ESR Meter

[electronwaster]

Hi,

I would like to build an ESR meter, to measure the equivalent series resistance of electrolytics (you know, the ones that fail all the time). Apparently an ESR meter can check these in circuit, since to AC, they should look as open-circuit as possible. One of these could be a very handy addition to the tool kit, I think.

I found a nice simple circuit, which I could make a PCB and case for:

http://ludens.cl/Electron/esr/esr.html

The problem is, I don't have a 20:1 transformer, as specified in the circuit. There is a square wave oscillator at the input, which is creating the 4 or so volt on-off signal, and this needs reducing to 200mV, which is what the transformer does.

What I'd like to do is eliminate the transformer (as much for the fun as the convenience) use a resistive divider to reduce that voltage, then an opamp in voltage follower mode (or just a BJT voltage follower) to increase the output impedance. Insane?

Can anyone explain why this shouldn't work? If people think it should work, I might just give this a go. If no-one knows, I might simulate it and see what happens. If there's any interest, I'll post results here.

Cheers,
electronwaster

[kripton2035]

hi,
yes you can try a simulation and give us the results, it is always interesting
or you can also try it on a breadboard?
anyway this schematic is quite old, and I made a web site with all (?) available esr meter schematics
most are really better than this one, just pick one.
here it is : http://kripton2035.free.fr/esr-repository.html
regards,

[electronwaster]

Wow - that's awesome, thanks! (I'm glad I asked)

So - have you made many (any?) of these?

Can you recommend one? My desired specs are as follows:

Simple circuit (understandable by a beginner - me), ideally no transformers, ideally standalone, I don't want to have to plug into the oscilloscope or multimeter. I'd like it to have a small analogue needle (I have one of these, and have made my own scales for them before, so that's no problem), or even an LED bargraph, though that needs another chip, probably.

Thanks,
electronwaster

[kripton2035]

the poptronix esr meter is what you need...
http://kripton2035.free.fr/analog%20esr/esr-poptronix.html

I made this one and it is enought for me - I made a bargraph with a lm3914 because I didnt have a level meter ...
works fine, found a lot of bad caps with this one ... that's the main usage !

I'm currently planning to build some of the others (the digital russian "go", the "kinza", and the cxem.net)
to give impressions and compare them really after that on the web site.

[jimmc]

For a (perfect) transformer with a 20:1 turns ratio, the secondary voltage will be 1/20th of the primary voltage as you have said.
However what you may have missed is the secondary current will also be 20 times the primary current, so substituting a resistive divider would require a square-wave generator with at least 20 times the drive current.

Jim

[electronwaster]

kripton: I will look into that one, and the others you plan to make, thanks for the recommendation.

jim: No - I realised there would be a step up in current, to balance the step down in voltage, hence the second part of my suggested replacement, the voltage follower (current amplifier). I'm guessing that you're suggesting that this won't be sufficient? I suppose I should breadboard/sim the circuit and find out myself, but it would be great to hear your insights.

cheers,
electronwaster

[firewalker]

The transformer doesn't have to be 20:1. The critical part is the ferrite core. I just coiled the secondary to a coil of a compact fluorescent lamp. It works just fine.








Alexander.

[w2aew]

This has been posted here before, but here's a video of the ESR meter that I built in 2006 - no transformer required:

[firewalker]

This has been posted here before, but here's a video of the ESR meter that I built in 2006 - no transformer required:

An idea for your avatar. Use the XYZ monitor you made to display your face!

 

Alexander.

[w2aew]

This has been posted here before, but here's a video of the ESR meter that I built in 2006 - no transformer required:

An idea for your avatar. Use the XYZ monitor you made to display your face!

Ooooh - I like it!  I just might have to do that!

[jimmc]

Sorry I didn't explain a bit more.

If you look at the diagram below, the Ludens ESR meter uses ' Option 1' with Rs =10 ohm (Because of the action of the transformer the source resistance is very low (<<10 ohm)). This gives a theoretical mid-scale reading of 10 ohm and a 91% scale reading of 1 ohm.
Changing to a resistive divider and keeping R5 at 10 ohm changes to 'Modified Option 1' with Rs' = 10 ohm and Rm = 10 ohm, hence Rs = 20 ohm cramping the lower resistance readings even more. Lowering R5 to <<10 ohm would require quite high currents to maintain the drive voltage.

Note that because there is little gain before the rectifier diodes the 'Vm to meter reading' conversion is non linear which does tend to cramp the upper resistance end of the scale giving the 5 ohm mid scale of the meter.



For comparison a meter with Rs =2.5 ohm and a linear AC voltmeter has the scale shown below.



Jim

[T4P]

http://kripton2035.free.fr/digital%20esr/esr-go-russian.html

[free_electron]

here's my 2 cents on all this 'ESR' meter nonsense. Yes, i call it nonsense... as it tells you nothing. ESR is a property of a capacitor that you need to account for when designing circuits that will function under pulse loading or signals with high crest factors. the reason being that the ESR * the current gives you the power dissipation in the capacitor. and you need to accomodate for that. Note the usage of the word 'DESIGNING'.  not 'troubleshooting'.

Now, what do these 'ESR' meters really measure ? They measure how good the electrolyte is... if there is any left at all.

Before things will get clear we need to know how a capacitor is constructed and the basic physics behind it.
Everyone knows a capacitor is built out of two conductive plates separated by a dielectricum ( a non-conducting area ). the physics laws are very simple :
Capacitance is proportional to the surface area of the plates , and inverse proportional to the distance. in other words :

- make the plates bigger and capacitance increases
- place plates closer together and capacitance increases.

this causes problems. bigger plates take up more space... putting them closer together decreases the voltage that can be put across the isolator ( the dielectricum , latin for 'void of electrons').

The classic ceramic of film capacitor construction is exactly that. layers of conductive material stacked on layers of ceramic or film material ( the dielectrics ) with the electrodes broudht outward alternatingly and connected in parallel.

Code: [Select]

       
+----- |
| -----+
+----- |
| -----+
+----- |
| -----+
+----- |
| -----+
|

now there is only so much you can do.... once you hit the 10uF barrier things start getting problematic... and we need another approach. Enter the electrolytic capacitor.
Now this is a totally different animal ! before i go into detail here is a little puzzle.

take a 1uF electrolytic capacitor and take a 1 uf film capacitor. peel them apart and unroll them... you will end up with a lot more material in the film capacitor than in the electrolytic capacitor. here is the strange bit : we know larger plates means more capacitance... we know plates placed closer together means more capacitance.
if you look at the small plates in the electrolytic capacitor and the thcikness of the paper isolator between them ( very thick compared to the tin film in the classic capacitor ) this seems to contradict the laws of phyics ! what is going on ?

well, we need to look at a molecular level... i'll come back on that in a moment. Let's first take a look at the electrolyte. What is electrolyte ? it is a chemical compound ( closely held secretive by capacitor manufacturers... every manufacturere has their own 'mix'. remeber the scores of motherboards with exploding ping-pang-wong capacitors ? that was a classic case of a 'stolen' chemistry. Mitsubishi chemicals of japan had a new electrolyte that was long term stable and offered better perfomance than the classic ones (lower esr and such). To protect this new chmistry the full formula was never avaialble. it was bits and pieces. Well guess what : part of the formula was stolen and leaked out.... some shady manufacturers started making caps based on this 'half' formula... the thing they dodn;t know was that the part missing in the formula was the part that made the electrolyte long-term stable.. sure it performed fine, but it degraded after a few months... conductivity went down , internal dissipation went up and the capacitors boiled themselves dry .. with exploding caps and failing computer motherboards as a result...

back to electrolyte. here are the properties of electrolyte that are of interest
- it is CONDUCTIVE and it must be . the better the electrolyte the better it conducts !
- it is corrosive ! leaking electrolyte from a acapacitor will eat away copper traces on pcb's. ( for people that have repaired tektronix 24xx scopes and tds400 scopes : they know what i am talking about. these scopes are notorious for a type of SMD electrolytic capacitor that has bad seals around its pins. the reflow soldering damages the seals and the electrolyte leaks over time , eating away the power plane connections with catastrophic failures as results.. )

wait, hang on... electrolyte is conductive ? but we need an isolator to build a cap ? so if we squirt electrolyte between two plates ... where is the isolator ?
well. here is how it works.

when an aluminum-electrolyte capacitor, as that is the correct description: the electrodes ( plates ) are aluminum ( they have to be ) and they use an electrolyte to build the dielectricum, is made it is not a  capacitor at all.

you take a strip of paper ( actually a cotton or synthetic porous weave ) , dip it in electrolyte and place an aluminum foil above it. repaeat this once more. then roll the thing up.
you now do NOT have a capacitor. the magic happens in the next step : a controlled current is sent through this stack. the electrolyte conducts but something else happens. the electrolyt , being an acid, attacks the positive polarized aluminum plate and etches it ... during the etching two things happen : on a microscopic scale the plate becomes very 'rough'. this means the surface area is increasing.... up goes capacitance value. a second thing that happens is that we are growing an oxide layer ( oxygen atoms from the electrolyte bind with aluminum atoms from the plate....) and aluminum oxide is ... a perfect isolator ! it is so good that aluminum oxide is used as the stubstrate material for hybrid circuits..

this process is called 'forming' the capacitor. they start with a low voltage and a very low current. as current decreases they up the voltage.. how long they grow determines the isolation voltage. : the thicker the oxide the higher the voltage it can hold. so the maximum voltage used during the forming determines the max working voltage of the capacitor.  ( essentially they use a constant current source with a max voltage limit. when current has fallen to zero for the set voltage the growing of the oxide is complete. )

electrolytic capacitors that have not been used for a long time develop 'leakage' in the oxide. that is why people that restore old equipment often 're-form' the plates using a current source... ( or sometimes more curudely using a tiny lightbulb... )

So. ESR is actually a property of the ELECTROLYTE. a wet electrolyte aluminum capacitor has one aluminum plate that is now porous and coated in an aluminum oxide layer, that forms the dielectricum, and a wet electrode consisting of the electrolyte itself. the other aluminum plate is only there to contact the electrolyte. so the electrolyte itself is the second plate of the capacitor ! the dielecticum es extremely thin as it is grown on a molecular scale. the plates are very large since they have become puttet on an a microscopic level. thinkf of it as taking an aluminum foil sheet and crumpling it up ... that's what happens. for a fixed volume the surface area shoots up tremendously.

So , how and why do electrolytics go bad :

If the electrolyte can leak out then you are essentially losing the second plate of the capacitor. that plate becomes smaller and capacitance goes down.
mechanical damage can cause this. But also electrical damage. Capacitors are often run at high temperatures and the electrolyte will evaporate.... two things happen : the electrical resistance goes up... since these caps are in switched power supplies and have to deal with pulsed currents you get power dissipation in the electrolyte... causing faster evaporation... causing higher resistance in the electrolyte causing more dissipation ( voltage across the capacitor+ESR is constant... ) so it ends up in a self destructeve spiral. It boils itself dry ...

Or you can apply overvoltage , shoot a hole in the dielectricum and then it is also game over... you now boil the capacitor because the full voltage stands across the electrolytic... and that starts out very low ohmic...

Capacitors in switching power supplies fail because of the dissipation due to pulse currents. the dissipation is ESR * rms of the ulse current. in switcing power supplies the pulse currents are high. much higher than in a 50 Hertz sinewave rectifier.... that is why we have special capacitors built for that purpose.

you can buy cpaacitors 1000uF 16 volt. one will be squat and fat... the other will be a very tall slender tube... the slender tube is for pulse currents. even though the capacitance is still goeverned by plate surface, the electrode length becomes a factor. A capacitor is actually a lumped network of repetitive R-C elements. to get the R down you need a wide electrode... so that is what they are doing in those tall slender capacitors. instead of a narow long ribbon they use a wide shorter ribbon. think of it as using a long road with one lane. if you stack cars end to end they can only move in and out one at a time. ( this is the suqat thick capacitor. ) use the tall slender one and oy have a short 6 line highway. now you move 6 cars in and out simulateously. the road is much shorter ( for the same a mount of cars you can place on the road )

So , back to el-cheapo ESR meters : they don't really meaure the ESR ( as in characterising the electrolytic as done by the people designing the capacitor ) : they give you an indication of how much electrolyte is left....

The correct way : apply ac voltage and look at ac current. calculate away the reactance of the capacitor for the given frequency and you can calculate back the ESR. to do that you need to know the capacitance value. problem with bad capacitors is that that value has gone down too..... since the electrolyte is half gone... half of theplate is missing...
so you need to know the real capacitance. that is done by counting electrons in a given time and measuring the voltage at the capacitor.

so you need two measurements : a charge measurement to determine true capacitor value , and an ac current measurement to calculate back.

[BravoV]

here's my 2 cents on all this 'ESR' meter nonsense. Yes, i call it nonsense... as it tells you nothing. ESR is a property of a capacitor that you need to account for when designing circuits that will function under pulse loading or signals with high crest factors. the reason being that the ESR * the current gives you the power dissipation in the capacitor. and you need to accomodate for that. Note the usage of the word 'DESIGNING'.  not 'troubleshooting'.

You sound very knowledgeable bout these thingy, just a simple noob question if you don't mind, why I successfully "troubleshooted" (not designing though) and improve my voltage ripple output by a huge margin by replacing the caps at the switching psus using this so called "non sense ESR meter" in finding those bad caps ? Are those just my dumb luck if its just a snake oil non sense meter ?

.....<snip>....

So. ESR is actually a property of the ELECTROLYTE. a wet electrolyte aluminum capacitor has one aluminum plate that is now porous and coated in an aluminum oxide layer, that forms the dielectricum, and a wet electrode consisting of the electrolyte itself. the other aluminum plate is only there to contact the electrolyte. so the electrolyte itself is the second plate of the capacitor ! the dielecticum es extremely thin as it is grown on a molecular scale. the plates are very large since they have become puttet on an a microscopic level. thinkf of it as taking an aluminum foil sheet and crumpling it up ... that's what happens. for a fixed volume the surface area shoots up tremendously.

...<snip>...

How about other type like tantalum ? Above long explanation doesn't apply since it has no electrolyte and tantalum can have bad esr too, cause I've experienced in the past finding these rotten tantalum caps using this non sense ESR meter ?

So , back to el-cheapo ESR meters : they don't really meaure the ESR ( as in characterising the electrolytic as done by the people designing the capacitor ) : they give you an indication of how much electrolyte is left....

Whats wrong with this method ? I mean for electronic enthusiast level in order to yank out bad caps from their psu ? Its very unlikely that everyone here are working at big company that has million dollars of measurement instruments to play with.

[free_electron]

here's my 2 cents on all this 'ESR' meter nonsense. Yes, i call it nonsense... as it tells you nothing. ESR is a property of a capacitor that you need to account for when designing circuits that will function under pulse loading or signals with high crest factors. the reason being that the ESR * the current gives you the power dissipation in the capacitor. and you need to accommodate for that. Note the usage of the word 'DESIGNING'.  not 'troubleshooting'.

You sound very knowledgeable bout these thingy, just a simple noob question if you don't mind, why I successfully "troubleshooted" (not designing though) and improve my voltage ripple output by a huge margin by replacing the caps at the switching psus using this so called "non sense ESR meter" in finding those bad caps ? Are those just my dumb luck if its just a snake oil non sense meter ?

.....<snip>....

ESR is actually a property of the ELECTROLYTE. A wet electrolyte aluminum capacitor has one aluminum plate that is now porous and coated in an aluminum oxide layer, that forms the dielectricum, and a wet electrode consisting of the electrolyte itself. The other aluminum plate is only there to contact the electrolyte. So, the electrolyte itself is the second plate of the capacitor ! The dielecticum is extremely thin as it is grown on a molecular scale. the plates are very large since they have become pitted on an a microscopic level. Think of it as taking an aluminum foil sheet and crumpling it up ... that's what happens. for a fixed volume the surface area shoots up tremendously.

...<snip>...

How about other type like tantalum ? Above long explanation doesn't apply since it has no electrolyte and tantalum can have bad esr too, cause I've experienced in the past finding these rotten tantalum caps using this non sense ESR meter ?

So , back to el-cheapo ESR meters : they don't really meaure the ESR ( as in characterising the electrolytic as done by the people designing the capacitor ) : they give you an indication of how much electrolyte is left....

Whats wrong with this method ? I mean for electronic enthusiast level in order to yank out bad caps from their psu ? Its very unlikely that everyone here are working at big company that has million dollars of measurement instruments to play with.

It is not 'snake-oil. Sure those 'esr' meters can tell you if you have a rotten capacitor, but they are NOT measuring what is really the 'ESR".

The 'real' ESR is defined by capacitor chemistry and mechanical construction and is a fixed point for a specific manufacturer, capacitor build / electrolyte used.
The 'hobby ESR' is defined at 'how bad' a capacitor has become because it is drying out , or the electrolyte is degrading ( the chemical composition of electrolytes is susceptible to high temperatures. it chemcially breaks down and this changes its conductivity.. this is why caps are specified for x degreec at y number of hours. a 105 degree c cap at 5000 hours means it will still have its specified value after that amount of time and temperature. beyond that ... no warranties ...)

So , yes you can use those hobby esr meters to find 'bad' capacitors as they will give notably different readings from good ones. Can those meters give you the true ESR ? Like the number you need when doing design ? eh... no ...
They are no snake-oil , they are usefull troubleshooting tools, but i would not call them ESR 'meters'.

As for Tantalum. there are wet tantalum capacitors. Same problem as with wet aluminum electrolytic ( there are dry aluminum caps... ) . Dry tantalums suffer from other problems and yes you can find that with these 'esr' hobbymeters.

I just write these long posts to educate people. There are so many 'perceived' things floating around that are completely incorrect. As a troubleshooter you may not have an interest in them. I post them in the hope that people will evolve from 'blind-troubleshooting' to 'educated-design' and then information like this does become important...

when i design a switching power supply i do need to know what the di/dt is ,and figure out a strategy for my tank capacitor ( the tank capacitor is the energy storage capacitor after the rectifier in a switching power supply. this is the energy 'tank' that the attached electronics will draw current from. the switching regulators jon is to fill the tank very quickly , then sleep while the load draws energy from the tank , then do it all over again. so this tank capacitor sees a large AC current ( charge / discarge ) . have bad ESR in the tank capacitor and the voltage ripple goes up ( ac current * esr resistance = ripple voltage ..... and that is staying within the energy holding limits. if your capacitance is too small you cant even hold enough electrons.... and that's another story.) so , sometimes it is beneficial to put 4 x 10 uf in parallel as opposed to 1 47 uf capacitor. why ? 
let's see :

ESR of the 10uf is 1 ohm.
ESR of the 47 uF is 0.5 ohm...

If i put 5 times 10uf in parallel i also get 50 uf , but i now have 5 times 1 ohm ESR in parallel ... that is 0.2 ohm ! so my 5 times 1 uf wil outperform the one 47uf capacitor !
That is why you see in big switching powersupplies lots of capacitors in parallel. you could place 1 big fat one , but by placing mulitple small ones in parallel you shoot ESR down and this directly reduces you esr induced ripple noise !

See how all this stuff comes together ?
Now , with Ac currents capacitor become very funny elements. they are a cascade of R-C elements and , as frequency increases , the 'subcapacitor' at the end of the cascade doesn't do a  whole lot anymore. So here is another tidbit of information :
A capacitors effective value changes with frequency ! a capacitor at 100Khz may have a totally different value than at 1Mhz. And i'm not talking impedance here but raw capacitance !
combine this capacitance change with the ESr and you get a low pass filter.. so if your switching power supply changes switching frequency ... bingo. another thing to look at.
And it gets worse. they also drift under temperature and even under DC voltage.
A capacitor only has the specified capacitance under 0 volts DC. epsecially the ceramic capacitors are notorious for this. that is why we have classifications like X7R X5R Y5V and Z5U..

A y5V , 10uf , 16 volts capacitor can be 10 uf at 0 volts. but 8 volts on it and you have 3 uf left. put 16 volts on it and you might as well put a 100nF capacitor down ...
Now, you may say 'i am only repairing, not designing' , well ... you need to know what new part to put in ! blindly slapping a component in will give you a good chance for 'repeat business' ... it'll fail again...

The same goes with blindly replacing caps in switching power supplies of tv's dvr's and computers. The forums are full with 'slap a 105 degree panasonic blalba in' ... but nobody explains why , and it may not always be the best option.... a little bit of education helps in these cases. and that's what i'm trying to do: Change peoples behavior from 'replace the 220v lightbulb when its broken' - repeat -repeat .....at infinitum ....  to 'check your line voltage. maybe you are better off placing a 240 volts rated bulb... you know lightbulbs are invers thermal coefiicent. their resistance drops a bit after heat up and stabilizes then. that is why you never should turn off and immediately back on... you create hotspots on the filament that will cause it to fail early. let a bulb cool down first (especially halogens ) or have an inrush current protection. the cooldown creates a natural 'soft start')

And i haven't even touched 'soakage' , electron dispersion, leakage  and ESL of a capacitor yet ....

[SeanB]

My "ESR" meter uses a sine wave generator that uses the Wein bridge lamp current across a resistor as the excitation supply, and an ac millivoltmeter on the other side. All made with 741 opamps, and works well enough to detect failing units. The dual 12V supplies are provided by an old modem power supply that provides 10VAC, with a pair of diodes to ptrovide an unregulated + - 15V rail that feeds a 7812 and a 7912 regulator.

[Ajahn Lambda]

Funny, I'm about to build one of these ka-jiggers myself.  Good timing.   

It is not 'snake-oil. Sure those 'esr' meters can tell you if you have a rotten capacitor, but they are NOT measuring what is really the 'ESR".
(...)
They are no snake-oil , they are usefull troubleshooting tools, but i would not call them ESR 'meters'.
(...)
Dry tantalums suffer from other problems and yes you can find that with these 'esr' hobbymeters.

Much like the dashboard gauges on most Fords then. 


No really!  Beyond being a basic indicator, they're not even close to what I'd call reliable, and their accuracy is a joke non-existent.  The only things they're good for are identifying trends (if they work long enough), and keeping the automotive gauge aftermarket afloat.


For troubleshooting, however, sometimes they ARE helpful, much like the cheap little home-made ESR. . . indicator. . . things. . . ?  For example, when you hear a lot of engine clatter, and the oil pressure gauge goes straight to 'L', it's a valid summation that something serious is occurring.  Further testing will likely be necessary, but at least you've got a place to start.


Another example:  using a compression gauge to determine the overall condition of an engine, and if something is wrong, what parts are likely at fault.  Do I care if the gauge is absolutely accurate?  Not really, especially when I'm trying to determine a relative relationship, which is more important in this circumstance.  As long as it's a consistent (offset) error, the gauge will serve its purpose; if it allows me to read the difference in cylinder pressures down to a resolution of ~10 psi, I can perform a valid test.


On a side note, if they're any of MY measuring tools or instruments, they're calibrated on a regular basis by a certified facility, simply because I want to know that what I'm measuring and recording is accurate to some degree.  If I come back 10 years later, for instance, and redo that compression test on the same engine, I'd like to know that both sets of measurements can be relied upon.  It's much to do with my engineering background, and perhaps a bit of overkill for most people, but from experience, dealing with unreliable data due to shitty instrumentation is a trip through Hell I'd rather avoid if possible.


Much the same, using a tool -- I think that's a better name, instead of 'instrument' -- to indicate a failing or failed capacitor is all right with most guys and girls just trying to figure out whether some component has failed.  It's at least some verification that there is a problem with the component, and it's one that will not be revealed by most low-cost DMMs with a built-in capacitance testing feature.

I just write these long posts to educate people. There are so many 'perceived' things floating around that are completely incorrect. As a troubleshooter you may not have an interest in them. I post them in the hope that people will evolve from 'blind-troubleshooting' to 'educated-design' and then information like this does become important...

I know this, and I really, honestly do appreciate your thoughts on the matter.  It's about as bad with automotive subjects, believe me.  But since electronics and electrical engineering are, in general, sort of 'black magic' in comparison, i.e. everything is unseen and largely conceptual, it's often very difficult to drive home the 'why' of it all.  There is so much math, physics, material science, core engineering concepts, etc. behind it that it's a bit crazy (of us) to expect a regular Joe to fully understand, when all he's trying to do is fix his ka-jigger.  I believe this is largely why we see a separation between the 'technician' and 'engineer' mentalities and/or modi operandi.  In other words, one takes care of the theory, and the other actually builds/services the thing.  Maybe left-brain vs. right-brain, if you will.

Now, you may say 'i am only repairing, not designing' , well ... you need to know what new part to put in ! blindly slapping a component in will give you a good chance for 'repeat business' ... it'll fail again...

Possibly, but for most people, good enough is good enough, especially given the temporal nature of modern electronics in the First World.  Despite that, I am with you; I don't have such a 'typical temporal' relationship to my electronics and devices & instruments I've created.  I take care of all things I own, service them when I need to, and design things to last.

And i haven't even touched 'soakage'. . . leakage. . . .

Sounds like diaper ratings. . . . 

[SeanB]

My industrial capacitor test method involves first discharge the capacitor for a minute with a 56K resistor, then check capacitance with a meter. If it is within 5% then proceed to next step. Apply 250V with an insulation tester, followed by 500V. If it passes ( meter drops to left side of scale) then turn to 1000v range and power it for 30 seconds, and listen if the thing starts ticking and crackling. When finished discharge with the 56k 10W wirewound resistor for a minute. Most new ones pass, and the majority of those that have been in use for under 5 years do so as well. i have some 1960's vintage GE units that passed, but are not going to be used as they both are massive and contain PCB oils.

[jimmc]

OK free-electron, I’m not quite sure the point you are trying to make, but here are my thoughts…

Firstly I totally agree with the idea that the ESR meter is a faultfinding tool and not a design aid, then again I don’t think that I have ever seen it claimed otherwise.


There are a couple of areas that I disagree with…


What is ESR?

I start from the premise that for any real capacitor (under fixed conditions) having a sinusoidal voltage impressed across it, the fundamental component of the current flowing through it may be duplicated, in magnitude and phase, by an equivalent circuit consisting of a perfect capacitor (C) in series with a pure resistance (R).
This resistance is the Equivalent Series Resistance (ESR) of the capacitor.

The value of C takes into account such things as the parasitic inductance of the terminations and electrodes and indeed is negative at frequencies above the self-resonant frequency of the capacitor.

The value of R takes into account all losses in the capacitor such as the resistance of the terminations, electrodes and electrolyte (if present) as well as losses in the dielectric.

Neither C nor R directly corresponds to any physical property of the capacitor.



Does an ESR meter measure ESR?

The ESR meter mentioned in the original post effectively applies a voltage at a frequency of 100kHz across the test terminals via a 10 ohm resistor (Rs) and measures the current through it; so what effect does reactance have?

The meter is not phase sensitive so we are only interested in the magnitude of the total impedance (|Z|) = SQRT ((Rs+R)^2 +X^2)

If a pure resistance R is connected to the meter then |Z| is simply Rs + R and obviously the meter must be scaled to read R.

Now replace the resistor with a perfect 1.6uF capacitor; this has a reactance (X) of  1 ohm at 100kHz.
|Z|= SQRT ((10+0)^2 +1^2) =SQRT(101) =10.0498
This means that the reactance will cause the meter to read high by a worst-case 50mohm

Change to a 16uF capacitor, the reactance drops to 100mohm and
|Z|= SQRT ((10+0)^2 +0.1^2) =SQRT(100.01) =10.0005
Now the reactance will cause the meter to read high by a worst-case 500uohm.

I would suggest that these figures show that, in practice, this type of meter can give an acceptably accurate reading of ESR for any capacitor with an effective capacitance greater than 10uF.

Jim

[G7PSK]

Why is it that we do not see barium titanate capacitors used as they have so many advantages compared with aluminium or tantalum capacitors especially in these days of micro miniaturization of all components and the requirements of high esr.

[free_electron]

OK free-electron, I’m not quite sure the point you are trying to make, but here are my thoughts…

The point i am trying to make is the following
- there is more to capacitors than meets the eye.
- the 'esr' meters are NOT ers meters at all. yes you can find a bad capacitor with them but you are NOT measuring ESR ....

I start from the premise that for any real capacitor (under fixed conditions) having a sinusoidal voltage impressed across it, the fundamental component of the current flowing through it may be duplicated, in magnitude and phase, by an equivalent circuit consisting of a perfect capacitor (C) in series with a pure resistance (R).

The problem is that C in this network is not a constant... c is subject to frequency, dc bias , tempearature and other factors. The fact that you use a 100Khz test frequency has an impact...

losses in the dielectric.

losses in the dielectric are in parallel.... they are not ESR
capacitor model:

Code: [Select]

                   
            Rp
                -/\/\/\-
                     |       |
0---nnnn---/\/\/\/\--+--||---+----0

    ESL     ESR       C



I would suggest that these figures show that, in practice, this type of meter can give an acceptably accurate reading of ESR for any capacitor with an effective capacitance greater than 10uF.

Amend that to : provided esr is larger than an ohm or so... try measuring a 10uf x7r where ESR is in the milliohms range ... Then try the same with an Y5V cap that wildly fluctuates under applied voltage ...

Here is some reading material on how ESR and ESL are measured.

http://www.emcesd.com/tt020100.htm

This technique is used by parameter analysers. there you can start fomr a bias voltage and make a ' step' to see how the capacitor reacts.

[free_electron]

Why is it that we do not see barium titanate capacitors

What do you mean not used ? Ceramic caps are barium titanate. Any capacitor of the X5R, X7R series is barium titanate.
Modern electronics are full of them. Very little tantalum or electrolytics around.
NP0 ( C0G) capacitors are typically titanium dioxide delectric. This has no voltage dependency and no tempco... These are the so called class-I capacitors.
Tio2 cannot be deposited thin enogh to make anyt kind of capacitor above roughly 4700pf ...

X7R X5R and anything that changes less than 15% over its operational temperature range is a class II capacitor.
More than 15% is a class 3. ( Z5U Y5V are examples here )
BaTi can be deposited thin and you can make capacitors up to 100 uf with ease. Above that they become bulky.
For large capacitance values the cost of the material becomes dominant, you can't make them as cheap as wet aluminum electrolytics.

The digital boards in tv's are full off class two caps. almost all decoupling is done with these. they are perfect to scatter all over the board and put them close to the point of load. but since they are not avaialble in large capacitance values ( too expensive ) they are not usable as 'tank' capacitor in switching power supplies.
Actually the are... i have switchers running at 5 ampere that use 4 little 2u2 X7R capacitors as 'tank' .. of course those switchers run at 2MHz+ ... not something you will find in an off-line smpsu.. those are still in the kilohertz range...

[jimmc]

I would recommend that anyone interested in the definition of ESR (Equivalent Series Resistance) reads this Application Note  http://www.low-esr.com/QT_LowESR.pdf

Jim

[free_electron]

That's a pretty good document.

things to take away from it : ESR is ONLY defined for a given frequency ( like i said earlier ). Actually ESR is a curve of resistance vs frequency.
In practice , ESR is only a 'young' term. The most commonly used parameter is D ( dissipation factor ) and this is the parameter used during design of a circuit. This is also what real LCR bridges can give you ( Like Agilents 4263B , and those Quadtech machines ) these machines can do frequency sweeps and have the possibility to apply DC bias to the capacitors under test.

So the 'hobby' meters are merely indicators of a good cap / bad cap type. They are NOT meters. ( which is one of the points i was trying to make ) . That doesn't mean they are bad or 'snake il'. They are very useful tools. But they only work under certain circumstances and with restrictions. This is also clearly stated in the quadtech document : However, when ?C is large at high frequencies, high capacitances or some combination, the actual series resistance can cause the largest part of the total D.

AVX, TDK and murata actually have software avaialble where you select the capacitor and they give you the exact plots and a spice model of the selected capacitor for your selected dcbias and or ac current. These manufacturers go through the trouble to fully characterize the components ( using even more sophisticated equipment than an LCR bridge. Most commonly Agilent 4294A or equivalent... starts at 44000$ ... hardly a 3$ cobbled up 'esr meter' )

AVX's tool is called spiCap and is public : http://www.avx.com/spiapps/spicap/spicap3.exe
Murata's SimSurf is also public : http://www.murata.com/products/design_support/simsurfing/download.html
As is TDK's Seat : http://www.tdk.com/seat.php

Seat ( SElection Assistant TDK) does also ferrite beads , inductors and much more and is a very exhaustive piece of software.


Most capacitors manufacturers have this software available but for some it is under NDA or only available if you are in the 'big boys' league... ( meaning a big corporation that uses tons of components, or a corporation that designs Silicon and releases reference designs for the chips including and verified the BOM with 'preferred parts' to end users.

now, as for this 'in-circuit' measuring... that is real snake-oil. what if you have more than 1 capacitor in parallel ? good luck ! or a big one and a small one...

i have repaired a bunch ( yeah besides design i also fix stuff ) of power supplies in lcd monitors ( samsung mainly ) . they are notorious for busted capacitors post secondary rectifier. we had like 50 of the same monitors all fail within one year. all the same defect : popped capacitors. They are running them much too close to their rated voltage and they run too hot. so i replaced them with a capacitor with double the working voltage and actually a lower capacitance value. I measured the caps temperature with an IR thermometer. The original cap runs almost 100 degrees, my substitue 40 ... of course my substitute costs 3x as much and is double the size. The monitors have now been working for 3+ years 24/7 ( these were, at the time, very expensive 1920x1080 with DVi and YcrCb inputs. They are reference monitors ued in broadcast and are color calibrated.)

If you are repairing your own flat panel : pick a capacitor with a higher working voltage , 105 degree and look for one rated at 5000 hours. it'll never come back. no need for an esr meter.
any capacitor post rectifier in the SMPS is suspect... all this stuff is designed wit the cheapest part to survive the warranty term.. after that : all bets are off...

[T4P]

Don't forget to glue the cap if possible, will stabilize the cap if there wasn't any glue before.
As dave said

"The parts can wiggle themselves off the board under vibration"

[free_electron]

hah! don't get me started on microphonics ...
You don't evenneed vibration. pulse currents trhough BaTi caps cause them to flex... the BaTi material is actually piezoelectric...
Send pulse current and the cap flexes... it can flex itself loose.... or they start singing. audibly ...
The opposite is also true. knock on the cap and it produces voltage.
The caps can pick up vibrations and inject these as electrical noise in a system !
we have that problem in harddisk. the positioning servo filter cannot use baTi based cpacitors we need Class-I as these do not have microphonic effect. the motor vibration is induced in the feedback loop of the positioning. if you have a rattling computer fan this is visible as read errors.... due to servo malfunction.

Filmcapacitors also have the microphonic pickup problem. the membrane is flexible. vibrate the cap and you get voltage out of it ( change the delta between the plates : capacitnce changes and for a given charge on the cap this translates into a voltage delta... so .. no filmcaps either...

there is so much things about components that most people do not know.
Things like voltage rating of a resistor. Noise in a resistor. why thick film is better than thin film in certain circumstances. why an l-cut trimmed part is less stable than an r-cut trimmed part.
Where noise in resistors comes from. not thermal induced noise but things like popcorn noise or flicker noise. how the resistancematerial actually causes the noise.  susumu has special lownoise SMD resistors that use a different allow than most other manufacturers. this reduces noise created by the electron flow. We use those in the servo loops.
i could go on for weeks...

[T4P]

hah! don't get me started on microphonics ...
You don't evenneed vibration. pulse currents trhough BaTi caps cause them to flex... the BaTi material is actually piezoelectric...
Send pulse current and the cap flexes... it can flex itself loose.... or they start singing. audibly ...
The opposite is also true. knock on the cap and it produces voltage.
The caps can pick up vibrations and inject these as electrical noise in a system !
we have that problem in harddisk. the positioning servo filter cannot use baTi based cpacitors we need Class-I as these do not have microphonic effect. the motor vibration is induced in the feedback loop of the positioning. if you have a rattling computer fan this is visible as read errors.... due to servo malfunction.

Filmcapacitors also have the microphonic pickup problem. the membrane is flexible. vibrate the cap and you get voltage out of it ( change the delta between the plates : capacitnce changes and for a given charge on the cap this translates into a voltage delta... so .. no filmcaps either...

there is so much things about components that most people do not know.
Things like voltage rating of a resistor. Noise in a resistor. why thick film is better than thin film in certain circumstances. why an l-cut trimmed part is less stable than an r-cut trimmed part.
Where noise in resistors comes from. not thermal induced noise but things like popcorn noise or flicker noise. how the resistancematerial actually causes the noise.  susumu has special lownoise SMD resistors that use a different allow than most other manufacturers. this reduces noise created by the electron flow. We use those in the servo loops.
i could go on for weeks...

Good read for me ! Thanks electron 
Yeah, all this while i knew about most monolithic caps are piezoelectric

[SeanB]

There are articles about how data centre drives can actually detect people singing in the racks from the increased data error rate in the drives near them, and that making the data centre quieter improves read rate considerably.

[M. András]

wow thanks the info on the topic, how much book covers it like this? i guess the handheld lcr meters arent good for design assistance too

[kripton2035]

handheld lcr meters arent able to measure at 100KHz, thay mainly stop at 10KHz
except the latest models, or the very expensive ones...

[G7PSK]

And in my ignorance I have been using ceramic capacitors and not even known what they are made of. Thanks for all the info, more learn on this thread than from any book on the subject that I have read. I will have to cut a ceramic cap open, I did this to an smd one and all I could see was a black mass and with 40X magnification a few striations in the material.

[jimmc]

It is always difficult to hold a discussion on an internet forum, misunderstandings abound. First let me say that I agree with 99.9% of all you have written, it contains a great deal of useful information not readily available.

However I cannot reconcile the quote below with the universally accepted definition of ESR based on the simple series R-C model as given in the App note I quoted.

losses in the dielectric.

losses in the dielectric are in parallel.... they are not ESR
capacitor model:

Code: [Select]

                   
            Rp
                -/\/\/\-
                     |       |
0---nnnn---/\/\/\/\--+--||---+----0

    ESL     ESR       C



Specifically the resistor labelled 'ESR' above is not what is generally understood as ESR and that the 'usual' ESR does include a portion attributable to losses in the dielectric.
Again I agree that ESR is valid only at the frequency of measurement and tells you nothing about how the capacitor behaves at other frequencies.

Jim

[free_electron]

yes and no... if you look at 'true' parameter analysers they will never measure 'ESR'.
ESR is a 'rug' you sweep losses and other 'fuzzy bits' under in an attempt to simplify things. ESR is just that. 'Equivalent'. in reality there is much more going on , but we simplify it by taking all the 'fuzzy bits' , piling them up and write down an 'equivalent' for it.

 i made a mistake in that diagram. ESR should read Rs and ESL should read Ls. brainfart on my part.

this equivalent model does not tell you everything. and during design phase the ESR is only a rough indicator. there are many other factors that need to be taken into account.

anyway

[T4P]

handheld lcr meters arent able to measure at 100KHz, thay mainly stop at 10KHz
except the latest models, or the very expensive ones...

The Mastech MS5308 is not relatively new in anyway. Few Tonghui's can do 100KHz but they are more expensive then the mastech

[robrenz]

handheld lcr meters arent able to measure at 100KHz, thay mainly stop at 10KHz
except the latest models, or the very expensive ones...

IET DE-5000  does 100kHz $330.00

[kripton2035]

DE-5000 is among the recent ones ...

[jimmc]

Thanks for your patience, I've rescaled my meter appropriately.

Jim

[CampKohler]

Thanks for taking time to write up your juicy comments, f_e. Can't read it all now, but I scraped it into a Wordpad file on my thumbdrive for later reading without having to dig back into the forum.

[typeglob]

Thanks for you posts free_electron, very interesting.

There are articles about how data centre drives can actually detect people singing in the racks from the increased data error rate in the drives near them, and that making the data centre quieter improves read rate considerably.

As for hairy men shouting at disks in a datacenter:

[StubbornGreek]

I've really enjoyed reading through this thread.

[helion]

And just like that, this thread has halted?!!

[donald]

This is ESR meter currently building , --> http://www.users.on.net/~endsodds/esr.htm

I did find postage prices steep for a "pcb" that only weigh approx 200grams or so..

[The Electrician]

yes and no... if you look at 'true' parameter analysers they will never measure 'ESR'.

What would be an example (manufacturer and model number) of a 'true' parameter analyser?

[dr_p]

handheld lcr meters arent able to measure at 100KHz, thay mainly stop at 10KHz
except the latest models, or the very expensive ones...

IET DE-5000  does 100kHz

I  have seen this on many LCR meters (IET DE-5000, Mastech MS5308, Uni-T UT612, CEM 9935, Agilent U1731): the datasheet basically says measuring large capacitors (more or less >200uF) is only done at 100Hz or 120Hz. That's 0.1kHz, just to be clear.

I understand ESR is treated as a secondary parameter so that is basically showing while measuring the capacitor. So for a large electrolytic ESR is only measured at 100Hz?

And if so, isn't that useless, since a switch-mode power supply works at 50-100kHz? 

edit:

I see that the Peak Atlas ESR meter mentions measuring ESR at 100kHz, with no mention of capacitance limitation.

[robrenz]

I just measured a 6000µf cap on the DE-5000  It will only measure the capacitance at 100Hz and 120Hz in auto range.
But when you switch to Rs for ESR you can go to 100kHz. You also get milli ohm resolution once you get to 1kHz.

Rs readings (same as ESR readings in ESR mode)
100Hz = 0.08 ohm
120Hz = 0.07 ohm
1kHz = 0.069 ohm
10kHz = 0.061 ohm
100kHz = 0.066 ohm

[dr_p]

Thanks for that, that's great to know.

But now I'm wondering if all of them do that... I guess it's all up to the main controller chip.

[T4P]

Or firmware because the MS5308 is based on the same platform and chip (NOT A COPY!)

[robrenz]

On paper the DE-5000 specs are better than the MS5308. So that makes me wonder if even though the chipset is the same are there accuracy grades of that chip and is it a higher grade chip in the DE-5000?  The fact that you can get a cal certificate from IET for the DE-5000 tells me its specs are valid and not just hype.