Hi All, I am a bit confused after I did some reading about ceramic capacitors. I am playing with PCB designs by making/trying simple voltage regulators for my projects, and of course part of voltage regulator design is proper selection of input/output caps, filtering caps, etc. Most of that stuff is already in datasheet as suggested values for input, output and other capacitors, some even suggest exact part numbers, etc... Sometimes it leaves things open ended like "larger values can be used if less ripple desired" ......
I do understand that there are different package sizes, different temperature and voltage ratings, different tolerances, with such great capacitors variety available on the market today up to this point capacitor selection for me was very easy, I just narrowed list down to capacitance that I need, temperature rating, package size, voltage rating, and at the end I get what I need.
I am aware that capacitance greatly depend on temperature, frequency, and even DC voltage but I had no idea that DC voltage can almost completely throw capacitance out of the window, for example in the article that I just read (and I also confirmed that by checking few datasheets) if we look at capacitor with for example 4.7uF capacitance and 6.3V rated voltage, turns out that its actual capacitance under such voltage would be about 1uF or even less.... Basically it seems that rated capacitance is nowhere near actual capacitance you get from your capacitor once it is in the circuit. Looks like if my voltage regulator outputs 12V DV voltage and I place 10uF, 16V capacitor I am actually will be getting much less than 10uF out of it, almost like 80% less than that?
More so, apparently package size does matter too, I used to choose smallest package I can handle (which is 0603 in my case) so I can get less esr and all that, but apparently larger packages offer less capacitance degrading for higher voltages.
So in the end, how do I properly choose the capacitor? For example if design sheet suggests to place 22uF output cap, do I just choose 22uF rated capacitor with rated voltage about 30% more than my regulator's output voltage or do I place few 22uF capacitors so that their actual capacitance at my voltage is 22uF summarized?
If the datasheet suggests for example 10uF input cap, 1uF filtering cap, 47uF output cap, does that already take into account that those won't be the actual capacitance values of those capacitors once they placed into circuit?
alexmg2:
in the article that I just read (and I also confirmed that by checking few datasheets) if we look at capacitor with for example 4.7uF capacitance and 6.3V rated voltage, turns out that its actual capacitance under such voltage would be about 1uF or even less.... Basically it seems that rated capacitance is nowhere near actual capacitance you get from your capacitor once it is in the circuit.
Which article said this? Maybe they're talking about practical effects, parasitic effects etc. Or using the capacitor outside the recommended operating conditions.
Here is the article:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527
I don't see where the uncertainty is.
It's a capacitor for Pete's sake.
If you want to measure the capacitance there are several ways. The simplest is measure the RCtime constant with a scope. If you didn't have a scope you can use a comparator to detect the threshold and use millis() to measure the elapsed time. You can use a relay controlled by a digital pin to connect +5V to the input of the RC circuit and measure the time it takes to rise to the Vref which triggers the comparator which switches the input of a digital pin which triggers the millis() elapsed time measurement which gives you the RC time constant which can be used to calculate the capacitance using the value of the resistor in the RC circuit.
The Vref should be set for 66% of R*C. (I think).
Alternately you can .make an oscillator. Personally I think the easiest way is to make a monostable vibrator with a 555 chip and measure the time using an arduino to trigger it and measure the time. I don't see any value in posting your doubts when it is so simple to verify it.
I didn't see you state that you measured the caps.
Your post makes it sound like you just accepted that the values are not correct instead of verifying it.
Just for the record , I am NIT disputing the article.
I am simply suggesting you do some testing.
Is this issue limited to ceramic caps ? Is it only certain caps or all caps ?
It's common knowledge that the capacitance of a ceramic cap decreases when the voltage across it increases.
How much depends on the type of material used (see graph in post#2).
All you can do is use the graph to select the capacitance you need.
Smaller caps, bigger problems. There is no free lunch.
Leo..
alexmg2:
Here is the article:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527
Thanks for that. Yeah..... just glancing at it.... looks like the temperature is sorting of going to extremes or something. Also..... figure 1 says 'Temperature variation vs. DC voltage for select 4.7µF capacitors.' But the plot doesn't actually show temperature variation versus DC voltage........ it's more like capacitance variation versus DC voltage. I'm wondering what the real deal is..... like...how come the graph doesn't show 'temperature'?
Ok..... I see..... they just put the wrong graph title... that's all. Was probably a rush job or casual write-up..... "for select 4.7µF capacitors".... for "selectED".
Somebody talks about mechanisms behind things like this here: click here
These "MLCC" capacitors are those surface mount stuff.
All. caps have a temp rating . So are they saying the rating is wrong ?
There are several sorts of ceramic capacitor, but the sort the article is talking about are those based
on ferroelectric dielectrics (barium titanate and relatives).
These are non-linear, highly temperature dependent, but have vastly greater capacitance for a given
geometry than the other ceramic types. For a 10pF capacitor they wouldn't be used, due to the
very imperfect behaviour, but for 10uF bulk decoupling they are universal since no other solid dielectric
can touch the relative permittivity of ferro-electrics (1000's, not just 5 or 10 for most ceramics).
Repeat, these are ferro-electric materials, which are just as imperfect as ferro-magnetic materials,
ie have non-linear reponse, saturation, hysteresis, large losses and are microphonic. For decoupling
this doesn't matter one bit - and this sort of cap is only used for decoupling...
If you ever see a design for an audio analog filter using capacitors, you'll see plastic-film capacitors are
specified, because ceramic would be hopeless.
At RF frequencies the sorts of ceramic used (we talking about 100pF and below) are fine, so RF analog
filter circuits use ceramics, but not BaTiO4 dielectrics!
Use a low ESR electrolytic capacitor, then.
I haven't seen anything in this post to suggest the OP has any real issue to speak of other than apprehension and concern. He is not posting because some circuit failed , is he ?