Ceramic capacitor dimension for a buck converter

[jmaja]

Is there any difference between different packages with the same capacitance and voltage rating? Some datasheet recommend or have in their example design very big packages, especially for input. E.g. a 4.7 uF 100 V capacitor is 2220 for one design with just 500 mA output current, while you can find 1210 packages with the same rating and 0805 packages with 50V rating (the converter can only handle 65 V anyway).

Looking at the data from Murata it seems that quite small caps can handle several A RMS current. E.g. this 1206 4.7 uF 50 V capacitor can handle 2-3 A RMS for 20 C temperature rise: http://psearch.murata.com/capacitor/product/GRM319R61H475KA12%23.html

 

[amyk]

You need to consider ESR.

[mikeselectricstuff]

You need to consider ESR.

And dielectric type - needs to be X5R or X7R.  ESR is rarely an issue for ceramics.

Note that over time caps have been getting smaller, so if an appnote specs a particular size, it could just be that was the smallest available when the appnote was written.
Another factor is there can be some sharp bends in the density vs. price curve.

[jmaja]

Yes, I'm only using X5R/X7R. Isn't the temperature rise vs. RMS current essentially the same as ESR? There is typically no ESR value given for ceramic capacitors.

[peter.mitchell]

Mike, i'm just wondering, what are the disadvantages of choosing larger package sizes for these components? (aside from the obvious more board space)

[AndyC_772]

The current / temperature rise characteristic is worth looking at, as it will give you an indication of the ESR (or at least, whether or not the ESR is actually likely to be a problem). Bear in mind that temperature specifications are notoriously vague, so check under what conditions that 20 degree rise is measured - the PCB footprint will have a major effect.

A physically larger part with a higher voltage rating will retain its capacitance better as the voltage increases. Depending on the dielectric, some ceramic caps have a capacitance which is quite a strong function of voltage. Of the common types, C0G/NPO are best, then X5R, then X7R, then there's a very large gap indeed, then Y5V which is almost always best avoided IMHO.

[mikeselectricstuff]

Mike, i'm just wondering, what are the disadvantages of choosing larger package sizes for these components? (aside from the obvious more board space)

You generally want to use the largest part you have room for, although packahes over 1206 can sometimes get more extensive. Space is the only significant disadvantage - this may have knock-on effects like trace loop area.

[jmaja]

For my current poject I need 20 uF and 25 V input. The RMS current can be up to 3-5 A, but normally is 1-2 A.  The concerns are price, size and availibility. Just looking from a local distributor I can get:

1. 2 * 10 uF 0805 for 0.30 euro (for two)
2. 2 * 10 uF 1206 for 0.80 euro
3.  22 uF 1206 for 0.60 euro
4. 2*10 uF 1210 for 1.50 euro
5. 2*10 uF 0603 for 1.80 euro
6. 22 uF 1210 for 1.10 euro
7. 2 * 10 uF 1812 for 3.80 euro

I was originally going for 7. since the datasheet suggest using several low ESR ceramic capasitors. But now I'm tinking why not go with 1.? Also I can get smaller current loop with these, although the mosfets are too big for much of a gain.

[jmaja]

I did some more research and it I found more data: http://ds.murata.co.jp/software/simsurfing/en-us/

A 10 uF 25 V X5R capacitors have quite small difference in heating, 0805 handles 2-3 A RMS and 1812 3-4 A RMS for 20 C temperature rise. But the real difference seems to be in the DC bias. At 20 V DC (typical for this application) 0805 has only about 1 uF left while 1812 still has 6.5 uF! Already at 5 V DC 0805 has dropped to 5.3 uF while 1812 is actually a bit above 10 uF.

So is the 0805 version useless as a buck converter input capacitor due to DC bias?

[sorin]

http://www.maximintegrated.com/app-notes/index.mvp/id/5527

[jmaja]

This is crazy! A 25 V or even 35 V rated capacitor has -90% capacitance already at 20 V and this is not shown in anyway in most datasheets. You really have to dig in to find out this. Reading the link by sorin it seems to be a surprise also for more experienced people.

Why so much about the temperature ratings, but nothing about DC bias?

[Memphis]

This is crazy! A 25 V or even 35 V rated capacitor has -90% capacitance already at 20 V and this is not shown in anyway in most datasheets. You really have to dig in to find out this. Reading the link by sorin it seems to be a surprise also for more experienced people.

Why so much about the temperature ratings, but nothing about DC bias?

Wait, you wanna tell me that if i put 10uF 35V of 6 peices into parallel and apply 24VDC on it, it will not have 60uF at sum? But just 6uF??? 

[BravoV]

Wait, you wanna tell me that if i put 10uF 35V of 6 peices into parallel and apply 24VDC on it, it will not have 60uF at sum? But just 6uF??? 

I guess thats why they manufactured something looks like this, if you're running out of 2D area, then build it upward vertically.

[jmaja]

Wait, you wanna tell me that if i put 10uF 35V of 6 peices into parallel and apply 24VDC on it, it will not have 60uF at sum? But just 6uF??? 

YES!

From Murrata selection of  10 uF 25 V X5R capacitors you would get 0.85-5.8 uF/piece thus 5.1-35 uF for 6 pieces at 24 V depending on the package selection. According to many sources there is no need to overrate the voltage thus 25 V is OK for 24 V.  Murrata doesn't supply 10 uF 35 V smaller than 1206, thus you'll get 1.9-6.1 uF/piece at 24 V.

TDK provides 10 uF 35 V X5R as 0805 and 1206. At 24 V you'll get only 0.7 or 0.8 uF/piece thus 4.2 or 4.8 uF for 6 at 24 V! With TDK you are better of with 25 V version of the 1206, which gives you 3 uF at 24 V!

[Memphis]

http://www.maximintegrated.com/app-notes/index.mvp/id/5527

Now was reading the blog about the capacitance dropping by DC voltage. I am f***** up, i need to redesign all my boards   

EDIT: Is ESR changing too?