Does capacitor size matter?

Go To Last Post
22 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

When you find that capacitors with the same value of capacitance come in packages ranging from the tiny SMT 0402 to those large radial packages, is there a difference? The reason I'm asking is that I'm trying to design my first standalone PCB project... and would like to use the smallest components possible, including things like capacitors. But when I look at commercial products, such as evaluation boards I have worked with in the past, some have very large capacitors in places (such as next to the voltage regulator) even though there are much smaller capacitors in other places. So it's made me wonder if there's some advantage or disadvantage associated with the physical size of a capacitor? What would you recommend for decoupling at the input/output of a voltage regulator? And what's a good value to use at the AREF pin? or AVCC pin? My datasheet speaks many times of placing one there for better noise suppression but does not mention a recommended value? Or package size, etc...?

Thanks,
James

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There are exceptions but in general package size = voltage. Put too much voltage on a cap that isn't rated for it and BANG! Most times ICs like regulators will tell you the value of the cap to put on the input and/or output. The package size you pick will depend on the voltage the cap will see.

Also, material will change the size of the cap. You just won't see huge ceramic caps for example.

Also, electrolytics tend to be large but not always.

In general 0.01uF is a nice decoupling cap. I don't want to scare you off by starting to talk about self resonance and stuff like that. You can go a long way before you have to worry about those kinds of complexity.

My last home project was done with all 0402 size caps and resistors. By far the smallest thing I've done. usually i use 0905 for 90% of my decoupling caps. A surface mount electrolytic will show up near a regulator somewhere too.

Go electric!
Happy electric car owner / builder

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I just got some 0603 caps and I'm amazed at how small they are, I don't even want to think about hand soldering a 0402 (yes, I know many people say it can be done, but I still don't want to try ;'] )

Speaking of large electrolitics I just got couple 50V 1.8mF (1800uF) for switched supply smoothing, 25mm diameter by 50mm long..

Edward

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
A surface mount electrolytic will show up near a regulator somewhere too.

See... why is that? It's that very thing that has me a little confused. If you can get the same value capacitor rated at the same voltage in the 0402 size... than why instead use a huge electrolytic near the regulator instead? Is there a real benefit? Or does it just "feel" safer to see something of that size?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

First point: Electrolityc caps usually have an order of magnitude (or two, or three) over usuall SMD packages non electrolytic (100uF Electr. against 10nF ceramic 0402). This is why usually one found big electrolytic capacitors close to the voltage regulator, since this is the part that needs a lot of energy (voltage x capacity) stored.

Second point: Real capacitors are also inductors (specially electrolytic) and have some resistance, thus they don't react exactly equal in front of consumption peaks. That's the main reason why 'small decoupling' caps are spread all over the board, close to the IC's.

I strongly recommend you to read about 'decoupling caps' in this fantastic forum, and also google a little bit about that. PCB routing is not as easy as many people thing.

Guillem.

Guillem.
"Common sense is the least common of the senses" Anonymous.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jdowns wrote:
Quote:
A surface mount electrolytic will show up near a regulator somewhere too.

See... why is that? It's that very thing that has me a little confused. If you can get the same value capacitor rated at the same voltage in the 0402 size... than why instead use a huge electrolytic near the regulator instead? Is there a real benefit? Or does it just "feel" safer to see something of that size?

I think part of it is tradition. Traditionally the way to get large amounts of uF's cheaply was by using a fat electrolytic. A reason to keep using them is that electrolytics and tantalum capacitors tend to be lossy which helps cut down on ringing/resonance.

On RF designs, I've noted that using large ceramic 'filter' capacitors on the power supply annoyed the snot out of the radio, presumably because of ringing between the capacitor and inductance on the traces to vcc and gnd.

For a non RF (no radio) design it probably doesn't matter, except that some switching power supplies also like having fat filter capacitor with the 'right amount' of impedance.

The way I like to think about filter caps is as an extension of the power supply. All power supplies have a cut off frequency beyond which the supply doesn't regulate. (AKA at 1khz the regulator responds, at 1MHz, unuh, nope) Thats where the filter caps come in, their impedance goes down with frequency, compensating for the regulators inability to respond to high frequency transients.

Hint: For two layer boards I like to make one side a semi-ground plane and 'tack' vcc to ground with filter caps place close to the components power supply pins.

Last Edited: Tue. Jun 5, 2007 - 07:39 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here you can find useful info about power supplies decoupling:

http://www.piclist.com/techref/p...

Gintaras

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It's not the size of the cap, but how you use it..

- Jani

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One issue should be mentioned (based on my own experiences in the past). The smaller the components become the harder they are to read - and SMD ceramic capacitors have no value printed on! So if you eyes start making problems (your arms become too short for normal reading) SMDs are boaring. Manually soldering SMDs while having a cold can become a nice experience...

If your board consists of components that can not be manually soldered (e.g. BGA packages) it is best to stay on SMDs and let a machine place your components. For home business or small lots, board space might count but if nothing of this is counts stay as long as possible on radial components - much easier at the design stage to measure, solder a second one parallel,...

Knut

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@jani:

http://www.karakal.com/client/im...

Quote:

It's not the size of the cap, but how you use it..

Cap Size matters !

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks guys for all your responses!

- James

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No problem James. Don't get to stuck on this. Unless you are getting into analog design or RF you really can't go too wrong. Sprinkle some 0.01uF caps around (on every voltage rail). Toss a 1000pF in parallel with the 0.01uF for good measure on your AREF. Put at least 10uF on the input and output of your regulator (unless you get one of the cool ones that say you don't need it.) And finally, if you are doing your own PCB layout, keep analog ground and digital ground separate. There should be plenty of info out there on PCB layout tips. Google is your friend. Have fun!

Go electric!
Happy electric car owner / builder

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

sgomes wrote:
No problem James. Don't get to stuck on this. Unless you are getting into analog design or RF you really can't go too wrong. Sprinkle some 0.01uF caps around (on every voltage rail). Toss a 1000pF in parallel with the 0.01uF for good measure on your AREF. Put at least 10uF on the input and output of your regulator (unless you get one of the cool ones that say you don't need it.) And finally, if you are doing your own PCB layout, keep analog ground and digital ground separate. There should be plenty of info out there on PCB layout tips. Google is your friend. Have fun!

Now there's something I never took the time to understand. A .01uf cap with a 1000pF in parallel is the same as a .011uf cap. So why no just toss one of those in there? Or something even bigger? I'm sure there's something analog related I'm missing and would love an explanation.
Thanks guys...
-Adam

-Adam
"Please don't judge my God by my inability to follow him" - Chris Mollins
================
www.onecircuit.com
================

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The trick of big electrolytic with a .01 in parallel uses the small cap to short out the hi freq noise that gets past the higher resistance of the big cap. I've seen some 'careless' designs where they forgot any caps on vcc and the board still worked (somehow?) and of course the other extreme... certainly you need at LEAST one cap on VCC!

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here is a reference:

http://hsi.web.cern.ch/HSI/s-lin...

There are many papers about "paralleling of capacitors".
Use the following keywords in a search:

capacitor paralleling decoupling impedance

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Generally, excepting RF, physical size does not matter. Generally, for a given technology (aluminum electrolytic, tantalum electrolytic, film, ceramic, etc), there is a fairly firm proportionality between the volume of the cap and the maximum product C*V (capacitance and voltage rating). This is another way of saying that, for a given technology, energy storage is proportional to volume.

OK, thats the academic stuff. Practically, the small surface-mount Cs are more "ideal" compared to the larger ones with leads. And, as previously suggested, larger capacitance and larger voltage rating generally means bigger cap.

On power supplies for most run-of-the-mill applications (not including motors and other higher current loads), it has been common to put an electrolytic near the regulator (say 10-100uf), a few ceramics (say 0.1uf) near major loads (like micros) and a bunch of smaller ones (say 1nf) as close as possible to each load. More recently with the new high value ceramic caps, a few large ceramics (say 1uf to 4.7uf) are placed near each major load and the 1nf caps are used close to individual loads.

If your voltage regulator is a switcher or LDO, there may be specific capacitance requirements.

All of this does NOT touch on caps used in audio circuits or RF circuits.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The short answer Adam relates to the fact that capacitors are not just capacitors. They are also inductors. At certain frequencies the L and the C can resonate. So paralleling two caps is not the same as one big cap of the same value. The 0.01uF will decouple a certain band of frequencies and just as it is becoming resonant the 1000pF cap will kick in and decouple.

Again, I just want to say this stuff is all in the "noise". (pardon the pun but I just could NOT resist!) Like Bob said, if you did NONE of this your board probably will work just fine. But there will be those days where it won't and that will be a long long long day trying to figure out what is going on. Caps are cheap. Get a salt shaker and fill it with caps...sprinkle them on your next design.

Go electric!
Happy electric car owner / builder

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ajcrm125 wrote:
Now there's something I never took the time to understand. A .01uf cap with a 1000pF in parallel is the same as a .011uf cap. So why no just toss one of those in there?
It all has to do with the inductance of each kind of capacitor. The smaller cap will (tend to) have less inductance, ergo a higher frequency before impedance starts to increase (i.e., it stops behaving like a capacitor). Of course, you're betting that the higher frequency noise will be smaller, too; usually a good bet, but...

Also, the physical distance from the cap to the IC it is protecting comes into play as well, for pretty much the same reason. In this particular case, inductance is your enemy. There are power supply isolation methods where inductance is purposely introduced to isolate two power supply domains, but we won't get into that here.

As others have said, for simple "low" frequency projects, the 0.01 uF bypass cap will work fine.

Stu (IC designer in another life, hope I remember this correctly :wink: )

Engineering seems to boil down to: Cheap. Fast. Good. Choose two. Sometimes choose only one.

Newbie? Be sure to read the thread Newbie? Start here!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, Stu has it right.

If you look at the self-resonant frequency of various caps (even smt in the same package size), the frequency goes up, as you would expect for smaller values. But it goes up more than proportionately. Often, above a few MHz, a 1nf cap will have lower net impedance than a 0.01uf (10nf) cap. Inductance is the difference.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Note that not all capacitors are created equal and you need to consider how and where they are used. Operating voltage and temperature can be important. Small physical size implies a high dielectric constant. A high dielectric constant usually has other disadvantages. For example consider the X7R and Y5V in this data sheet:
http://www.vishay.com/docs/45156/elecdata.pdf
A 50V rated X7R operating at 25V has 95% of its initial capacitance, compare that to a Y5V which is down to 40%. Similar story with temperature. Using a Y5V near rated voltage and high temperature leaves you with practically no capacitance.

Arthur

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jdowns wrote:
Quote:
A surface mount electrolytic will show up near a regulator somewhere too.

See... why is that? It's that very thing that has me a little confused. If you can get the same value capacitor rated at the same voltage in the 0402 size... than why instead use a huge electrolytic near the regulator instead? Is there a real benefit? Or does it just "feel" safer to see something of that size?

In part ... cost. Here at work we use a lot of 1uF, 2.2uF and 4.7uF ceramics usually rated at on 6-10V, but they are used when very low ESR (Equivalent Series Resistance) is called for. In less demanding applications, .1uF or 100nF ceramics are common, we rarely use 10nF as someone else suggested, these are usually rated at 50V in larger sizes of 15-25V in 0402 and 0603 sizes.

Tantalum is better in terms of ESR, and smaller than the equivalent electrolytic - but is more expensive. Electrolytics are larger, available in higher voltages and generally cheaper than tantalum, but have higher ESR.

For ceramics, the choice of dielectric is also very important, as [adebeun2] points out. One LDO (low dropout voltage) regulator we use requires 1uF on the output at any temperature or voltage. So we use a 2.2uF near ts rated voltage to ensure we always have at least 1uF. Read voltage regulator data sheets VERY VERY carefully! Many of the newer LDO's and the like have very specific requirements.

For the engineer I work with, size is important, so we use nothing bigger than 0603, plenty of 0402 and now a lot of 0201 as well. But we also use QFN packaged devices to keep size right down.

Dean

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

Now there's something I never took the time to understand. A .01uf cap with a 1000pF in parallel is the same as a .011uf cap. So why no just toss one of those in there? Or something even bigger? I'm sure there's something analog related I'm missing and would love an explanation.
Thanks guys...
-Adam

Actually a 1000pF is pretty ineffective in parallel with a 10nF for any practical purpose. I would generally use a 100nF with a 1nF in parallel, but on AVCC on many AVR's you should probably also use a 10uH choke in series as well.

The 100nF provides bulk bypassing of the lower frequency noise. But at increasing frequency its self-inductance means that it looks like a lesser capacitor. The 1nF remains more 'capacitive' at higher frequencies than the 100nF does.

Most serious low noise design will also use ferrite bead type RFI-suppressors (essentially broadband chokes) not only on the supply lines but on signal lines leaving the board as well.

Dean