Capacitors in parallel, different types. - long puzzled me?

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Capacitors in parallel - long puzzled me?

I have seen in loads of commertial circuits, published and derived by examination, with large and small capacitors in parlell, usually diferent types.

I know how to calculate paralell cap values, IE add them, and I apriciate that a circuit requireing a precice value may use multiple caps to acheve it but on the output of a voltage regulator surly this isnt the case.

I suspect it is something to do with the freequency responce and hence impedance, of the diferent types and sizes but havnt found a explanation 'simple enough' to get me started with understanding the concepts involved.

Can anyone give me a leg up so I have some hope of understanding the more complex info that ploliferates on the web and in the relavent data sheets.

I probably should have learned this years ago .... usually just play with values when building which is hardly efficiant or clever.

Thanks guys
Al
 
Last edited:

beenthere

Joined Apr 20, 2004
15,819
If you are referring to the usual practice of placing a relatively large value tantalum or electrolytic plus one or two small value ceramic caps near the input power pin of a device like a microprocessor, the idea is not to trim to some desired value of capacitance, but to more effectively eliminate noise from the supply line.

The larger caps act as a reservoir to keep the voltage input constant while in operation, and the small ones provide a low impedance path to ground to pass off high frequency noise.
 

SgtWookie

Joined Jul 17, 2007
22,230
Sure thing, Al.

The larger values of capacitance (say, >3uF) that you see are generally aluminum electrolytic caps. These are good for low-frequency transient response, but their impedance increases with increasing frequency.

The small-value capacitors you see (typically <= 0.1uF, or 100nF) are meant to be metal poly film or ceramic capacitors. These take care of the high-frequency transients.

Some types of regulators, particularly LDO regulators, are very sensitive to the amount of capacitance on their outputs. If it's too small or too large, they will oscillate.

Even standard regulators like the venerable 78xx series (7805, for example) can and may oscillate if they just have very small capacitances (say, 100pF to 10nF) on their outputs. In cases like these, the idea of the small cap is to swamp any tendency that the regulator may have to oscillate.

Some ICs like the venerable 555 timer require a minimum of two caps, one 1uF and one 0.1uF. Most digital ICs like the 74xx and 4000 series require a minimum of one 0.1uF (100nF) capacitor, metal poly film or ceramic, across their Vcc/Vdd/GND/Vss connections.

If you don't use the capacitor(s) specified in the datasheet, you are begging for trouble.

We very frequently have n00bs that wonder why they run into problems when such bypass caps are omitted; some threads even go on for many pages until someone asks whether proper bypass caps were used.

Even if bypass caps aren't shown on the schematic, they are still required; at the very least one per IC.
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
AHhhhh .... The whole big eletrolytic presenting a large impedance to higher freequencies is what I had completly missed .... I always assumed the bigger the better because it would swamp everything, it probably explains why many of my circuits have issues.

Dood I feel so silly now......

Right away I realise I should have small bypass caps, in adition to the smoothing caps, on my amp circuit for the generator control.

Is there a repository of 'standard' capacator impedance curves somwhere?
Is ther a rule of thumb govning the type of cap for spaciffic applications?
Should I be looking at anything other than impedance curves?

I realise I could check individual data sheets but untill I have a better idea what I am doing that would not be that practical.
I supose what I am realy asking is how can I narrow the field a little when looking for a cap for a spaciffic application.

I hope you realise every time you answer a question I am condemmed to hours of searching / reading to make good use of my new knolage.

You are a star ... Thanks so much
Al
 

SgtWookie

Joined Jul 17, 2007
22,230
Well, there are lots of different types of capacitors, but I try to keep things somewhat simple around here. For complete n00bs, it's difficult enough for them to understand that there is a difference between polarized aluminum electrolytic capacitors, and another type - say metalized poly film or ceramic capacitors.

Metal poly film are actually preferred below around 100V; they have a bit better response than ceramics, but their cost/size goes up quite a bit as their V-rating or capacitance increases. Ceramic caps are generally less expensive, and are easy to find even rated for several thousand volts.

The single 0.1uF/100nF metal poly film or ceramic are the general "rule of thumb" for digital circuits, where currents are generally <20mA. If you're using a device that has higher currents capacity, you need to parallel an aluminum electrolytic with it. It's really not optional; as without it/them your circuit performance will go to a hot place in a handbasket.

Something that many people overlook is that wiring and PCB traces have inductance. The longer and more narrow (or smaller in diameter) the trace or wire is, the more inductance that it has. This can wreak havoc on high-speed signals, particularly square waves, like those on the outputs from comparators, or MOSFET gate drivers, or digital logic.

It is often necessary to slow the rise times of such signals by using resistors in order to avoid excessive "ringing", or unwanted oscillations. For example, a MOSFET gate has a "gate charge" that is usually specified in nC's, or nanocoulombs. It's basically a capacitor. The wiring from the driver to the gate has inductance. Therefore, you have a series LC circuit; when suddenly charged or discharged, it will "ring" for quite a while afterwards at high frequency, causing the MOSFET to heat due to staying in the linear (resistive) region for too long. Adding a small resistor in series with the drive signal right at the MOSFET gate (say, 10 to 22 Ohms) will slow the rise/fall time of the signal enough to eliminate much of the oscillation to begin with, if not snub it entirely.
 

Thread Starter

Dyslexicbloke

Joined Sep 4, 2010
566
Your right about n00bs being confused ... it a permanant state with me.
Sad thing is I'v been tinkering with electronics, for years, but stuck to published circuits and circuits from data sheets. I would use a volrtage reg as directed with the suggested caps, but never realy understood why.

I'v built loads of stuff that works just fine and modified or repaired even more but usually its been case of joing A to B with a couple of passivs to 'match' them.
The problem with that is there are huge holes in what I know, due to lack of experiance, and worse still I probably harber some misconceptions as well.

I have decided, because I have time right now, I want to start getting things right and building the sorts of circuits I'v avoided in the past, which is basically anything that isnt broardly DC if we are talking componant level.

Thanks for the primer, I have some spaciffic things to look for now which will mushroom once I strt searching as these things do.
I get the impression that caps, both use and selection is both a big and important topic, probably more important than I ever gave it credit for previously.
MUCH more reading I think.

'O' and as far as your example comment go's:-

" Therefore, you have a series LC circuit; when suddenly charged or discharged, it will "ring" "

Well I guess that means once I feel a little more comfortable with my understanding of caps I wll have to move on to inductors and resonant circuits.

Last question on this subject ...

Placement of bypass caps important but exactly how should it be done?

Obviously close to what their bypassing is a given but what if the Vcc and Vss pins are on opersit corners of the chip and to compound the issue your working on stripboard.
There isnt a short way to connect both sides of the cap. Which pin Vcc/Vss gets the cap leg, forcing the other side to have a link of some sort.

Al
 

Ghar

Joined Mar 8, 2010
655
There are actually a lot of sources that discourage paralleling different varieties of capacitors.
By doing that you get multiple resonant peaks at every combination of parasitic inductance in series with each capacitor.

If you parallel identical capacitors you reduce the inductance and increase the capacitance without introducing extra resonances.

It's going to depend on the application of course but it's a point to consider.
 

Audioguru

Joined Dec 20, 2007
11,248
In the 60's, Philips published impedance curves of their capacitors in their data books.
I have a few of their data books but since I haven't looked at them for about 30 years I don't know where they are hiding.
 

marshallf3

Joined Jul 26, 2010
2,358
Obviously close to what their bypassing is a given but what if the Vcc and Vss pins are on opersit corners of the chip and to compound the issue your working on stripboard.
There isnt a short way to connect both sides of the cap. Which pin Vcc/Vss gets the cap leg, forcing the other side to have a link of some sort.
l
The distance is generally close enough that it doesnt matter, it becomes wherever placement is most convenient.

I've also been known to use those tiny AXIAL caps to bypass across the ICs by tacking them to the pins n the rear of the IC.

Matter of fact they used to (and probably still do) sell IC sockets with these caps incorporated in them.
 
Top