Decoupling capacitor: Tantalum or Electrolytic?

Hello folks,

I need a decoupling capacitor to catch lower frequency transients in my main power rail. The circuit runs at 5V - 2Amps.

Currently I’m using a Tantalum 100uF-16V TAJC107K016RNJ capacitor. I like it because it is SMD and very compact.

Having said that, I have read that Tantalum capacitors tend to flame out if they are hit by a voltage spike higher than then rating. Sparkfun recomends not to use them if one can avoid them.

What do you recommend? Is it safer to switch to an Electrolytic capacitor instead?



I should probably go for a ceramic, maybe something like: 100uF/6.3V CC1206MKX5R5BB107.

1 Like

@sazp96, I recommend a polarized aluminum SMD capacitor rated at a minimum of 2x the operating voltage. So a 16V rating is a minimum IMO.


I realized from peekay123 answer that my suggested 6.3V might be on the low side if the 100uF is a hard requirement at 5V. Then you should consider at least 10V. But I would still stick with a ceramic for long life with less risk of problems IMO.

Example: GRM32ER61A107ME20L


Thank you guys!

@janne68, thanks for the suggestion. I’m using several ceramic capacitors to decouple high frequency transients. But for the main power rail I want to use a polar capacitor. I’m not 100% sure why, it just seems to be a best practice from looking at other circuits.

@peekay123, I found a 220uF 16V SMD Aluminum capacitor CS1E221M-CRE77. The dimension of the capacitor work well and it in stock at Seeed Studio :slight_smile:

Is there a downside of having a capacitor that is too big? E.g. 47uF vs 220uF?

@sazp96, I usually size based on how much low/mid frequency noise I can expect. In the case of decoupling caps, too large is never a problem except for cost and size. In your case, 220uF is probably overkill but will work just fine IMO.

1 Like

Great. Thanks @peekay123

@sazp96, the very smart @BDub pointed out the following:

If the regulator requires low ESR, then typically you don’t want to use electrolytics… but you can get low ESR electrolytics if you need a lot of capacitance

After a little research, it turns out that there is a MAXIMUM capacitor size for USB 2.0 of 10uF. So you may want to reconsider by using a 4.7uF cap. And don’t forget to be generous with the 0.1uF caps.

1 Like

You are right that it’s a common solution to use polar capacitors for decoupling, but I would not say that it’s “best practice”. I would say that if you should pick one most common reason for failure in devices that include decoupling capacitors I’m pretty sure it will be a electrolytic capictor (I have no statistics to back it up, just my own experience). They fail; it’s just a matter of time.
Reason they are used it mostly due to lower cost and that it will be “good enough”.
And of course when talking about higher capacitance, there’s not much option then electrolytical capacitors.
That said, there is of course nothing really wrong with going with an electrolytic. A good brand with high enough voltage will probably last as long as your device will be in use.

Regarding the size: it’s really hard to tell from what you told in the circuit diagram. But I think that you might get away with a much smaller than 220uF.

I’m a bit slow poeekay123 has written more during my write :smile:

As peekay123 had found there is of course a limit in how much inrush current a USB outlet on a computer can supply and that is specified as hi listed to 10uF. So if youare using it as an USB-port and not running it from a USB charger for supply only then that is the limit I guess :slight_smile:

Very interesting @peekay123.

Here is my dillema. My PCB connects to a Adafruit neopixel LED strip. In there documention, they recommend using a 1,000uF capacitor right at the power source.

Before connecting NeoPixels to any power source, add a large capacitor (1000 µF, 6.3V or higher) across the + and – terminals as shown above.

So it looks like I need the capacitor to be <10uF to comply with USB 2.0 limits, but ~1,000uF to comply with neopixel best practice.

Am I thinking about this correctly?

@sazp96, I’m confused also. The capacitor limit is primarily to control inrush current. However, USB 2.0 was designed prior to having these new multi-amp USB power supplies. If you are using a high quality, high power USB supply then the 220uF will be fine then. If you don’t control the power supply then you may need to rethink your design a bit.


If you are going to use that much current then I guess you will be using a USB Power supply/charger and not connect it to the USB port of a computer.

I would put that 1000uF capacitor as close to the Neopixel string as possible and a smaller on the incomming power.

1 Like

Thank you for the answers.

@janne68, let me give you a bit more background of the solution. It is powered by AC to DC wall power source. It has a USB input, runs on 5Volts and provides 2.1 Amps.

Simplifying things a bit, there are 3 main devices powered by the source. 1) The Photon, 2) 3.5" LCD screen, 3) A power hungry Adafuirt Neopixel LED strip. It is because of the LED strip that we need the 2 Amps.

@peekay123, I do control the power source. I’m using [this one] ( from Seeed Studio.

The good news is that I already have two capacitors in place. The one closest to the power source is C1.

C8 feeds from C1 and is the closest to the LED strip.

So, is the best path forward:

  • C1: 10uF
  • C2: 1,000 uF

Could there be a problem by having a big 1,000 uF capacitor feed from a small one (10uF)?


@sazp96, adding two large caps across (parallel) the 5V supply is like having one big one. The large one is not “feeding” off the small one! The difference is where they are placed to reduce noise.


I support what peekay123 says here. There is no problem having them in parallel but placed at the correct place. Decoupling capacitors should always be placed as close to the current draw it is supposed to decouple to avoid the spreading of “noise” in the powerlines. So the 1000uF goes as close to the NeoPixel as possible.

1 Like

There’s a podcast where the hosts interview a capacitor expert. Short version: tantalum caps won’t explode unless you overheat them severely while soldering. For through hole parts, this is not a problem, but for SMT it can be.

The new materials in tantalum caps aren’t flammable like the old ones. It is not the tantalum that burns, btw.


Great podcast!
The short videos from Dave Jones from eevblog is also worth looking at IMO.
And there are probably more stuff on the page that are good!

Thank you @janne68 and @naikrovek!

The podcast is awesome.