Tantalum Capacitors, what I've learned

[Dreato]

I’ve been an electronics enthusiast for > 50 years, and inhaled a lot of solder flux fumes! Over the years I’ve learned a lot about both passive and active components. Specifically, electrolytic capacitors. I know if you install a conventional electrolytic backwards and apply the rated voltage, they will explode. Like many of you, I found this out the hard way! When installed properly, they have a limited lifetime. In my experience, after about 30-40 years, they can ‘dry out’ or otherwise fail. I’ve always assumed that tantalum electrolytics had a much longer lifetime and were immune to this type of failure. I’ve learned that this is not the case. Full disclosure – I am an electrochemist by training. I am NOT an electrical engineer, and I claim no special knowledge. Some of what I state below may be incorrect. Feel free to make corrections and to relieve me of some of my ignorance!

I assumed that all tantalum electrolytics were ‘dry’, ie: did not contain a liquid electrolyte and were not subject to drying out. This was not true. There are 2 types of tantalum caps, one with dry electrolyte and one with wet electrolyte. I assumed that the wet electrolyte tants would be subject to ‘drying out’ and would have a limited lifetime. This is true. I assumed that the dry electrolyte tants would have a longer lifetime. This is NOT true. My experience has shown me that they have the same 30-40 lifespan that all the other types of electrolytics do.

HP uses both dry and wet electrolytics in their calculators. They cost more that traditional electrolytics, but can pack more capacitance in a smaller space. Tants have a lower series resistance, too. This is important in switching power supplies and digital circuits.
Recently I have been working on some HP25 calculators and replacing the dead ACT chips with the 25E ACT from Panamatik (highly recommended). Even after replacement, the calcs were still dead or intermittent. A quick check showed that the internal switching power supplies were not working properly. I actually found a cracked/broken resistor in one! (replaced and now works 100%), but most often I found bad tantalum capacitors. Rather than replace them one by one, I ordered the full complement of tants from eBay sellers (60uF 10V axial, 22uF 15V axial, 2.2uF 10V radial) and replaced them all. This revived each calculator. In EVERY case, the tants I removed tested good in my capacitance meter. EVERY case! I should note that some HP25C calculators use a different style capacitor (radial instead of axial) on the board, and later woodstocks use a completely different power supply daughterboard, which uses a couple of the old style, cheaper, electrolytics.

In another case, I had 2 HP97s with intermittent or bad card readers. One gave intermittent read errors, the other would stop pulling the cards part way through, both showing ERROR. CLX cleared the error and complete the card feed process. Teenix, a very helpful forum member, suggested checking the tantalum caps on the card reader board. I ordered these (22uF 10V radial, 6.8uF 10V radial, 3.3uF 10V radial) and swapped them all. Both readers started working 100%. Again, all the removed tants tested 100% GOOD on my capacitance meter. Actually, I used 2.2uF caps rather than 3.3uF, since I had them from my HP25 power supply adventure.

Why did these tantalum caps test good but still fail in the circuit? I don’t know. I do know that next time I open a classic or Woodstock series or HP97 calculator, the tants are all gonna be swapped.

Questions, comments or corrections are welcomed.
Best Regards,
Dave

[Garth Wilson]

A Kemet engineer gives a talk about capacitors at https://www.youtube.com/watch?v=ZAbOHFYRFGg . Starting at 22:20, he's talking about tantalum capacitors for the next 18 minutes, mostly about failure modes.

[Dwight Sturrock]

(05-07-2021 04:42 PM)Dreato Wrote:  Why did these tantalum caps test good but still fail in the circuit? I don’t know. I do know that next time I open a classic or Woodstock series or HP97 calculator, the tants are all gonna be swapped.

Questions, comments or corrections are welcomed.
Best Regards,
Dave

I checked several capacitors with my Fluke on capacitance setting, also measuring across the capacitor with another meter. It looks like the Fluke uses a very low AC voltage (or pulsating DC) possibly <10 Hz. An oscilloscope could confirm this.

Meters sometimes give results different than real world scenarios. Sometimes the reason is the difference in voltage between the test equipment and the actual circuit.
Sometimes an oscilloscope can clear up some ambiguity depending on the measurement being made.

[teenix]

There's quite a bit going on for these capacitors when in use.

See some oscilloscope traces at the end of my Classic notes doc.

http://teenix.org/ClassicNotes.pdf

I guess they are under more stress in circuit than in a tester. Maybe like driving an old car on a smooth highway compared to an old corrugated dirt road.

They probably act a bit like my old body. It worked for ages probably with lots of faults, then sooner or later the bugs start to show :-)

Good to see it is working again.

cheers

Tony

[Kostas Kritsilas]

(05-07-2021 04:42 PM)Dreato Wrote:  I’ve been an electronics enthusiast for > 50 years, and inhaled a lot of solder flux fumes! Over the years I’ve learned a lot about both passive and active components. Specifically, electrolytic capacitors. I know if you install a conventional electrolytic backwards and apply the rated voltage, they will explode. Like many of you, I found this out the hard way! When installed properly, they have a limited lifetime. In my experience, after about 30-40 years, they can ‘dry out’ or otherwise fail. I’ve always assumed that tantalum electrolytics had a much longer lifetime and were immune to this type of failure. I’ve learned that this is not the case. Full disclosure – I am an electrochemist by training. I am NOT an electrical engineer, and I claim no special knowledge. Some of what I state below may be incorrect. Feel free to make corrections and to relieve me of some of my ignorance!

I assumed that all tantalum electrolytics were ‘dry’, ie: did not contain a liquid electrolyte and were not subject to drying out. This was not true. There are 2 types of tantalum caps, one with dry electrolyte and one with wet electrolyte. I assumed that the wet electrolyte tants would be subject to ‘drying out’ and would have a limited lifetime. This is true. I assumed that the dry electrolyte tants would have a longer lifetime. This is NOT true. My experience has shown me that they have the same 30-40 lifespan that all the other types of electrolytics do.

HP uses both dry and wet electrolytics in their calculators. They cost more that traditional electrolytics, but can pack more capacitance in a smaller space. Tants have a lower series resistance, too. This is important in switching power supplies and digital circuits.
Recently I have been working on some HP25 calculators and replacing the dead ACT chips with the 25E ACT from Panamatik (highly recommended). Even after replacement, the calcs were still dead or intermittent. A quick check showed that the internal switching power supplies were not working properly. I actually found a cracked/broken resistor in one! (replaced and now works 100%), but most often I found bad tantalum capacitors. Rather than replace them one by one, I ordered the full complement of tants from eBay sellers (60uF 10V axial, 22uF 15V axial, 2.2uF 10V radial) and replaced them all. This revived each calculator. In EVERY case, the tants I removed tested good in my capacitance meter. EVERY case! I should note that some HP25C calculators use a different style capacitor (radial instead of axial) on the board, and later woodstocks use a completely different power supply daughterboard, which uses a couple of the old style, cheaper, electrolytics.

In another case, I had 2 HP97s with intermittent or bad card readers. One gave intermittent read errors, the other would stop pulling the cards part way through, both showing ERROR. CLX cleared the error and complete the card feed process. Teenix, a very helpful forum member, suggested checking the tantalum caps on the card reader board. I ordered these (22uF 10V radial, 6.8uF 10V radial, 3.3uF 10V radial) and swapped them all. Both readers started working 100%. Again, all the removed tants tested 100% GOOD on my capacitance meter. Actually, I used 2.2uF caps rather than 3.3uF, since I had them from my HP25 power supply adventure.

Why did these tantalum caps test good but still fail in the circuit? I don’t know. I do know that next time I open a classic or Woodstock series or HP97 calculator, the tants are all gonna be swapped.

Questions, comments or corrections are welcomed.
Best Regards,
Dave

Capacitance testers do what they say they do, they test capacitance. What they don't do, is test whether a capacitor will work in a circuit. To be more specific, they test at one frequency (usually), and with one applied voltage. That is most often not reflective of what the circuit the capacitor will be used in. What is not really mentioned often, is that varying the frequency and applied voltage (ac or dc) will produce a different capacitance value, and that may be significantly different from what the capacitance meter shows. Most of the lower end capacitance meters also do not test for leakage, which is a pretty good indicator of how a polarized capacitor is doing.

Most tantalum capacitors, whether liquid electrolyte or solid, are spectacularly intolerant of reverse voltage, to a far higher extent than other electrolytic capacitors. While it may take a few seconds for a regular electrolytic capacitor to fail when reverse biased, the failure on a tantalum is instant, and sometimes quite noticeable.

Many times these days, you can find multi-layer surface mount capacitors that are in the lower range of tantalum and electrolytic capacitors (up to about 10u), and are in a smaller form factor, and often at a higher working voltage. You do have to work out the connection type, but with the smaller form factor, this is not a big issue.

[jebem]

I usually start by checking for leakage with a ohmeter.
Then I check for capacitance and ESR.
Capacitance many times looks good but ESR is high, therefore I do not want this cap on a DC-DC converter.
Usually it is more efficient to just swap the old caps with new production ones. Caps, like batteries, suffer performance degradation with passing time, even when not in use.

[AndiGer]

I have a capacitor tester that also gives ESR as a result. But: what is high or what is low ESR?
Didn't find a realistic answer ...

[jebem]

(05-09-2021 07:14 PM)AndiGer Wrote:  I have a capacitor tester that also gives ESR as a result. But: what is high or what is low ESR?
Didn't find a realistic answer ...

No definitive answers, agreed.
The lower ESR, the better.
But ESR value depends on the reading equipment quality. Most units use a fixed frequency for testing, but from a practical point of view, we compare ESR for fresh production batches of new caps for each capacitance value and working voltage and select the best ones for our application.
So there is no absolute reference as I see it.
But the same is valid for the capacitance real value. A lot of say, 10 caps of 100uF nominal value may present deviations like +50% on some of them, and usually the value will be at least 10% higher.
Many vendors used to care and published charts in the good old days.
ESR impact would depend on the circuit design and application type of course.

For small electrolytic cap values, a low ESR would be a value below 0.5 ohm.
Again, it all depend on the circuit requirements.

[CMarangon]

I had purchased some supercaps via internet and they are abou 1 Farad.
They keep a led working for hours.

Worse thing is that it is no reter than 5 volts.

It believe that it will be used, in the future, as batteries.


Sincerely yours,

Carlos- Brazil

(05-11-2021 07:05 AM)jebem Wrote:  

(05-09-2021 07:14 PM)AndiGer Wrote:  I have a capacitor tester that also gives ESR as a result. But: what is high or what is low ESR?
Didn't find a realistic answer ...

No definitive answers, agreed.
The lower ESR, the better.
But ESR value depends on the reading equipment quality. Most units use a fixed frequency for testing, but from a practical point of view, we compare ESR for fresh production batches of new caps for each capacitance value and working voltage and select the best ones for our application.
So there is no absolute reference as I see it.
But the same is valid for the capacitance real value. A lot of say, 10 caps of 100uF nominal value may present deviations like +50% on some of them, and usually the value will be at least 10% higher.
Many vendors used to care and published charts in the good old days.
ESR impact would depend on the circuit design and application type of course.

For small electrolytic cap values, a low ESR would be a value below 0.5 ohm.
Again, it all depend on the circuit requirements.