Author Topic: Nominal / Rated Capacitance in power converter datasheet  (Read 4223 times)

0 Members and 1 Guest are viewing this topic.

Offline jusacaTopic starter

  • Supporter
  • ****
  • Posts: 69
  • Country: de
Nominal / Rated Capacitance in power converter datasheet
« on: February 17, 2022, 09:20:37 am »
In the datasheet of the TI LMR43620 buck converter on page 29 is a table with typical external component values.
For output capacitance the table says "NOMINAL COUT (RATED CAPACITANCE)". For input capacitance it just says "Cin":
1416475-0

How exactly are these values to be interpreted? Does nominal capacitance mean, that I don't have to look for a cap that has this capacitance value after DC bias derating? So I can simply use the "marketing" value? And the column of the table where it simply says "Cin" would mean that I have to take DC bias derating into account?

Or why would they choose these two different ways of naming the capacitance?
 

Online Vovk_Z

  • Super Contributor
  • ***
  • Posts: 1395
  • Country: ua
Re: Nominal / Rated Capacitance in power converter datasheet
« Reply #1 on: February 17, 2022, 08:27:12 pm »
I understand it as rated capacitance but not real. They really mean 3 caps of 22 uF each  for Cout.
If they wanted to say real capacitance they would write like "50 uF". 
« Last Edit: February 17, 2022, 08:37:12 pm by Vovk_Z »
 

Offline ANTALIFE

  • Frequent Contributor
  • **
  • Posts: 506
  • Country: au
  • ( ͡° ͜ʖ ͡°)
    • Muh Blog
Re: Nominal / Rated Capacitance in power converter datasheet
« Reply #2 on: February 18, 2022, 01:21:43 am »
In the datasheet of the TI LMR43620 buck converter on page 29 is a table with typical external component values.
For output capacitance the table says "NOMINAL COUT (RATED CAPACITANCE)". For input capacitance it just says "Cin":
(Attachment Link)

How exactly are these values to be interpreted? Does nominal capacitance mean, that I don't have to look for a cap that has this capacitance value after DC bias derating? So I can simply use the "marketing" value? And the column of the table where it simply says "Cin" would mean that I have to take DC bias derating into account?

Or why would they choose these two different ways of naming the capacitance?

When I see capacitor values like 22u & 4.7u I would think they are nominal capacitor values (as in ones not taking DC bias into account). So Cout would need x3 22u in parallel, and Cin would need x1 4.7u & x1 100n in parallel

A good way of checking if this is true is seeing how the dev/evaluation board is designed https://www.ti.com/tool/LMR43620MQ3EVM-2M
 
The following users thanked this post: jusaca

Offline jusacaTopic starter

  • Supporter
  • ****
  • Posts: 69
  • Country: de
Re: Nominal / Rated Capacitance in power converter datasheet
« Reply #3 on: February 18, 2022, 08:30:44 am »
A good way of checking if this is true is seeing how the dev/evaluation board is designed
That sounds pretty reasonable, good idea. They use exactly the cap values speced in the datasheet, so they are really just talking about nominal capacitance for Cin as well as Cout.
Thanks.
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 8110
  • Country: fi
Re: Nominal / Rated Capacitance in power converter datasheet
« Reply #4 on: February 18, 2022, 08:42:54 am »
It says it twice: "nominal" and "rated". Both mean the rating the capacitor is sold with. No confusion here. This is better than usual, often the datasheets just talk about "capacitance", but values like "22uF MLCC" are the giveaway: they must be nominal ratings. If it was an actual value, it would say a round number like 20uF for example.

Except, of course, this holds a hidden assumption! As they nowhere tell the actual amount of capacitance required, you need to assume they used some "typical" capacitor available on the market.

But typical high-capacitance MLCCs, that would get used in typical converters, under typical DC bias, vary from maybe 15% to maybe 60% of rated capacitance! It's quite a range of variation. Also note even avoiding just Y5V is not enough. A decently rated X7R part may well drop into 20% under bias. You need to have the curves, or failing that, assume based on ratio of package volume / (Vbias * Cnominal), in other words, "too good to be true" energy density realistically means it is not true.

I would just assume C_actual is 40% of that rating. So 22uF nominal = 9uF actual. So leave the worst offender caps that drop down to 15% under your DC bias on the distributor's shelf, but OTOH, don't waste time trying to find those that perform best-on-the-market or are in massive package sizes, because low inductance (small package) also matters.
« Last Edit: February 18, 2022, 08:44:54 am by Siwastaja »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf