How to select filter capacitors?

Benj

I am designing a board using the 40 pin DIP prop and would like to know how to properly select the filter capacitors. The prop demo board for example uses a .1uF before the 5v regulator, and 10uF capacitors after the 5v and 3.3v regulators. The Propeller Education Lab Setup says to use 1000uF caps after the regulators and to add a .47uF cap before the 5v regulator if using a wall power supply and not a battery. How do I know what to use?

Mike Green

Read the datasheet for the regulator you intend to use. The manufacturer will generally discuss the issues appropriate to the specific regulator. If you use the same regulator as used on the Demo Board, use the same capacitor values. A lot depends on what you're going to drive from the power supply. 1000uF capacitors are often used when there's a large surge anticipated (like a servo motor).

Timothy D. Swieter

Yes, read the datasheet for the regulator that is going to be used. Also note that it isn't only the capacitor value that is important but also the type of capacitor (ceramic, tantalum, etc). For the LM1086, which is the one used on the Propeller Protoboard, the input to the regulator only needs a ceramic cap if the voltage source is "far away" and the output needs a low ESR tantalum system cap. The datasheet details this. I have gotten away without using a tantalum on the output, but you risk ringing in the circuit, even to the point where it is audible. There was another thread about ringing noises coming from the Propeller Protoboard regulators.

What regulator are you considering for your design?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Timothy D. Swieter, E.I.
www.brilldea.com - Prop Blade, LED Painter, RGB LEDs, uOLED-IOC, eProto for SunSPOT, BitScope
www.tdswieter.com

Benj

Thank you for the info.

I will have to check when I get back on Monday, but I am pretty sure that they are the LM7833 and LM7805.

Sleazy - G

To first elaborate on the·note·of ringing tim brought up.· This happens to be an underdamped harmonic,·seeded by·inputs from·most likely a power-switching process within your power regulator. Some unaccounted inductances in the circuit are also necessary. In performance apps, Youve got an RC circuit at the power bypassing··location,·from what i believe·should be·in almost all·cases a paralleled tantalum / ceramic pair,· Up until just about a month ago, Demoboards and Protoboards came with·one ceramic·cap per power decoupling junction.··However,·Just the last protoboard I recieved· i noticed·tantalums were·soldered-in between the power rail gaps, assisting the ceramic,·however the datasheet nor the silkscreening of the new boards·reflect the change.·The price did go up to 30$

My opinion is using just·a ceramic alone will not provide the efficiency of tantalums with highly efficient switchers,··On the other hand, the tantalums low·ESR can result in ringing from underdamped transfer functions. such a low series resistance/impedance with unaccounted-for inductances forming a 2nd order system, and harmonic·oscillations will follow if underdamped.· If you use the tantalum alone, you will most likely be underdamped.··There is a solution, use a ceramic along side a tantalum.· The pair will exhibit a better ESR together, and·still lessen the chance·of ringing. ·I think this property of hybrid·capacitive sections working better than others spawns from the fact that·the strength of ceramic capacitors in·damping harmonics through heat dissipation·perfectly complements the tantalum's strength in·providing low ESR efficiency, Ones weakness is the others strength.· AND!·you can never make the ESR·higher or the·capacitance lower of the pair by adding more capacitors of either type··since the·Capacitances of·capacitors in parallel add by direct rule of mixture,·but their·combined ESR·adds by inverse rule of mixture - So you could only get more efficient, either way.·You would eventually reach a capacitance·which would cause·the oscillations damped by the·circuit's resistance·to·re-emerge.··This might sound a little confusing,·try to look at it like an optimization problem.·The design process entails selecting a low enough ESR tantalum·which meets·noise rejection standards and which·which also has·enough capacitance to meet your circuit's dc power demands without depletion (this would result in losing voltage regulation),·if an·underdamped harmonic oscillation occurs by luck, then you would slowly substitute small·ceramic capacitors in parallel for·capacitance which the tantalum·would otherwise make up. The·point where you·could see the circuit being damped by the ceramic cap's heat dissipation JUST enough to cease·resonating, then stop.--· If there was no·harmonic oscillation originally after running the circuit with just the tantalum capacitor, you·can work your way back to oscillation by·adding more tantalum capacitors in parallel, thus reducing total·ESR, or just getting another tantalum with less ESR.··If you did things right you should·have a CRITICALLY DAMPED SYSTEM - AKA PERFECT, at least for this application.· That way you get the most efficient operation with the lowest noise.

Another circuit non-intrinsic method of damping would be the snubber circuit.· Most data sheets will mention snubbers, especially·if they are for extremely efficient switchers, for these chips have the most problems with ringing when such problems are not addressed.· The snubbers, which use just a few capacitors,·but are usually chip specific.··It helps to check·datasheets for snubber reccomendations.



Post Edited (Sleazy - G) : 11/28/2008 5:05:27 AM GMT

ElectricAye

Sleazy G,

I didn't originally post this question, but I want to thank you for posting a detailed explanation about the use of paired capacitors. I recently went through a problem with ringing/instability on a high speed amp + comparator circuit, and I found the following to be of interest in solving the problem, especially the section on selection of bypass capacitors:

http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1154,C1009,C1028,P1219,D4138

I found the soldering and wiring examples near the end of the document to be of interest, too. The concept of "air wiring" and bending pins on ICs seemed ridiculous until I used it to salvage a section of circuitry that I accidently bridged with solder and couldn't get clean.

Benj might also find the following to be of use, too:

http://www.linear.com/pc/downloadDocument.do?id=24956

The following explains a little about ESR:
http://www.kemet.com/kemet/web/homepage/kechome.nsf/vapubfiles/F3101_GoldMax.pdf/$file/F3101_GoldMax.pdf

have fun,
Mark

PS. Sleazy G, in your photo, are you just deep in thought or... or are you kissing your electronics?

Post Edited (ElectricAye) : 11/28/2008 3:04:31 PM GMT