Hi, my project has reached a point where I'm using more power than the arduino can provide, so I'm having to add an additional 3.3v and 5v regulator. I'm doing "cut 'n' paste" engineering and copying from datasheets and other designs. Most of the designs seem to include a couple of capacitors on the 'input' side of the regulator (in this case a 12 volt supply, a battery). One is a 100 mico farad, the other 0.1 mirco farad, as I understand it, the former is to filter any low frequency noise, the latter for high frequencies.
The designs I'm copying both include these two extra capacitors, since they're both filtering the same supply voltage do I really need both pairs? won't a single pair of 100 and 0.1 micro farad capacitors be adequate?
Also, since the result of two capacitors in parallel is to sum the values together, I'm effectively getting a single 200 micro farad cap aren't I ? so can't I just make do with one capacitor?
Just to make it clear, the design I'm talking about has already been built (with two pairs of capacitors) and appears to work just fine, however my circuit board is getting a bit crowded, and I'm looking for components I can drop. So my questions are...
- Can I get away with just a single pair of 100 and 0.1 micro farad capacitors serving both regulators?
- If not, can I use a single 200 micro farad capacitors to replace the two capacitors in parrallel?
Thanks
Since you are using a battery, my guess is that you could get away with just one set of caps. Is your project something that requires very low noise on the input?
my guess is that you could get away with just one set of caps
that's my guess to (I've snipped the extra caps off, and it still seems to work) However, years of experience has taught me that there's a difference between 'just working' and 'working properly' - I'm hoping there's an experienced electronics engineer out there who may be able to give a definitive answer.
Cheers
Absolutely, use a 200 - 1000- uf "Input Filter" capacitor and a single .1uF cap to filter Input Noise on the PSU Input wiring and 'local' 1uF and 100nF filters placed "Close" to EACH IC USED in the Design. My personal choice is to use about 100 uF/100mA Input load current. Switching regulators are recommended as they as a class waste less power with typical efficiencies @ 70 to 90+ %. Consider a 9V source for a 5V Load @ 100 mA. The load power is P = E X I or P = 9V X .1A = .9W. The Load power is 5V X .1A or .5W. Therefore .4 W is WASTED as HEAT in the regulator ! so the battery is doing almost twice as much work as is needed to power the load. Now consider a Pre Regulator and a "Low Dropout" voltage regulator, If the Pre Regulator is set to 6 - 6.5V out (this will NOT work with the 78XX series of regulators) and the same 100 mA load, Now the input volatage is 6.5V and the load current is .1A X 6.5V = .65W load to the Pre regulator and the same .5W load power now we have .65 - .5W = .15 W waste power, a difference of .25W of power that isn't wasted in heating the regulator and environs... 1/2 of the total 5V device load current. Small switch type buck and boost mode converters are available from Ebay for a little less than $2.00. These devices typically will deliver 1 - 2A of output current and most will operate with >12V input voltage. The Boost Mode devices are most preferable for battery operated devices as they will generally work from a 3V input to deliver 5 Volts to a project. This is great if one has a Li-Po 3.7V battery and a boost mode device then with a suitable charger... it is Hard to put too much filtering on a PSU, however it is very easy to not have enough filtering so my thought is that it is much BETTER to err on the "Too Much" side than it is on the "Too Little" side and the parts are inexpensive and easy to use.
IMO and some Food for Thought.
I'd ditch all the linear regulators and go with switching regulators for both supplies, directly feeding the 3.3V and 5V header pins (if additional 3.3V is needed right on the card, which I would doubt).
The arduino's caps would then take of its own decoupling needs, and 10uF cap at the additional loads with 0.1uF across each IC would be all you need.
There are several here that can source high current amounts if needed, up to 7A.
I use one of the Boost regulators to make 12V from 5V for one of my designs, needed the higher voltage to drive banks of strings of 5 LEDs in series.
It is in my Direct experience a good idea to use about 2 -3 X the value that you "Think" you Might need for input and source filtering. The components use no power themselves (unless they are defective) and WILL save a great deal of wasted time in trying to find issues later. If you look at the individual cost of a 10 uF VS a 100 uF the difference is small... Cheap insurance and if one is using batteries, Required as a battery has a varying internal resistance as it's capacity diminishes/discharges.
The other thought is, as I described in my last post to "Distribute" carefully the actual 'device' filtering... DON'T PLACE ALL the caps in the area allotted to PSU in your project, spread them out . In many cases 1 to 10 1uF / .1uF 'combination filters are more effective than a single 470uF cap on the PSU alone, especially if you use SPI or I2C communication and DO NOT use the filtering or the power supply on the Uno or other CPU boards for external and/or especially high current loads (Large numbers of LED's or Motors/relays). This is even MORE important if you use 'bread-boards for project construction or development because of the long wire length's. Filter capacitors are cheap insurance... As Always IMO
Doc
I have a design that drives an LCD display and a GPS receiver for a time server device as well as radio link... all from a 3.7V Li-Po battery and all the rest of the design operates from 5V... with a proper Li-Po charger. Without the very special charger the Li-Po battery is DANGEROUS. The batteries WILL EXPLODE if not handled properly but make a wise choice for battery operated devices.... But Only if YOU KNOW HOW TO USE THEM.... and oh yeah, the supply rail noise is about 10 mV and that noise is I suspect 'common mode' noise or ground differential noise that is 'common' to the ground circuit. IMO
Doc
Thanks for the responses - all useful stuff. I get the impression that the selection and placement of smoothing and filtering capacitors is more of an art than a science!
Regards
Yeah, But you can BRUTE FORCE them... just double up the value's and spread them out. DO NOT use a remote breadboard without them... unless you have a lot of time to figure out whether it's the sketch or it's implementation that is not working properly. Ya know I am an old engineer anfd I hate as much to have to tell someone else why i think their project isn't working properly as I do being told that my "Sketch was apparently written by someone afflicted with terminal STOOPID. I did do for many years successful engineering (sometimes)... But when it comes to writing sketches... I still don't understand why there is no toilet paper alongside most 'Hot Rocks' as I get them confused with the item that does require TP, frequently... In Everyone Else's Opinion...
Doc