Electrolytic Capacitor Tolerance Question for Cap Experts

Go To Last Post
42 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello 'Freaks,

I have received some aluminium electrolytic capacitors which are rated at 4.7uF, 20%. When I measure them on my six digit fully calibrated (ie I'm pretty sure this is accurate) LCR Meter they measure from 3.09uF to 3.4uF.

My subcontractor says they're okay because when they reflow them the capacitance increases to just over 3.76 (not sure if they stay high after they cool down). I think he's talking rubbish and they're just plain dud (i.e out of spec).

Can anyone shed any light on this? Are they out of spec, or does reflowing them permanently raise the capacitance?

Appreciate any answers or 'ancient wisdom' going.
Murdo.

This topic has a solution.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

Last Edited: Thu. Nov 17, 2016 - 10:53 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

News to me!

 

What does the manufacturer's datasheet say?

 

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You'd want to talk to the capacitor manufacturer about this. Another trick you might want to try is apply power to the capacitor for 24 hours and re-measure. There's a process called 'forming' that creates the high surface area on the aluminium foil that creates the capacitance. This happens with applied voltage. Temperature might also assist.
If the capacitor is of dubious heritage, compare with a similar device of known brand like Nichicon or Nippon Chemi-Con.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Basically the data sheet says the usual stuff - voltage, esr tables, temp derating, etc but nothing about permanent changes following reflow.
When I tested them in situ, I did notice that freeze spray reduces capacitance and letting them warm up raises it again (ok, not news, but at least that much is normal)

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Do you have any other (same technology) caps to compare with? Older deliveries, same maker, different supply channel. Where they from a reputable supplier/agent or from evilBay?

 

Ross McKenzie ValuSoft Melbourne Australia

Last Edited: Wed. Nov 16, 2016 - 09:57 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks Kartman,
I'll look into the forming in the morning.
They are from Farnell - their inhouse brand Multicomp, which might be good or bad depending on what they bought and branded...but I'd have thought they'd be more obviously within tolerance; I mean these are heading over 30+% off the target.
When I measure the (small) stock of other electrolytics in the lab they are much closer and within reasonable tolerance.
This is causing my product to restart repeatedly and the ripple is killing the processor occasionally.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Valuesoft,
No I have no other exact same values in the lab, but other capacitors (e.g. 10uF s etc) measure much closer within 20%
Bought from Farnell or Element 14, but they are the in house brand Multicomp.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MurdoMcLeod wrote:
measure from 3.09uF to 3.4uF.
Sounds like mismarked 3.3uF to me...

David

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well giving Farnell the benefit of the doubt, you could perhaps suggest to them that they have been wrongly labelled and request a new delivery batch for comparison.

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Contacting Farnell is probably the best idea, but getting replacements is difficult - technically we didn't buy them - the subcontractor did, so it ends up in the usual four way argument: engineering dept, production dept, subcontractor and supplier. With everyone wanting a least effort solution except Engineering who want a sustainable quality driven solution...I hate to think what this would be like if we manufactured in China instead of the U.K.
But contacting Farnell's tech support people might be a good idea - that way they can decide if 35% is way off or not.
Thanks,
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

>>measure from 3.09uF to 3.4uF.
That's what I said but got poo-pooed by production - don't I know what quality processes the suppliers have, much better than ours, etc, etc,
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Your test method may not be to the standard!
https://en.tdk.eu/download/530704/1dfcee1242ded725859f8e450c03ca01/pdf-generaltechnicalinformation.pdf

They say the test is done at 100Hz.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Does your contract/agreement with the subcontractor include their acceptance testing to your spec? If your spec said 4.7uF +/- 20%... hold them to it or at least request proof that they did it. In the end "we" are talking pennies in the actual component cost but multi-pounds in resolving the rejected/damaged product. If the subcontractor wants your future business they should not fob you off with salesman "talk" aka bulls^%t.

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Kartman,
Ooooh! That is a great datasheet. How come I never find this sort of stuff first off when I search google?
My LCR does 100Hz, 120Hz, 1kHz and 10kHz so I tend to test electrolytics at the 100 or 120. My BoM specifications use the esr or df as requirements (much to production depts annoyance) to filter out crap caps - they're more trouble than they're worth. I'll have a read tomorrow but I'm 'fairly' sure my measurement technique is right.

What is interesting is that guys I trust (my own hardware team, the parallel team we share lab space with, and people like Kartman, valuesoft, etc) have not heard of this thermal jump in capacitance after reflow. I smell Rattus norvegicus...
Thanks guys,
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

.. or... taurus excretum

Ross McKenzie ValuSoft Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Valuesoft,
Yes, exactly - I want the sub contractor to sort it out but production dept want an easy, don't ruffle the feathers, life.
I think a conference call with Mr Subby is in order tomorrow.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Stupid question do they have + and - ends ? (if yes how do you make sure that they don't have rev. voltage?)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi sparrow2,
Yes, they are polarised. The black mark is fairly clear on all of them, which usually denotes the cathode on electrolytics.
My LCR's four terminal smt tweezer probes have marked the positive bias and negative bias so I can be sure not to reverse polarise them.
Actually that's a good line of thinking - I will measure them tomorrow using a low level AC so I can check if they are mistakenly marked. Could explain a lot you know...
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Oh, there's a lot of taurus excretum in 21st century engineering!

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Maybe the caps were stored for too long at a too high temperature and became degraded. In that case the leakage current will also be higher than spec.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks El Tangas,
I'll have a look at that tomorrow. Might be a part of the problem.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Wikipedia has quite a story about elco's

https://en.wikipedia.org/wiki/El...

Wikipedia suggests:

"The capacitance value measured at the frequency of 1 kHz is about 10% less than the 100/120 Hz value."

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One thing to consider is that the RLC meter may not measure electrolytics accurately. If the meter does not apply any bias, the value might not be representative, particularly if the test voltage swings both plus and minus from zero.

 

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Paulvdh, ka7ehk,
My LCR meter does low frequency (100 & 120Hz) so that should be okay.
I'll need to check what DC bias is used to test them - not sure if my meter can do that.
Thanks,
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

MurdoMcLeod wrote:
This is causing my product to restart repeatedly and the ripple is killing the processor occasionally.

And this hasn't stopped you sleeping at night worrying about warranty and product lifetime.

Your design is "on the edge" in this aspect - Just 0.7 uF down from minimum spec. and your device stops working, You've not got nearly enough margin there.

 

Aluminium Electrolytic Capacitors generally suffer large degradation in capacitance over time (especially at higher operating temperatures) and is a common failure mode.

Now is your opportunity to do a change note and sort this out by fitting a 10uF or even 22uF.

 

Last Edited: Thu. Nov 17, 2016 - 10:23 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

How close to the rated voltage are you measuring. These sometimes don't exhibit their full capacitance at low voltages.

 

If you don't know my whole story, keep your mouth shut.

If you know my whole story, you're an accomplice. Keep your mouth shut. 

This reply has been marked as the solution. 
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well folks, I know you've all been waiting for the results and here they are...

Firstly I think I now know significantly more about electrolytic caps than i did before this problem, and boy are they complex.

Secondly, I finally got my hands on a quantity of the caps removed from the products this morning. They were measured using the techniques described in the iec60384 standard, and lo and behold they measure between 4.0 and 4.3uF. Other electrolytics in the lab were even closer matches to their expected values.
The key here is not to trust he LCR's auto mode - it decided not to use the frequency I wanted all by itself, but put it in manual mode and force the 100/120Hz test frequency AND turn on the biasing function at the standard levels. After that it's simply a matter of hitting NULL, and measuring. I did some "advanced" testing and they are marked correctly polarity wise.

Thirdly, their ESR was awful - 30 Ohms or more, which was close to their datasheet limit and far worse than the 10uF caps we bought in to replace these ones which were at 10 Ohms or so.

Fourth, sometimes you don't get a definite explanation even when you have a solution. Changing the caps for 10uF ones and all the problems went away - my colleague and our placement student spent all day testing the units in the environment chamber and we got zero false starts and very stable. I suspect a confluence of lowish value caps (as N.Winterbottom pointed out it is right on the edge value wise) and some very high ESR was enough to make a few boards erratic. Possibly some of the other ideas suggested by you all contributed (maybe old stock, poor storage before fitment, the Forming process mentioned above is also very interesting and the possibility of increased leakage current with it).

Solution: we are going to change the BoM to 10uF in future. Is is awkward because the person who designed the product used the physically smallest sizes he could for all the components so we will have to buy the special miniature ones that cost more just to fit this footprint but cost la vie we can't change the pcb so we'll put up with it. There are three versions of this product so I am getting my colleague to roll out the ECN to all of them to avoid it happening again.

Lessons: be liberal with reservoir caps and don't shrink the can size in case you need to go up in value. Capacitors are much more complex than they look. And finally, go manual with your test equipment.

Many thanks everyone for your suggestions. I cannot emphasise how useful that TDK app note is - it explains everything so clearly.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

>>MurdoMcLeod wrote:
This is causing my product to restart repeatedly and the ripple is killing the processor occasionally.
And this hasn't stopped you sleeping at night worrying about warranty and product lifetime.
Your design is "on the edge" in this aspect - Just 0.7 uF down from minimum spec. and your device stops working, You've not got nearly enough margin there.

Aluminium Electrolytic Capacitors generally suffer large degradation in capacitance over time (especially at higher operating temperatures) and is a common failure mode.
Now is your opportunity to do a change note and sort this out by fitting a 10uF or even 22uF.

Yes, I totally agree with you. I normally double or more the reservoir capacitance I need.

To be honest I stopped worrying about this product a long time ago - I didn't design it, and the eejit who did is long gone and best forgotten. As you can tell from my wording this is not the first problem I have had to fix but about the twentieth! I suspect I could have redesigned it from the ground up with all the time I have spent on it over he last few years. Its one of those really annoying ones whereby you fix one problem and then two more show up in the wake...

Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you had no problems, you would've learnt nothing! I'm glad you've found the root cause and gained something in the process. Having fixed lots of computer monitors and power supplies in years gone by, measuring the capacitance is not the best way of determining the health of a capacitor - ESR is. I've had capacitors that the capacitance measures spot on but have been near open circuit when it comes to ESR, so I have a trusty ESR meter that also measures low Ohms. For critical application that have high ripple current ( like switchers), the physically larger caps generally have a longer life. if you're designing a power supply, sometimes it is better to use a number of smaller caps vs one big one - especially in regards to ESR.

If you're feeling bored, you can read up on the evils of tantalum caps. These have some nasty traps for the unwary!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For the capacitance range in question, ESR spec, and tolerance spec as given wouldn't it be prudent to go with a ceramic?  You can easily get devices that are cheaper, smaller, and tighter specifications.  What bit of capacitor knowledge am I missing that wouldn't make this the obvious choice?

Letting the smoke out since 1978

 

 

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Some regulators don't like ceramics due to the low ESR. If doing a design today, I would consider using ceramics over electros. Nevertheless, the Hi K ceramics have their own set of traps for the unwary. Capacitance can be related to applied voltage, piezo resistivity, mechanical issues - cracking and production issues - like cracking and cost as the value gets over around 10uF.

[edit] Whilst talking to a collegue on an unrelated topic, he mentioned that a problem with ceramics is that they don't have any damping like you'd get with a tantalum or electro, so circuits tend to 'sing' or oscillate more easily. This probably explains why certain regulators don't like ceramic output caps.

Last Edited: Fri. Nov 18, 2016 - 05:58 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You need to check the spec of you LRC meter if you can measure electrolites or tantalium caps for that matter.

Reverse polarizing a elco is not good, a tantalium reversed is an interesting thing you best do in a safe environment.

They are more made to measure non polarized caps.

you also need to have a look at the derating tabels, most of the time they are bypassed and in the end you turn out to run into problems when the caps do not work as expected.

 

we recently had some fun with Y5V ceramic caps. if you quickly read the datasheet they will state -20% to +80% tollerance, but if you include supply voltage ( also a derator) and temperature they actually are -80% + 20% on top of the earlier mentioned tolerance. those are interesting things when you produce in a chineese factory were they say it can be 35C and tell us stuff is not working, we get units shipped to us and in our 21C office we see no problems......

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Low Dropout regulators generally need an output cap with some minimum ESR. A few (but increasing number) are stable with very low to zero ESR. This is one specific case where ESR is "A Good Thing". ESR will also limit the maximum inrush current on power-up. On the other hand, ESR will contribute to output ripple in switch-mode converters. In RF power circuits, ESR causes power loss and heating in capacitors, especially in power amplifier output matching networks. 

 

Hopefully, this list helps to demonstrate that one (usually) cannot blindly choose a low ESR cap. It is a parameter that is often as important as value, voltage rating, or temperature coefficient. The degree of importance varies from application to application, circuit to circuit.

 

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One of the problems I encountered was working out what is good or bad esr. None of the graphs or small tables you see printed on esr meters agree - it seems each table is specific to that meter. In the end I went back to df and worked it from there after I had scoped the meter to check the test frequency and sorted that little problem...
The regulator specifies an aluminium electrolytic cap as being required, probably for the stability reasons mentioned above.
My LCR meter is definitely a good instrument, but like all these things yoo need to know how to set it up. The auto mode was going away from the lowest frequency - that was a trap for me. Working out how to use the built in DC biasing function involved a trip to the attic to find the manual...
My polarity test was similar to the tant polarity sensitivity discussed above - bench power supply set near the caps' rated voltage and some really thick magazines to muffle the shot blast...
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In general you should test the cap the way you use it!

Charge to used voltage, discharge through a known resistor for the current peek time, (or until a drop of ? 200mV), and then measure the voltage, and calc the size of the cap.

Next is that what you want/need.

 

And the 30 ohm could be why you get a wrong number , try to take an old and add 20 (22) ohm and see if you LCR meter get the same result. 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
[edit] Whilst talking to a collegue on an unrelated topic, he mentioned that a problem with ceramics is that they don't have any damping like you'd get with a tantalum or electro, so circuits tend to 'sing' or oscillate more easily.

Yeah, sure, but why the edit?

This is exactly the same as low ESR.

You can also call it high Q.

(Yep, according to wikipedia a single capacitor has a Q-factor)

https://en.wikipedia.org/wiki/Q_...

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What test equipment do you use to measure ESR of a cap? 

Any recommendations for such?

 

Jim

 

(Possum Lodge oath) Quando omni flunkus, moritati.

"I thought growing old would take longer"

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Jim,
I have a couple of small hand held test units: one is specifically for in circuit esr testing, and there is a general component tester which is also quite good for transistor testing etc as well.
At work my lab has a five or six digit TTI made LCR-400 unit, with smt tweezers, etc. This unit can give you two parameters at a time e.g. capacitance and esr, or inductance and q-factor. I think it is very good, although I mostly use it to check COG or NPO ceramics for my RF work, and as/when needed at times like this.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

kiObk - this is what I use:

http://www.capacitorlab.com/esr-...

I have the 'original' version from the 90's. One of the few electronic kits I've built that has been very useful over many years. Devilishly simple, but tricky to understand. There's been plenty written on how it actually works. Simply magic for a single chip micro and a handful of discretes.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

kiObk, Kartman,
That's the standalone esr meter I have as well. I tried making up the Blue esr kit version but I could never get it to work accurately - the EVB one works perfectly every time.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kartman wrote:
kiObk - this is what I use:

Meh, that one has a zilog heart.

 

These are based on an avr and < USD10.

https://www.aliexpress.com/whole...

 

An older version of these has an HD44780 display and even needs less pesetas.

Sourcecode for these older versions is easy to find (github?) I dunno about te versions with the dot matrix display.

These are not high quality / accuracy instruments but they are great fun to use.

These "transistor testers" do much more than you would imagine.

 

You can pop in about any discrete piece of silicon and it will tell you what it is, which connectons are where and some of the most important characteristics.

It can test:

resistors

capacitors (capacitance down to pF & esr for elco's).

transistors

fets

diode's

thyristors

triac's

 

I've done some tests with a small resistor in series with an elco and the change in ESR on the display seems fairly accurate.

Especially for this amasingly simple hardware.

 

For the user manual, watch some youtube video:

https://www.youtube.com/results?...

 

Even the power supply is fun. They use a momentary switch to turn the AVR on, which drives a PNP transistor parrallel to the switch and after a few seconds it turns itself off again.

 

 

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Paulvdh,
I suspect those won't work in circuit - how could they with DC only operation?
The actual ESR meters do work in circuit. I have a similar general component tester but it isn't as good as a proper ESR device.
Murdo.

There are already a million monkeys in front of a million keyboards, and the internet is nothing like Shakespeare!