ESR Values for Electrolytic Caps

[madires]

Considering these reference values, my guess is that modern 400V capacitors will actually have an ESR in the range of 1-10Ω. It is a bit strange, since usually I read that proper ESR values should be less than 1Ω, eventually down to 0.1Ω, so I guess this applies only to lower rated voltage capacitors.

An ESR of roughly 3-5 Ohms is typical for electrolytics around 1-10µF at 400V. Haven't you checked the datasheets you've linked?

[m3vuv]

just saying i repaired the trigger circuit on my tek 453 using the 5 transister esr meter from here,the cap tested was a 100pf disk ceramic,the esr readings were nonsense but when testing some known good 100pf ceramics i got some meter deflection although meaningless but at least there was some meter deflection,on the bad cap there was none,it was a sort of go no go test,anyway doing that fixed my scope.

[WindWalker]

Haven't you checked the datasheets you've linked?

Before doing these experiments, I didn't know how to interpret the datasheet information. Now I understand which values are to be expected (in the worst case, since dissipation factor is given as maximum value).

[masster]

See below my contribution to this topic. The tables are based on a very large batch of ESR measurements of new capacitors from various manufacturers.The ESR meter was Peak Atlas ESR70.

[free_electron]

Tables like this are useless.
It heavily depends on manufacturer, family of capacitor, physical construction , fat squat body or slender tall body but most importantly : what electrolyte is being used !

The ESR of a capacitor used to couple an audio signal is irrelevant. You are talking 2 ohms on an input impedance of hundreds of kilo-ohm ...
Where it matters is power supply , especially for capacitors that are working hard : current in current out. Ripple current is important because ripple current x esr gives a voltage ripple but more importantly : a power dissipation ! the damn thing will heat up ,and no matter how good the seal is , the electrolyte will evaporate over time...
So: capacitors after rectifiers are working hard, whether this is a bridge at 50hz or a switcher
Capacitors after a linear regulator are not working so hard but they can still endure pulse currents from the load side.

Don't go randomly swapping out capacitors for "better esr" the esr may be there for a reason ! a low esr cap on a lm317 output is a reason to drive the thing into oscillation ! you need 1 to 2 ohms there.
A low ESR cap on the input of a circuit fed by a power supply over a long cable (like a laptop power adapter ) can have disastrous effects. When you hotplug such a thing ( adapter is powered and you plug the cable in the jack) you will get a massive surge current. Combine that with the inductance in the cable and you will get a pulse voltage that can easily double or triple the supply voltage. if you have a 19 volt supply and your regulators are designed for 24 volts max ... be prepared for them to fry.

The electrolytes today are almost all water based. no more glycol/ DMF or DMA unless you buy specific long life / high temp capacitors.

and for cleaning off leaky capacitors : what to use also heavily depends on what electrolyte was used. electrolyte is not only conductive ( so there is electromigration when voltage is applied. ) but it can also form copper salts when reacting with the traces.
Capacitors have gotten a bad rap because of the stolen chemistry in the early 2000's , but that has long been resolved. Those were the early days of water based electrolytes. the stabiliser was missing and the water was split infot oxygen and hydrogen and the capacitor popped. Not becasue of the explosive mixture but purely because of the pressure build-up. The top just split.

A much more common failure these days is leaking at the electrodes. The rubber seal between the case and pins leaks. This can happen if the capacitor has been mis-treated : too long in the  reflow or wave and board cleaned with an incompatible cleaning agent that damaged the rubber. Or they run hot , the rubber dries out , becomes brittle and cracks.

it all depends on application and the capacitor. you cannot simply go 2200uf 16 volt : that 0.2 ohm esr.

so please : throw that ESR meter in the drawer until you learn how to read datasheets. and NEVER mix manufacturers and families. Oh i want nichicons .. ok , which one ? there's ate least 25 families and another 25 that are obsolete... and they are all different but they all are 2200uf 16 volts...

[masster]

I really don't like your disrespectful tone. You can have an opinion without being a jerk.My reply addressed the original poster message. If you don't like my reply, ignore it. Oh, and you might go for some anger management classes.

[mag_therm]

Thanks masster.
One table like that is worth > a page of hot air.

My contribution is that the ESR increases as temperature reduces.
I  worked on an outdoor tracked vehicle with VS chopper drive.
We had trouble at below -10C as I recall, sorry I have no data about it, but recall testing in the hot/cold chamber

[rsjsouza]

Masster, this is a long forgotten thread that originally had problems by perpetuating an incorrect belief that tables will be relevant without many proper disclaimers. The first analogy from free_electron nine years ago summarizes the same rant today:

that's like making a table giving the weight of a cow .. depends on the subspecies, race ,where it grew up , what it ate , when it ate , if it farted yet ...

Sure this is tongue in cheek, but it gives the proper sense of the usefulness of this data in the light of the immense variations mentioned nine years ago and in today's rant as well. Nothing in the principles of the measurement and the capacitors' inner workings changed at all.

If it is not obvious, I too find ESR tables very limited in relevance due to the exact reasons pointed out years ago and in many fora around the internet, but somehow this is still a sticking point that can demonstrably cause more problems and waste than solve them - especially if someone without experience is reading this.

If you feel you can apply this safely in your environment, then feel free to do it. Unfortunately the boundary conditions are too vast for general applicability.

See below my contribution to this topic. The tables are based on a very large batch of ESR measurements of new capacitors from various manufacturers.The ESR meter was Peak Atlas ESR70.

[free_electron]

I really don't like your disrespectful tone. You can have an opinion without being a jerk.My reply addressed the original poster message. If you don't like my reply, ignore it. Oh, and you might go for some anger management classes.

i am not being disrespectful. i am just stating facts.
You have two tables high quality and normal quality. That in itself is a problem.
- what were your selection criteria ?
- what brands , models were you using ? what is the spread per cell ?

These tables have no frame of reference.
You may have put in a lot of effort but the result is not usable.
If i pick a capacitor that needs comparison : does it fall in your definition of "quality" or is it just "average" ? which table do i use ?

Even assuming that you took the best of the best for the high quality table and the worst of the worst for the normal quality. What if i use one that falls in the middle ?  which table do i use ? do i interpolate ? is the gradient from normal to high quality linear ?

What if i have a very high quality 2200uf 35volt 105 degree 10000hour brand new capacitor and it measures 0.52 ohm. according to your high quality table it should be 0.05 ohm . my capacitor is an order of magnitude worse... should i return it to digikey ? complain to the manufacturer ?
What was the measurement frequency ? i don't have your particular machine, i do have a few others : A DE5000 , a Tonghui TH2821B and a HP 4263A bridge. From what i read that Peak Atlas measures at 50Khz (it's actually a range" but there don't specify what frequency is used for what)
Typically for big capacitors for bulk storage in power systems ( line voltage , not switching ) should be measured at 50, 60 or 100 and 120 hz : single phase rectification eu/us and double phase rectification eu/us. that is why ESR meters have settings for those frequencies.
Components for switching applications should be measured at higher frequencies.

The behavior of ESR for one and the same capacitor is not linear. it decreases (over frequency) very quickly in the first decade, then slows down as frequency keeps increasing.

So i can't even reproduce the readings unless i know what frequency was used for each of the cells. The peak atlas seems to have a range from 40Khz to 100Khz , so it looks like at can "adjust" depending on what it is measuring. That throws  another spanner in the works. Are all the capacitors in the same bin measured at the same frequency or did the meter decide to switch frequencies ?

Your effort no doubt took a lot of time but unfortunately the resulting table is not usable
- what were your selection criteria ?
- what brands , models were you using ? what is the spread between manufacturers/models per cell ?
- what is the frequency they were tested at ? was it locked ?
- how large is your sample size ? (not only between capacitors, but also within a same model. : i tested 3 capacitors 2200uf 16 volt. one teapio , one nichicon and one xicon... so you can see a spread between brands. but what is the spread inside the brand ? test 10 of the same make and model ,test 100 of same make and model )

There is not enough information to have a frame of reference.
Take the datasheet of the manufacturer and model and look it up, then measure it and compare. 

[floobydust]

I use tables like these for a ballpark when testing parts, because failed electrolytics are an order of magnitude off. Who's got time to track down the exact datasheet for the cap when model/series numbers are rarely even on the label.
OP can you recheck table 1, the 10uF 16V and 25V values look high in comparison to parts bigger 10uF 8Ω seems high to me. Would like values for the 10V and 16V down to 1uF filled in.
Graph is from Anatek Blue ESR Meter - log scale is gross to read but graph might be a better format.

[wraper]

Graph is from Anatek Blue ESR Meter - log scale is gross to read but graph might be a better format.

This graph is an utter nonsense. Generalizing<5uF, 5-10uF, 10-100uF and 100-500uF  . Then >500uF is something entirely different. Also why the hell >500uF has entirely different ESR relation to voltage rating?  Which actually is much closer to reality than the rest. then <5/10-100uF ESR/V relations are the same, 5-10/100-500uF are the same too but entirely different from latter  .

[floobydust]

The graphs are the latest Anatek Blue ESR Meter decal. I wasn't saying it's "accurate" or even readable (yuck!) but it's another approach compared to tables, if there were some grid lines and sanity added.
Another table was here https://www.badcaps.net/forum/showthread.php?t=95107&page=3
There, 10uF 16V is 1.6Ω

[wraper]

The graphs are the latest Anatek Blue ESR Meter decal. I wasn't saying it's "accurate" or even readable (yuck!) but it's another approach compared to tables, if there were some grid lines and sanity added.
Another table was here https://www.badcaps.net/forum/showthread.php?t=95107&page=3
There, 10uF 16V is 1.6Ω

I wonder why they couldn't just draw a bunch of identical plots shifted from each other on Y axis instead of drawing this nonsense.

[Vovk_Z]

1uF/16v.
ESR-2Ohm.

2 Ohm is fine for 1 uF cap.

[free_electron]

1uF/16v.
ESR-2Ohm.

2 Ohm is fine for 1 uF cap.

It is an incomplete specification and thus useless.
ESR is an AC parameter that is frequency dependent.

Answering the value without the measurement frequency is like giving directions to someone , but from an unknown start point.

Excuse me sire , do you know how to get to the dentist ?
Yeah, sure you go straight, then left at the crossroads, then two traffic lights over , then right at the intersection.
When the guy is about to leave you tell him , wait , those are the directions from where i live. I'm just a tourist on a guided tour...

2 ohm may NOT be fine for a 1uF cap. It all depends on the application. That little 10uf cap at the output of a lm317 or 1117 needs some ESR to keep the frequency loop stable. If you have too low an esr the regulator will oscillate.

Even modern regulators have this issue : https://www.ti.com/lit/an/slva214a/slva214a.pdf

[David Hess]

It is an incomplete specification and thus useless.
ESR is an AC parameter that is frequency dependent.

ESR is is largely independent of frequency above a minimum frequency.  Here are some measurements I took of a 15 microfarad solid tantalum capacitor that I had handy using my DE-5000, and my ancient ESI 250DA gives the same results at its measurement frequency of 1000 Hz:

100 Hz   0.022 D   2.2 Ohms   15.812 uF   101 Rx
120 Hz   0.024 D   2.0 Ohms   15.784 uF   84.1 Rx
1 kHz   0.106 D   1.08 Ohms   15.507 uF   10.3 Rx
10 kHz   0.911 D   0.96 Ohms   15.081 uF   1.06 Rx
100 kHz   10.42 D   0.91 Ohms   18.15 uF   0.0877 Rx

Rx (reactance) = 1/2PiFC
ESR = D * Rx

What this shows is a relatively constant ESR, which results in an increasing dissipation (D) as the capacitive reactance falls with frequency to approach the ESR.  It is the dissipation which changes with frequency, and not the ESR, but it is changing because the losses in the relatively constant ESR become significant when the capacitive reactance approaches the ESR.

Answering the value without the measurement frequency is like giving directions to someone , but from an unknown start point.

A dedicated ESR meter makes a direct AC scalar measurement of the ESR at a frequency high enough that the capacitive reactance can be ignored.  This is not as good as a vector measurement like I did above, but is close enough for diagnosis in a bulk decoupling application.  This scalar measurement of the ESR meter fails at low values of capacitance where the capacitive reactance is greater than the ESR, which for the 15 microfarad capacitor measured above is somewhere below 100 kHz.

[ebclr]

If the good or bad value to be expected can't be defined by a table. And many capacitors are extremely hard to link to a datasheet and are dependent on temperature, frequencies,, and other factors. Then is my question why measure esr , since the measured value can't be compared to nothing?  I guess is better to have a table as a reference than nothing

[rsjsouza]

1uF/16v.
ESR-2Ohm.

2 Ohm is fine for 1 uF cap.

It is an incomplete specification and thus useless.
ESR is an AC parameter that is frequency dependent.

Indeed.

@1kHz:
Brand new Rubycon 1µF/50V YXM series: D = 0,038; ESR = 6,25Ω; 981,2nF
10 year old Nichicon 1µF/50V PW series: D = 0,044; ESR = 7,52Ω; 947,7nF
20 year old Tantalun 1µ/25V (unknown brand): D = 0,030; ESR = 5,28Ω; 930,9nF
40 year old Siemens 1µF/63V XD series: D = 0,042; ESR = 6,44Ω; 1056,2nF
40 year old Philips 1µ/250V (MKT): D = 0,003; ESR = 0,49Ω; 1031,3nF

@100kHz:
Brand new Rubycon 1µF/50V YXM series: D = 1,520; ESR = 2,93Ω; 827,3nF
10 year old Nichicon 1µF/50V PW series: D = 0,568; ESR = 1,08Ω; 829,3nF
20 year old Tantalun 1µ/25V (unknown brand): D = 1,863; ESR = 3,50Ω; 846,5nF
40 year old Siemens 1µF/63V XD series: D = 1,053; ESR = 1,90Ω; 907,1nF
40 year old Philips 1µ/250V (MKT): D = 0,025; ESR = 0,03Ω; 1007,9nF

All of them except the last one (which is a MKT film capacitor) fail the vast majority of tables.

[free_electron]

15 microfarad solid tantalum capacitor

do that again with an electrolytic ... that's what those tables are for. comparing (solid) tantalium to electrolytics is apples and pears, more like apples and cows. totally different species , totally different behavior.
Every capacitor behaves differently. In an electrolytic Capacitor most of the esr is split between the mechanical construction and the electrolye. The non reactive portion (pure-ohmic)  is in the conductivity of the electrolyte. the rest sits (mostly.. the electrolyte has some too) elsewhere.
Wide body, squat capacitors are made for pulse loads at low frequency : large currents after 50/60hz rectifiers with loads drawing relatively low frequency currents. Tall slender caps are for pulse load at high frequency : switchers and capable of delivering fast transients. They have very low esl.

Every capacitor family and model has its application.

As for not being able to find datasheets : that too is nonsense. capacitors are always marked . i just turned around and grabbed a few random capacitors out of my drawer. i got these at the surplus swapmeet, so i do not know the exact part number

IC - RZSM MM 105C 470uf 35v : IC = Illinois Capacitor owned by cornell-dubilier. went to cde.com (manufacturer , search for rzsm . first hit gave me part number. googled for part number -> datasheet
logo , dont recognise it - google NRSA : second hit NRSA series. open website : NIC components is manufacturer , google for NIC NRSA : first hit : datasheet
logo could be nippon chemicon , could be nichicon , don't know, don't care : 470uf LXF 63V 105c : google LXF. third hit : datasheet : LXF series united chemic-con

The key is those letter codes. That identifies the series.
The same goes for SMD electrolytics there is (almost) always a letter code that lets you identify it. it may take a bit of work but it is doable. there is always service manuals too.

It is not that hard to find the datasheet. Better than trusting a random table made with random measurement of random components , from random lineage , random model, random type with random instruments.
and comparing them to your part with your meter ...

[wraper]

As for not being able to find datasheets : that too is nonsense. capacitors are always marked . i just turned around and grabbed a few random capacitors out of my drawer. i got these at the surplus swapmeet, so i do not know the exact part number

Not really, cheap general purpose Chinese capacitors often do not have series marking. Not to say there is often no website or no datasheets on the website even if you know who made them.

[Vovk_Z]

1uF/16v.
ESR-2Ohm.

2 Ohm is fine for 1 uF cap.

It is an incomplete specification and thus useless.
ESR is an AC parameter that is frequency dependent.

Answering the value without the measurement frequency is like giving directions to someone but from an unknown start point.

We measure ESR typically at 100 kHz, plus or minus some deviation (I mean it is given in datasheets at this frequency). Even if this frequency varies +-(30..40)% from this point, such a result (2 Ohm) still is good.

Of cause, we can measure the frequency at 1 Hz, or at 10000 MHz, but this is a bit, as you say "useless".

[free_electron]

We measure ESR typically at 100 kHz,

that is not correct. it depends on capacitor application. The datasheets will call out the measurement frequency.
That illinois capacitor part calls it out for 120hz . same for NiC. (they give it as dissipation factor wo you can derive ESR from that (provided you know test frequency)


not the scale is logaritmic !

It all depends on the application. if you have a capacitor after a bridge rectifier you are not interested in the esr at 100KHz you want to know it at 100 or 120hz. Thermal effect in a capacitor are mainly caused by the ripple current^2 x esr .
ESR also drifts over temperature.

https://article.murata.com/en-us/article/impedance-esr-frequency-characteristics-in-capacitors

[David Hess]

15 microfarad solid tantalum capacitor

do that again with an electrolytic ... that's what those tables are for. comparing (solid) tantalium to electrolytics is apples and pears, more like apples and cows. totally different species , totally different behavior.

The same pattern appears.  I tested 2 different new parts, and 4 very old parts.  The old Sprague 30D series compares favorably with the new modern low impedance part which explains its popularity in early switching power supplies.

Rx (reactance) = 1/2PiFC
ESR = D * Rx

100 Microfarad 50 Volt Nichicon PM (Low Impedance) with 2036 Date Code

100 Hz   0.024 D   0.40 Ohms   95.18 uF   16.7 Rx
120 Hz   0.025 D   0.36 Ohms   95.01 uF   14.0 Rx
1 kHz   0.071 D   0.122 Ohms   92.25 uF   1.73 Rx
10 kHz   0.547 D   0.097 Ohms   68.4 uF   0.233 Rx
100 kHz   4.86 D   0.087 Ohms   3.59 uF   0.443 Rx

100 Microfarad 25 Volt Nichicon KL (Low Leakage) with 2021 Date Code

100 Hz   0.037 D   0.66 Ohms   90.12 uF   17.7 Rx
120 Hz   0.041 D   0.61 Ohms   89.92 uF   14.8 Rx
1 kHz   0.192 D   0.354 Ohms   84.38 uF   1.89 Rx
10 kHz   1.570 D   0.315 Ohms   23.1 uF   0.689 Rx
100 kHz   23.6 D   0.284 Ohms   0.238 uF   6.69 Rx

#1 100 Microfarad 50 Volt Tektronix House Number with 7844 Date Code (1)

100 Hz   0.047 D   0.58 Ohms   125.46 uF   12.7 Rx
120 Hz   0.049 D   0.52 Ohms   124.80 uF   10.6 Rx
1 kHz   0.185 D   0.246 Ohms   115.33 uF   1.38 Rx
10 kHz   1.211 D   0.187 Ohms   41.6 uF   0.383 Rx
100 kHz   38.2 D   0.152 Ohms   0.277 uF   5.75 Rx

#2 100 Microfarad 50 Volt Tektronix House Number with 7844 Date Code (1)

100 Hz   0.038 D   0.47 Ohms   127.19 uF   12.5 Rx
120 Hz   0.042 D   0.44 Ohms   126.74 uF   10.5 Rx
1 kHz   0.181 D   0.235 Ohms   117.51 uF   1.36 Rx
10 kHz   1.208 D   0.180 Ohms   43.1 uF   0.369 Rx
100 kHz   25.9 D   0.147 Ohms   0.405 uF   3.93 Rx

#3 100 Microfarad 50 Volt Sprague 30D Series with 7421 Date Code

100 Hz   0.024 D   0.30 Ohms   119.48 uF   13.3 Rx
120 Hz   0.026 D   0.28 Ohms   119.16 uF   11.1 Rx
1 kHz   0.122 D   0.166 Ohms   115.10 uF   1.38 Rx
10 kHz   0.954 D   0.139 Ohms   57.0 uF   0.279 Rx
100 kHz   4.98 D   0.115 Ohms   2.64 uF   0.603 Rx

#4 100 Microfarad 50 Volt Sprague 30D Series with 7421 Date Code

100 Hz   0.026 D   0.32 Ohms   127.73 uF   12.5 Rx
120 Hz   0.028 D   0.29 Ohms   127.32 uF   10.4 Rx
1 kHz   0.142 D   0.181 Ohms   122.19 uF   1.30 Rx
10 kHz   1.065 D   0.148 Ohms   53.3 uF   0.299 Rx
100 kHz   6.26 D   0.125 Ohms   1.979 uF   0.804 Rx

[smile]

What about ESR for low value caps like, typical used in ATX PSU?

50V 0.33uf; 50V 0.22uf; 50V 0.1uf; 50V 1uf

[mariush]

Those are most likely not in high frequency paths, where you need low esr, so it doesn't really matter.  But I'd expect 4-10 ohm for such capacitors.

For example, you may see such small capacitor on the input of a controller chip, to smooth out the input voltage before the chip gets its power from an auxiliary winding on the transformer. don't need low esr for that.
Could be used in a comparator or opamp for protections, basic simple things that don't need low esr.