Author Topic: Tantalum caps: Still problematic?  (Read 1161 times)

0 Members and 1 Guest are viewing this topic.

Offline ubbutTopic starter

  • Regular Contributor
  • *
  • Posts: 100
  • Country: de
Tantalum caps: Still problematic?
« on: October 11, 2019, 10:09:44 am »
Hi,

looks like we need to replace 0805 10µF ceramic caps with tantalum caps. (Reasons are outlined below). I'm am aware tantalum have the reputations of failing spectacularly*, but I wonder if this is still the case today? These are the parts we tested our circuit with and intend to use:
https://www.rohm.com/products/tantalum-capacitors/conductive-polymer-capacitors/tcto/tctop/tctop1a106m8r-product

thanks

*and also of having negative impact on the environment

Reasons:

We have a voltage reference, but need half that voltage for bias purposes. This is achieved with a simple resistor divider, followed by a 10µF ceramic (X7R) cap. What could go wrong?
The boards are already manufactured and ready to be assembled. And of course now we realized there is a problem.
It is not the voltage dependency of the caps that causes problems, but their temperature stability.
The board warms up a bit when powered, about 50°C, if I now blow some air on it to cool it down, the voltage at the divider changes for a few seconds and then goes back to its original value. This is of course because the capacitor changes its capacitance which in turn changes the voltage at the divider for a short time.
Electrolytic caps do not show this problem. But these are not available in 0805 (as far as I know).

Not wanting to re-spin the board I tried tantalums and they also do not show the temperature dependency (at least not measurably in this setup)
« Last Edit: October 11, 2019, 10:58:27 am by ubbut »
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 21226
  • Country: gb
  • Numbers, not adjectives
    • Having fun doing more, with less
Re: Tantalum caps: Still problematic?
« Reply #1 on: October 11, 2019, 10:23:48 am »
There are other possible explanations for what you observe.

A decent diagnosis will involve removing adjectives (e.g. "bit") and replacing them with numbers, a schematic, a photo of the relevant part of the board, and preferably an infra-red image of the hot board.

The quick test you should do is to remove the capacitor, and observe if the problem is still there. It may be necessary to use a different (external) source for the half reference bias voltage.

BTW, a 10uF X7R is unlikely to be 10uF when there is a significant voltage across it. Some lose 80-90%(!) of their capacitance when at their rated voltage.
« Last Edit: October 11, 2019, 10:26:35 am by tggzzz »
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Online wraper

  • Supporter
  • ****
  • Posts: 17952
  • Country: lv
Re: Tantalum caps: Still problematic?
« Reply #2 on: October 11, 2019, 10:24:16 am »
Tantalum capacitors (except tantallum polymer) need voltage derating. At least 50%, preferably 70% derating.
Quote
We have a voltage reference, but need half that voltage for bias purposes. This is achieved with a simple resistor divider, followed by a 10µF ceramic (X7R) cap. What could go wrong?
You probably don't need 10uF and much smaller capacitance (with less dependency) would suffice. Effect would be less noticeable also due to lower capacitance as such. Also you could simply reduce resistance of divider.
 

Offline ubbutTopic starter

  • Regular Contributor
  • *
  • Posts: 100
  • Country: de
Re: Tantalum caps: Still problematic?
« Reply #3 on: October 11, 2019, 10:38:08 am »

The quick test you should do is to remove the capacitor, and observe if the problem is still there. It may be necessary to use a different (external) source for the half reference bias voltage.

That has of course been done and the capicitor also replaced with an electrolytic one (as described above). The effect vanishes.

BTW, a 10uF X7R is unlikely to be 10uF when there is a significant voltage across it. Some lose 80-90%(!) of their capacitance when at their rated voltage.

Well aware of this, but the absolute capacitance is not that relevant here. The dynamic capacitance is the problem. Also only 2.5V, so not as much an issue.

Quote from: wraper
You probably don't need 10uF and much smaller capacitance (with less dependency) would suffice. Effect would be less noticeable also due to lower capacitance as such. Also you could simply reduce resistance of divider.
I do, this is to reduce noise (of the resistors themselves).
 

Offline Ian.M

  • Super Contributor
  • ***
  • Posts: 13216
Re: Tantalum caps: Still problematic?
« Reply #4 on: October 11, 2019, 10:45:13 am »
@ubbut,
Tantalums were always problematic* if run close to their rated voltage or from a low impedance high current power rail, or if passing high pulsed currents.   Provided you've chosen one with a rated voltage of *AT* *LEAST* double (and preferably three times) your max working voltage, none of those conditions apply in your circuit.

N.B. your Mouser.de link above is was borked - it appears to contain a session ID and stripping that still wasn't valid.
However searching for the part number found it on the manufacturer's website:
https://www.rohm.com/products/tantalum-capacitors/conductive-polymer-capacitors/tcto/tctop/tctop1a106m8r-product
and the datasheet and faq are particularly helpful.

* with a significant risk of loss of magic smoke and unintentional illumination!   >:D  :popcorn:
« Last Edit: October 11, 2019, 11:14:18 am by Ian.M »
 
The following users thanked this post: ubbut

Offline ubbutTopic starter

  • Regular Contributor
  • *
  • Posts: 100
  • Country: de
Re: Tantalum caps: Still problematic?
« Reply #5 on: October 11, 2019, 10:57:58 am »
Thanks, the FAQ is great.

So quick checklist:
Voltage rating is 10V, actual voltage 2.5V. So 4 times --> OK
Impedance is several kOhms --> OK

N.B. your link above is borked - it appears to contain a session ID and stripping that still wasn't valid.

Thanks, fixed the link in my first post
 
The following users thanked this post: Ian.M

Online Gyro

  • Super Contributor
  • ***
  • Posts: 10173
  • Country: gb
Re: Tantalum caps: Still problematic?
« Reply #6 on: October 11, 2019, 11:57:40 am »
Being on the output of a reasonably high resistance voltage divider sounds like a very benign environment for a Tant. They usually fail due to overvoltage or excessive ripple current being demanded of them. I'm sure 2x voltage rating would be fine in this situation and that mainly to minimize leakage current.
Best Regards, Chris
 
The following users thanked this post: ubbut


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf