Choosing a 100uF supply decoupling capacitor

[lukegalea16]

Hi - I tried browsing this forum to find my answer (which I partially did) but I need some additional info.

So, I have a 3V6 battery operated PCB design and the transceiver that I'm using requires a 100uF low ESR (0.7OHM) tantalum capacitor. However the manufacturer is using 2 x 100uF MLCC on his EVK.
Initially I also started testing with the 2 x 100uF MLCC capacitors. However I'd like to move away from this as 100uF is a large value for MLCC (generally more difficult to find good spec parts like AEC-Q200, X7R and overrate voltage like > 10V).

Now I've started looking into tantalum capacitors and settled on some AVX Kyocera and Kemet models. Here are my concerns:

• In-line after my battery I've got a PPTC fuse and P-MOS reverse polarity. After this I've got MLCCs decoupling connected across the supply (I plan on placing the tantalum capacitors in parallel with the supply as well). Will the tantalum get damaged from this in-rush current? I was going to use a 10V-16V one to derate the voltage.
• I've noticed that for a 10V tantalum capacitor, the maximum leakage is 10uA (which is a lot for a battery powered application). Is it typically less or do I have to live with this?
• Can someone elaborate more about the allowable RMS ripple current? It's the maximum current draw from the cap without damaging the cap right?

Some models that I'm currently looking at from Digikey are https://www.digikey.com/short/mdn7qn19.

Thanks!

[Zero999]

A 10V capacitor should be fine for a 3.6V battery. There's overcurrent protection, which is the main thing.

The leakage is the maximum specified at the maximum temperature and voltage ratings. It'll be much, much lower at 3.6V and just above room temperature.

Yes, the ripple current is the maximum continuous AC current rating of the capacitor, but it won't be damaged by shorter, higher current spikes.

[mikerj]

If this isn't a high volume/cost sensitive application then take a look at polymer aluminium electrolytic capacitors; they can have significantly better ESR and leakage values than tantalum.

[kimballa]

To elaborate on zero999's response, the rule of thumb I've heard is to pick a tantalum with a voltage rating roughly double your max expected voltage. That will withstand the inrush current with no further protection measures.

eg: I do a lot of 12V circuitry and a 20V tantalum decoupling capacitor hasn't ever given me issues.

I don't even know if they come as low-rated as 7.2V. (mostly because I haven't looked.) For a 3.6V battery, a 10V tantalum is more than robust enough to not have to worry about it, as long as the ESR meets your spec.

[David Hess]

In-line after my battery I've got a PPTC fuse and P-MOS reverse polarity. After this I've got MLCCs decoupling connected across the supply (I plan on placing the tantalum capacitors in parallel with the supply as well). Will the tantalum get damaged from this in-rush current? I was going to use a 10V-16V one to derate the voltage.

I think it is unlikely that it will be damaged.  Usually this comes up with an even lower impedance source like a big capacitor bank, or large battery.  And derating by 10 or 16 volts should make it immune to surge related failure

But also consider using a polymer aluminum electrolytic capacitor or a polymer solid tantalum capacitor.

I've noticed that for a 10V tantalum capacitor, the maximum leakage is 10uA (which is a lot for a battery powered application). Is it typically less or do I have to live with this?

That is the maximum leakage for testing.  Typically it is much lower.

Can someone elaborate more about the allowable RMS ripple current? It's the maximum current draw from the cap without damaging the cap right?

RMS ripple current relates to power dissipation and self heating.  It is the maximum RMS current that can be continuously drawn through the capacitor while still meeting its operating life specifications.  It is equivalent to the power rating of a resistor.

[SiliconWizard]

Tantalum caps will leak some current. Yes, it'll be less than the rated value in practice, but still may not be negligible depending on your use case, if you design something ultra low power.
One figure I specifically have mind, because I ran into it specifically, was a design with a 22µF/16V tantalum cap, voltage range was 0 - 10V. At around 3V+ we noticed a current leakage of about 500 nA. That's not a lot, but keep that in mind if you are looking at sub-µA "sleep" currents. That should give you an idea - in your case, the leakage should be typically <= 1 µA.

One thing I'd note from your description is having a relatively large capacitance after a PTC. I don't know how often your circuit is bound to be powered on, meaning the large capacitor will get charged and draw a relatively large inrush current. That may "trip" the PTC for short amounts of time at each power-on cycle and degrade its series resistance over time, which in turn will degrade efficiency. Of course if the device is meant to be powered on at all times and will rarely see power cycles, then that's not really an issue. Otherwise you could add some means of limiting the inrush current.

[lukegalea16]

I was quite settled on using a 100uF 16V tantalum however the maximum leakage at room temperature is 0.01CV which is 16uA. Derating this with temperature yields ~7 times more.
I know that running this at 3.6V the tendency will be to have a much lower leakage current (from your feedback) but I still fear that some parts might start turning up with this maximum leakage current (which is a lot for our application).

[Zero999]

I was quite settled on using a 100uF 16V tantalum however the maximum leakage at room temperature is 0.01CV which is 16uA. Derating this with temperature yields ~7 times more.
I know that running this at 3.6V the tendency will be to have a much lower leakage current (from your feedback) but I still fear that some parts might start turning up with this maximum leakage current (which is a lot for our application).

Well no. It'll be 3.6µA, since the voltage is 3.6V.

Is suggest you actually test some capacitors, rather than relying on data sheet figures.

Failing that, then you really can't get much lower leakage than ceramic or film capacitors, so you'll have to stick with the MLCCs.