EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: JohnS_AZ on April 26, 2011, 05:57:24 pm

Title: Building an ESR meter
Post by: JohnS_AZ on April 26, 2011, 05:57:24 pm
Hello guys,

I want to build an ESR meter and after searching the web (and this forum) have found two contenders. Analog design is not my strong suit, and was wondering if you would have an opinion between the two. One takes a wacky transformer and that's fine. My -only- concern is which one do you think would be superior in terms of accuracy/stability.

Thanks.

Option #1 http://ludens.cl/Electron/esr/esr.html (http://ludens.cl/Electron/esr/esr.html)

Option #2 http://members.shaw.ca/swstuff/esrmeter.html (http://members.shaw.ca/swstuff/esrmeter.html)

Title: Re: Building an ESR meter
Post by: joelby on April 26, 2011, 06:13:49 pm
The Bob Parker meter seems to be fairly popular here. You can read the original article at http://members.ozemail.com.au/~bobpar/k7214.pdf (http://members.ozemail.com.au/~bobpar/k7214.pdf) . The firmware for the microcontroller doesn't seem to be available unless you buy the kit, but you could certainly have a look at the circuit diagram if you need another option to choose from.
Title: Re: Building an ESR meter
Post by: sacherjj on April 26, 2011, 07:58:27 pm
The Bob Parker meter seems to be fairly popular here. You can read the original article at http://members.ozemail.com.au/~bobpar/k7214.pdf (http://members.ozemail.com.au/~bobpar/k7214.pdf) . The firmware for the microcontroller doesn't seem to be available unless you buy the kit, but you could certainly have a look at the circuit diagram if you need another option to choose from.

Not getting anything from that link.
Title: Re: Building an ESR meter
Post by: Bloch on April 26, 2011, 08:22:12 pm
Not getting anything from that link.

It works fine :-)

Try right click and save.

And be sure to have a Adobe reader installed.

/Lars G
Title: Re: Building an ESR meter
Post by: am2pgs on April 26, 2011, 08:56:08 pm
The Bob Parker meter seems to be fairly popular here. You can read the original article at http://members.ozemail.com.au/~bobpar/k7214.pdf (http://members.ozemail.com.au/~bobpar/k7214.pdf) . The firmware for the microcontroller doesn't seem to be available unless you buy the kit, but you could certainly have a look at the circuit diagram if you need another option to choose from.



IMHO the Bob Parker approach to ESR measurement seems to be more superior. Having said that, I don't like the Analogue to Digital conversion method described. It seems too crude, complicated and prone to inaccuracies.

These days it is not that difficult to use the same measurement principle but with nice and cheap uC (like a small PIC) with a proper built in A/D.
Title: Re: Building an ESR meter
Post by: jimmc on April 26, 2011, 10:40:47 pm
General-purpose ESR meters do not have to be precise; we are looking for gross changes from normal (in the region of 50%).
Either of the designs mentioned will be usable although both will drift a bit with temperature.

Bob Parker’s meter is an elegant design for those who like a digital readout.
(Because of the pulsed nature of the test current, a standard A-D converter would need to be preceded by a sample and hold circuit adding to the complexity.)

My own preference is for an analogue meter and below are some notes on  a design I posted on another forum a while back.  I did not write this up as a project for others to build, more to show the design considerations involved.


My requirements were:
Indication from 0.2 to 20 ohm
Measurement frequency 100kHz
Protection against charged capacitors and 240vAC
Temperature stability
Power from 9v PP3 with low current consumption
 
(https://www.eevblog.com/forum/index.php?action=dlattach;topic=3228.0;attach=17018)

A quick look at typical ohmmeter circuits showed that to achieve a scale 0.2 to 20 ohm, a source resistance (Rs) of around 2 ohm would be required.
The modified form of Option1 was chosen as the easiest to implement. As a starting point Rs’ and Rm were chosen as 1 ohm.

To avoid excessive drive power a sensitive detector would be required.  Experimenting with a millivoltmeter showed that with 1 ohm across its input, a FSD as low as 1mV rms was practical.  This meant that the drive current required would be 2mA rms (with Rx = 0 Rs’ and Rm are in parallel).

In my opinion OP-AMPs are not the best choice at 100kHz; the gain of most general-purpose types is low at this frequency (< 30). As DC precision is not required a single transistor can easily out perform them.
To drive a diode bridge from 1mV you need a gain of at least 1000 (in round terms) implying a gain-bandwidth product 100MHz minimum, no the province of a common Op-Amp. For good linearity down to 1/10 FSD this figure goes up to 1GHz!
(As an aside, I think this is how several of the other ESR meters 'get away' with Rs in the order of 20 ohm. Low gain before the diodes gives an expanded scale at the expense of temperature stability.)


A 5v power rail was chosen using a low drop-out regulator which will regulate down to 5.2v input with a quiescent current of < 200uA.

(https://www.eevblog.com/forum/index.php?action=dlattach;topic=3228.0;attach=17020)

The detector circuit was based on a design first seen in ‘Ferranti E-Line Transistor Applications’ 1970. (see 2nd image).  T1 & 2 drive a 100uA meter to FSD with 22mVrms on the base of T1.
T3 & T4 form a pre-amp with a gain of around 22 to give the required 1mV sensitivity.  (The LF gain is approx 28 (=R9/R8 + 1) with an HF roll-off provided by C7)
Capacitor values were chosen for 100kHz operation (100nF at 100kHz has a reactance of 16 ohm)
The circuit was built and found to perform as expected. Deflection for zero input with a 1 ohm resistor across the input was around 0.5%FSD. Surprisingly no supply decoupling other than C8 (tantalum) was required for stability.
Current consumption was under 2mA.

(https://www.eevblog.com/forum/index.php?action=dlattach;topic=3228.0;attach=17022)

The source was based on a CMOS version of the ubiquitous ‘555, swapping the normal roles of the output and discharge pins gave a 1:1 Mark-Space ratio and increased drive current at the expense of an active pull-up (T5) (3rd image)
(R21, a SOT resistor, was added later to trim the M-S ratio to exactly 1:1)

I used the CMOS version of the '555 (http://www.intersil.com/data/fn/fn2867.pdf (http://www.intersil.com/data/fn/fn2867.pdf))
Much lower power and no nasty current spikes on the supply rail, but the output starts to sag sourcing more than about 1mA from a 5v supply. The (open drain) discharge output can easily sink 15mA and an external active pull-up ensures a large enough source current.
Driving the timing resistor from the output of a standard (bipolar) '555 doesn't work as well, the output 'High' voltage is only about 3.5v which is only just above the 2/3Vcc upper threshold of the comparator section. This will give a very asymmetric M:S ratio.
see (http://www.unitechelectronics.com/NE...#internal-ccts (http://www.unitechelectronics.com/NE...#internal-ccts)).


An inductor L2, originally a 1mH unshielded inductor, was added to attenuate the harmonic content of the square wave.
This had such a large stray field that probing with a ‘scope was difficult. It was replaced by a toroidal inductor, which cured the problem. The DC block (C12) was chosen to resonate with this inductor.
The pot network was chosen to give an adjustment range of around –20 +40% to allow for the effect of test leads.
Protection against charged capacitors was provided by C11 an X2 rated capacitor.  Its reactance was cancelled by L1 another toroidal inductor.

Without L1 the reactance of C11 (3.5 ohm) would raise the source impedance 4 ohm.
This would mean that mid scale would be around 5 ohm and, for example, that the current for an ESR of 0.2 ohm would be 2.6% below FSD rather than 9.1% below FSD. (All assuming that the reactance ot the test capacitor is negligible.)


The loss associated with C11 & L1 added less than 0.2 ohm to the effective value of Rs.  With a loaded Q of less than 2 (Xc = 3.5, Rs = 2) minor frequency drift should not cause problems.  As a bonus the reactance of C11 at 50Hz is 6.8k so any mains pick-up will be attenuated by 76dB.
Protection was completed by D5 to D8.
(I had a senior moment here, a single pair of back- to-back diodes between R16 and R18 would have been better.)
 
Finally there was the problem of re-scaling the meter.
First I measured the meter reading for a range of known resistors.
Then I removed the scale plate and scanned it (with steel rules as a reference).
By importing the scan into TurboCAD I was able to accurately measure the dimensions (both linear and angular).
After a steep learning curve I found that it was fairly easy to draw the new scale and even tilt the numbers to match.

Current consumption for the completed unit was 3.7mA.

(https://www.eevblog.com/forum/index.php?action=dlattach;topic=3228.0;attach=17024)




A perfect 1uF capacitor should give an apparent ESR of 0.56 ohm with Rs = 2 ohm. I measured a film 1uF capacitor and got a reading just below 0.5 ohm. The slight discrepancy is caused by losses in L1 / C11 raising Rs to around 2.2 ohm.

 Jim

ps just noticed a mistake in my original post, T4 is a ZTX502 (not 302)

If the high voltage protection is omitted, the source may be simplified and no inductors are required.

(https://www.eevblog.com/forum/index.php?action=dlattach;topic=3228.0;attach=17029)
 

Title: Re: Building an ESR meter
Post by: Fraser on April 26, 2011, 10:47:39 pm
I built the Bob Parker design that was sold by Dick Smith Electronics in Austarlia. The unit appears to work well but I have no reference with which to compare it.

Considering that capacitors are known to be a relatively weak link in terms of equipment longevity & MTBF, I am surprised that thers is not more choice in ESR meters. I can find very few commmercial offerings that would be superior to the Bob Parker instrument. There is a chap in Canada building analogue and digital display ESR meters but some self proclaimed Capacitor Oracles on the NET have suggested that the available analogue ESR meters can't correctly display the very low ESR that is present in some modern high performance capacitors and so should be avoided. I have no comment on this as I deal with everyday consumer grade capacitors that work fine with both analogue & digital ESR meters.
Title: Re: Building an ESR meter
Post by: DaveW on April 27, 2011, 12:09:11 am

Option #2 http://members.shaw.ca/swstuff/esrmeter.html (http://members.shaw.ca/swstuff/esrmeter.html)


I've got this one sitting on my bench along with a commercial esr meter. You don't get quite the linear response you'd like( haven't tested this rigorously, but have compared to the commercial one), but works well enough for finding faulty electrolytics in and out of situ. So simple and cheap design, works well enough for most of what you need an esr meter for. I've heard that the Bob Parker design is better for more exact uses, no personal experience with that though...
Title: Re: Building an ESR meter
Post by: am2pgs on April 27, 2011, 01:10:59 am
General-purpose ESR meters do not have to be precise; we are looking for gross changes from normal (in the region of 50%).


may be so, if you are just interested in detecting gross failures. But then again the digital option is so much more elegant/simpler, more stable and without the all the trimming needed for the analgue option.

all the uC I have worked with have sample/track and hold circuits built in with configurale aperture down to a few hundred nano-seconds. It is just a matter of having a reasonably good A/D Driver with low source impedance and timing in software.

The only things to consider for accuracy are the current source tolerance/stability ,A/D reference voltage and DC offsets, all of which can be easilly taken care of with careful design. the timing tolerances will probably be pretty good with a reasonably fast uC and stable clock.

 

Title: Re: Building an ESR meter
Post by: sacherjj on April 27, 2011, 02:29:28 am
Not getting anything from that link.

It works fine :-)

Try right click and save.

And be sure to have a Adobe reader installed.

Working for me now.  I was getting server not found errors before.
Title: Re: Building an ESR meter
Post by: david77 on April 27, 2011, 10:03:12 am
I've built this
http://www.elv-downloads.de/service/manuals/ESR1/ESR1_KM_G_021017.pdf (http://www.elv-downloads.de/service/manuals/ESR1/ESR1_KM_G_021017.pdf)
and it seems to work quite fine, although I have no other ESR meter to compare it to.

I chose this circuit mainly because it uses only standard parts that I had lying around anyway.
Title: Re: Building an ESR meter
Post by: kripton2035 on May 16, 2011, 09:52:16 pm
you can have a look at my esr meter repository page
you can compare all available esr meters, ready to build or buy yourself.
regards,

http://kripton2035.free.fr/esr-repository.html (http://kripton2035.free.fr/esr-repository.html)
Title: Re: Building an ESR meter
Post by: Flavour Flave on May 28, 2011, 09:39:30 pm
Quote
(Because of the pulsed nature of the test current, a standard A-D converter would need to be preceded by a sample and hold circuit adding to the complexity.)

Why? Just interested because I thought you could do it  in software, pray tell!
Title: Re: Building an ESR meter
Post by: torch on May 30, 2011, 01:06:38 am
Is there any reason why the method of using a function generator and scope, as described here: http://geoffg.net/Measuring_ESR.html (http://geoffg.net/Measuring_ESR.html) could not be used at a higher frequency (say, 100KHz) to find the ESR of smaller value caps? (He is using a 1KHz wave and can only measure down to 10uF). Also, I note that he is using a 10v signal while the ESR meter designs all seem to generate a fraction of that (IIRC, Bob Parker blue ESR meter is 100mv output) so as not to turn on any semi-conductors in the circuit. Any reason not to turn down the output voltage to something similar?

I realize that a scope and function generator is not nearly as elegant and portable as an actual ESR meter, but will it work for the interim?
Title: Re: Building an ESR meter
Post by: jimmc on May 30, 2011, 03:27:35 pm
Quote
(Because of the pulsed nature of the test current, a standard A-D converter would need to be preceded by a sample and hold circuit adding to the complexity.)

Why? Just interested because I thought you could do it  in software, pray tell!

Senior moment I'm afraid, I hadn't realised that even a cheap modern A-D , such as in a PIC, could easily acquire a pulse of a few uS.

Jim
Title: Re: Building an ESR meter
Post by: torch on May 30, 2011, 08:08:30 pm
Is there any reason why the method of using a function generator and scope, as described here: http://geoffg.net/Measuring_ESR.html (http://geoffg.net/Measuring_ESR.html) could not be used at a higher frequency (say, 100KHz) to find the ESR of smaller value caps? (He is using a 1KHz wave and can only measure down to 10uF). Also, I note that he is using a 10v signal while the ESR meter designs all seem to generate a fraction of that (IIRC, Bob Parker blue ESR meter is 100mv output) so as not to turn on any semi-conductors in the circuit. Any reason not to turn down the output voltage to something similar?

I answered my own questions: it all works just fine. Larger caps show a virtually flat line with a very narrow spike at the start of each square wave input. A 0.47uf cap has a slight ripple waveform (with the aforementioned spikes). A 0.1uf cap produces a shallow triangle wave between the spikes. A 3pf cap passes the square wave virtually unchanged. Seems to work equally well in circuit as without.

So, I now have built an ESR with a component count of 6: 2 bnc connectors, 1 project box, 1 51 ohm resistor, 1 scope and 1 function generator  :D
Title: Re: Building an ESR meter
Post by: Mechatrommer on May 30, 2011, 09:54:53 pm
mine is 5, minus the project box :P