Do low-ESR ceramic capacitors in the uF-range exist?

Question

I am designing a power supply for my board and plan to use an isolated, regulated DC-DC converter. This converter from Murata fits the bill, but the datasheet doesn't provide any guidance on input or output caps.

So I looked at competitors and found this one from TI which is a more feature-rich version of what I'm using. The data sheet shows this example circuit:

The note on the input cap says:

\$ C_{IN} = 2.2\mu F \$ for 5-V input devices... Low ESR, ceramic capacitors are required.

Is there a typo here? On Digi-Key (search URL), the highest-value they have for low-ESR ceramic caps are 3.3nF, and they cost $43 USD each. What gives?

Answer 1

For your purposes, you should read that statement as asserting two things.

1. Ceramic capacitors have low ESR.

2. You can use a ceramic capacitor with this regulator because it is stable with low ESR caps.

Due to a lack of historical perspective, you didn't quite understand what this statement meant. It is not your fault. Once upon a time, most regulators used aluminum electrolytic caps or tantalum caps. Both of these have MUCH higher ESR than ceramic.

But aluminum electrolytics are bulky. And then, for a while, tantalum was very difficult to obtain (the great tantalum shortage). And the ceramic caps became much more available in higher capacitances. So everybody (the IC vendors) started designing their regulators so that they could use ceramics. This was a marketing point, and an important consideration.

"Our regulator is stable with low-ESR ceramics! No need to struggle with procuring unobtanium tantalum caps, or simulate a tantalum by adding a resistor in series with your ceramic cap (something that was also done at times)."

Hope this clears things up. I am sure you can use any X7R or X5R ceramic cap. If you are really worried (if you don't believe me), try to find out if the part has an evaluation board available. If it does, there will be a published BOM for the evaluation board. So you can use the exact same cap TI used on their evaluation board in your design.

Answer 2

As mkeith's excellent and accepted answer states, ESR is not your problem, any ceramic cap will give you the ESR the circuit requires.

However, there is a further gotcha with ceramic capacitors, especially high value ones, like 2.2uF. The ceramics used can have a colossal voltage coefficient of capacitance. A part that measures 2.2uF with 0v across it could drop to 40% of that at its rated voltage, and halve again at elevated temperature. This is rarely shown on short-form data sheets, and only becomes apparent when you dig deep into the manufacturer's detailed data on their website. Different case sizes and voltages will use different dielectrics, so you must look up the data for your specific capacitor part number. If you can't find the data, don't use that manufacturer!

General rules for regulator decoupling capacitors. Get the full data on the exact capacitor you're using. If you can't, go for large case sizes, only use capacitors up to half their rated voltage, avoid dielectrics with a '5' in the designation, and if the application note specifies a minimum capacitance, over-provide by a factor of 2. These are less important for general decoupling, where exact value is rarely a problem. However, for regulator decoupling, the value is often important for stability.

Answer 3

Yes indeed, they exist. I like to use K-SIM from Kemet for evaluating ESR and capacitance change vs. bias when I'm searching for part numbers.

Here's a quick search as an example, a 1206 4.7uF:

This part in 10% tolerance costs $0.46 or $0.126/pc in 2000pc reel. Quite expensive for a capacitor (probably you can relax its specs), but nowhere near those $46 you said.