next topic »

[on:]

Does a switching regulator like the one in the following link need decoupling capacitors like a linear regulator would?

http://www.digikey.com/product-detail/en/OKI-78SR-5%2F1.5-W36-C/811-2196-5-ND/2259781

[Reply #1 on:]

All electronics require decoupling capacitors. Switch mode stuff more than most.

[Reply #2 on:]

what values for input and output? same as linear? electrolytic?

[Reply #3 on:]

Yes I would always use a 0.1uF on both the input and the output with an electrolytic 47uF on the output as well.

[Reply #4 on:]

like a linear regulator would?

Generally speaking, be extremely careful with (output) decoupling on any voltage regulators, LDO and switching mode regulators in particular. You will need to read the datasheet to fully understand what specific requirements are for your particular regulator. (output) decoupling capacitors are inherently unstable in a voltage regulator utilizing negative feedbacks, as such, they should be as small as possible.

[Reply #5 on:]

Recommended Output Filtering
The converter will achieve its rated output ripple and noise with no additional
external capacitor. However, the user may install more external output capacitance
to reduce the ripple even further or for improved dynamic response.
Again, use low-ESR ceramic (Murata GRM32 series) or polymer capacitors.
Initial values of 10 to 47 μF may be tried, either single or multiple capacitors in
parallel. Mount these close to the converter. Measure the output ripple under
your load conditions.
Use only as much capacitance as required to achieve your ripple and noise
objectives. Excessive capacitance can make step load recovery sluggish or
possibly introduce instability. Do not exceed the maximum rated output capacitance
listed in the specifications.

This was on page 11 of the data sheet. Since I don't have an oscilloscope, I'll just have to try it both ways. Would you suggest maybe a 1uF ceramic cap in parallel with a 22uF ceramic cap on both the input and output side? I'll be using it for a couple maxbotix XL EZ 1 ultrasonic sensors. I need a separate power supply for them because they can draw 100mA peak. There will also be a couple 600mA brushed DC motors on the circuit through a motor driver (TB6612FNG), but those will be run through a separate but identical regulator.

[Reply #6 on:]

I would follow the datasheet and use no decoupling cap until you run into problems.

If I had to put something there, i would put a small electrolytic (<4.7uf) there. ceramics, particularly monolithics of that size, are quite expensive and not worth it.

[Reply #7 on:]

SMT Ceramic capacitors are really inexpensive and can be found on Ebay, I've purchased both 1206 and 0805 devices for pennies... assortments of values from 10pF to 47 uF 150 - 200 parts for 8 to 10 dollars US. I'd recommend using a 100 uHy choke similar to the switching inductor (so you can use the full current with minimum, loss)and put caps from both sides to ground, 10uF on the input side and 47uF on the output side of the choke. A simple Pi filter, inexpensive and effective. Tantalum caps can be had as well typically 6 - 10 dollars US for 20 parts and 10, 22 and 47 uF caps are available too, I know, I've recently bought them from a place on Ebay that is located in So. California... about 5 miles from my residence.

Bob

[Reply #8 on:]

Doc,

http://www.digikey.com/scripts/dksearch/dksus.dll

Here are the results from DigiKey for 100mH through hole inductors. In the current column, does that mean max current? For example, the inductor at the following link has 2A rated current. http://www.digikey.com/product-detail/en/5250-RC/M8271-ND/774811 Since I only need a little over 1A, that would be fine for the choke right?
 
Neither the ceramic or the tantalum through hole caps are too expensive for this (around $1 ea). Which do you suggest?

And what kind and value do you suggest for the output side of the switching regulator?  

This project isn't going into outer space so I'll probably be fine without decoupling like dHenry said, but I would like to get it right and learn a little along the way.
Thanks all!

[Reply #9 on:]

Neither the ceramic or the tantalum through hole caps are too expensive for this (around $1 ea). Which do you suggest?

Ceramic is the best type for decoupling (of common capacitor types) because it has the fastest response.

Tantalum is rubbish for decoupling.... and they catch fire when they fail.

[Reply #10 on:]

Tantalum is rubbish for decoupling.... and they catch fire when they fail.

Not true for polymer-based tantalums.  They have ultra low ESR and they are unable to ignite.  It's the MnO2, not the tantalum that is the issue.

However, if lifetime of the circuit isn't a concern there's no need to pay the price for a Ta.