = 10mV ! Now how's that for coolness factor.

[shovenose]

Re: Analog ESR meter design

Simple it is... Reminds me of this particular quote:



Henry Ford said it.

What all analog ESR meter circuits have in common is that they use a constant AC voltage source passed thru a shunt resistor, then thru the cap under test in series. The idea is that depending on the ESR of the cap, this circuit produces a certain AC current which makes a certain voltage drop across the sense resistor, and that is shown on the meter.

Now, to me this approach seems ass-backwards, because the source resistance of the voltage source isn't exactly zero so errors creep in, then if the shunt resistor is too high a value it won't read low ESR caps, and if it's too low the signal voltage becomes too low to be useful.

So instead of doing the same old thing i decided for a different and (at least to me) more straightforward approach, which is feeding the cap thru a constant AC current source which depending on the cap's ESR produces an AC voltage across which can be read directly across the cap itself, no shunt resistor involved!

And the true gem - the voltage source circuit produced maximum voltage output when the ESR was zero. This circuit produces maximum voltage output when the ESR is high, therefore, if you calibrate it properly, it can also be used with a DMM and display a voltage proportionally to ESR, for example 1mOhm = 10mV ! Now how's that for coolness factor.

I didn't understand a work of it but it sounds cool

[Th3_uN1Qu3]

Re: Analog ESR meter design

Then stop spamming. At least don't quote in reply when the previous post is right above... as admin somewhere else i thought you knew of those basic forum rules.

[momaka]

Re: Analog ESR meter design

Sounds like a great project!

Everyone post their wishlist here, and i'll see what i can make of it.

Perhaps easy to build - i.e. doesn't require any special, hard to find ICs or anything like that.
My spare parts bin is quickly growing. It would be nice if I could use some of them to built this meter.

[Scenic]

Re: Analog ESR meter design

What I've found so far was either ridiculously overcomplicated (µC with proprietary software, needing a programmer), or so crude you'd be lucky if it worked at all.
That, and because premade ones like the ESRMicro are still rather expensive if you don't do this all day long is basically why I don't have an ESR meter at all.

Something in between which doesn't need a microcontroller, but still gives useful results for checking "our" kind of caps would be great.

And like momaka said.. simple off the shelf components.. not some special über-rare ICs or something.

If this thread goes anywhere, it's gonna be interesting I bet

[Th3_uN1Qu3]

Re: Analog ESR meter design

Definitely no micros or hard to find ICs - heck, i'm lazy enough that it might even be no opamps, just a comparator and a few transistors.

The most advanced device inside will be a rail-to-rail comparator - so you can calibrate the AC current source at a lower frequency with a true RMS multimeter, without the need of an oscilloscope, and then switch to 100kHz resting assured that the current remains the same. I wanted to use a 555 timer for the oscillator at first (especially since it can provide enough current without the need of an output buffer), but the output voltage variance from 50Hz to 100kHz is too high.

Anyway this is a linear circuit - absolute accuracy of the current source isn't that important, because you can simply take some precision resistors and calibrate the output against them, but low drift over time IS something i'm aiming for.

[momaka]

Re: Analog ESR meter design

There's also this ESR meter which doesn't look all that complex:
https://www.badcaps.net/forum/showth...&highlight=esr
I wonder if it works well. The only thing I don't have to build that one is those 2 particular transistors and op-amps. I do have general purpose UA741 op-amps, though.

[Th3_uN1Qu3]

Re: Analog ESR meter design

741 is not suitable for this circuit. Use TL072/082 or NE5532. The basic idea however is the same as mine - i woulda been surprised if it hadn't been done before - but i want to take it one step further and use a current source instead of just a resistor. You see that in the comments it says that it needs a stable 12 volts power input - and a 7812 may not be enough for high precision. I'm going to feed the current source from a TL431 reference - that should keep it in check regardless of the input voltage.

And i'll likely be using simple transistor gain stages instead of opamps, no need to complicate things.

[momaka]

Re: Analog ESR meter design

741 is not suitable for this circuit.

Kind of anticipated that. I was hoping to reuse some of the stuff we got in our EE kits from school, but I guess not :P . I did use the UA741 in conjunction with a KA431 and a MOSFET to make a regulated 1.5V supply (similar to the 3.3v MOSFET-derived rails in PSUs). It worked fairly okay for the most part.

By the way, is your ESR meter going to be an addon for an analog multimeter or a digital one to show the ESR readings (or neither?)?

[Th3_uN1Qu3]

Re: Analog ESR meter design

Addon to digital meters.

[Th3_uN1Qu3]

Re: Analog ESR meter design

First prototype is complete. It works... sort of. For now it only works correctly at 50Hz so it is useless for testing ESR, but it does a good job as a low ohms meter. Currently it can do either 1 volt/ohm or 10 volts/ohm, so yes, it can measure with 0.1mOhm resolution on a cheap multimeter! Anyway, this proves that the concept is workable, but the realization could use some improvement.

Some of the trouble is also given by the build method - it's wired in a rather messy way on a perfboard, so i'll be designing a PCB and see what that gives.

There is another subtlety that i only realized after building the real thing - when measuring the ESR of a capacitor with a continuous AC signal, you also have to discharge the capacitor at the end of each cycle, otherwise the capacitor will charge up until it hits the limit of the voltage rail that the current source is powered from, and give completely bogus readings!

Edit: It's easy to overlook simple solutions when you're concentrated on something - the humble resistor connected to a pulsed voltage source works just fine... So the design mentioned above is technically correct, but you've got the drift of the 555 to take account of - you noticed that in the instructions it mentions using the relative function of your multimeter to calibrate the offset. What if your DMM doesn't have relative? I'm trying to avoid offset altogether.

[momaka]

Re: Analog ESR meter design

First prototype is complete. It works... sort of. For now it only works correctly at 50Hz so it is useless for testing ESR, but it does a good job as a low ohms meter. Currently it can do either 1 volt/ohm or 10 volts/ohm, so yes, it can measure with 0.1mOhm resolution on a cheap multimeter!

That's still pretty cool! Given that my cheap 830D multimeter can't measure anything under 2 Ohms, and for up to 50 Ohms it has a 2 Ohm offset, your prototype can be pretty useful for finding short circuits on a board.

[cmj21973]

Re: Analog ESR meter design

Sounds like the project is going well.
Post or PM some pics & schematics when you get a chance.

[Th3_uN1Qu3]

Re: Analog ESR meter design

Sure, i will post schematics. I just want it to be as simple as possible. Using the circuit that momaka mentioned above, with the relative function on your DMM and all, it reads negative when trying to test a capacitor, and has weird results with resistors as well. Negative as in lower than the relative value you zero'd your meter to - because this circuit has a huge amount of offset. Of course getting an actual negative voltage from the circuit ie below ground is impossible, but the point is that when measuring a cap the output voltage is lower than with probes shorted.

The circuit which i built, with the current source, running at 50Hz, works great for resistors, wires, PCB traces and the like, and has very, very low offset and drift. At higher frequencies, again, it goes nuts. Also, when testing a capacitor you also have to provide the means to discharge it after each cycle (which a current source does not), and all methods i've tried so far produce interference which translates into offset or worse, it makes the readings non linear. A simple resistor from the output of the 555 as in the circuit referenced above should work but yet it does not. As i no longer have an oscilloscope (had to sell it due to financial reasons in November), it's hard for me to troubleshoot the circuit - because it works in simulation.

Don't worry tho, the bag of tricks is not yet empty.

[tom66]

Re: Analog ESR meter design

If I were to design an ESR meter, this is how I would do it.

Constant current source pulses 1A thru cap, 100kHz @ 10% duty. Low-RdsON MOSFET (<5mohm) zeros cap on every cycle. Differential amplifier senses voltage across cap (through two less critical switches) and peak detector captures these. If the multimeter's impedance is known or guaranteed, then the output buffer can be removed.

Problems:
- the discharge and charge waveforms need to be even duty cycle (10% each) but out of phase, or the accuracy suffers
- the cap builds up a charge to about ESR * current (1 ohm ESR = 1V) which could bias semiconductors (a lower current may fix this.)

Main op-amp would be replaced with an op-amp and power stage (NPN follower, probably.) The op-amp needs to be fast (no overshoot.) An ideal model is used in the simulation.

Output is 1V/ohm and can measure from 1mohm to 1 ohm with decent accuracy (+/-2%.) At low esr offset error becomes a problem - with a 1mohm cap it tends to measure about 1.7mohm. Cap should be >100uF for best results. No overload protection, easy to blow up if you plug a charged cap into it.

Paste into falstad.com/circuit:

$ 1 5.0000000000000004E-8 24.46919322642204 50 5.0 50
r 224 256 224 336 0 0.0010
c 224 352 224 416 0 0.0010 0.0023163257658775777
r 224 416 224 496 0 0.01
g 224 496 224 512 0
w 224 416 272 416 0
159 272 400 272 272 0 0.0010 1.0E10
w 272 400 272 416 0
w 224 256 272 256 0
R 288 336 320 336 1 2 100000.0 2.5 2.5 1.7453292519943295 0.1
w 272 272 272 256 0
w 224 336 224 352 0
a 80 256 192 256 1 15.0 -15.0 1000000.0
w 192 256 224 256 0
R 80 240 48 240 1 2 100000.0 0.0050 0.0050 0.0 0.1
w 80 272 80 416 0
a 432 336 544 336 1 15.0 -15.0 1000000.0
159 288 256 352 256 0 20.0 1.0E10
159 352 416 288 416 0 20.0 1.0E10
w 288 416 272 416 0
w 288 256 272 256 0
r 352 416 432 416 0 100000.0
w 432 352 432 416 0
r 432 416 544 416 0 100000.0
w 544 416 544 336 0
r 352 256 432 256 0 100000.0
w 432 256 432 320 0
r 432 256 544 256 0 100000.0
g 544 256 560 256 0
R 320 272 320 304 1 2 100000.0 2.5 2.5 0.0 0.1
w 320 272 352 272 0
w 320 400 352 400 0
w 352 400 352 272 0
r 192 416 192 256 0 10000.0
w 224 416 192 416 0
w 192 416 80 416 0
a 592 352 704 352 1 15.0 -15.0 1000000.0
w 544 336 592 336 0
w 592 368 592 400 0
d 704 352 704 400 1 0.805904783
c 704 400 704 448 0 3.3E-10 0.003485802029299198
g 704 448 704 464 0
a 832 416 944 416 1 15.0 -15.0 1000000.0
w 832 400 768 400 0
w 832 432 832 464 0
w 832 464 944 464 0
w 944 464 944 416 0
w 944 416 976 416 0
w 704 400 592 400 0
r 704 400 768 400 0 1000000.0
r 768 400 768 448 0 1000000.0
g 768 448 768 464 0
o 46 64 0 35 0.02288728493827538 0.0 0 -1

[ben7]

Re: Analog ESR meter design

Nice, thanks for the clarification on how esr meters work! I just thought that I need an esr meter, but the price of around $50 put me off ... why not make it for less than $10? Then I came across this thread today.

A question though, where the heck do you find a 50 or 100uA meter movement?

-Ben

[mariushm]

Re: Analog ESR meter design

just a heads up for those interested in this
there's a thread on eevblog nowadays which mentions several circuits and designs in it.. on site linked there contains the source code for the device if one wants to make it.

http://www.eevblog.com/forum/product...cseen#msg90514

[severach]

Re: Analog ESR meter design

why not make it for less than $10?

China could add ESR in as a scale on any number of meters making ESR measurement essentially free. For some reason they don't.