# How to select a smoothing capacitor?

1. Feb 17, 2016

### grasscut

Im doing piezoelectric energy harvesting, with a circuit with full bridge rectifier, smoothing capacitor and resistive load. May i know how can i select a value of the smoothing capacitor for this situation? I read some article, they recommend 1000uF but didnt mention more information on why isit so

2. Feb 17, 2016

### DrDu

3. Feb 17, 2016

### grasscut

my piezo is generating around 28V of rectified voltage at 17 Hz at resonance. how do i justify if i were to use 1000uF?

4. Feb 17, 2016

### jim hardy

here's how to get a decent estimate without calculus
...... decide how much ripple you want at your anticipated load current
let's say you are happy with 10%
28 v average is 44 peak
but from what you describe i doubt you'll get that much
let's guess you get 30 under load >>>> just a guess, tune this up with your measured values
10% of that is 3 volts
and 17 hz is 29.4 milliseconds between full wave rectified peaks
if your voltage is to droop 3 volts in those 29.4 milliseconds between peaks , that's a voltage change Δv/Δt of 3volts/0294seconds = 102 volts/sec
i = c X Δv/Δt, so
c = i/(Δv/Δt )

You can work it backward and iterate in on an answer too,
since i = c X Δv/Δt , 1000 uf will give 3 volts ripple for load current of 102 milliamps

have fun

and believe in your basics. They solve 99 % of all problems.

old jim

5. Feb 18, 2016

### meBigGuy

By that do you mean the 17Hz signal from the pizeo, when rectified, supplies 28V? Is that with some kind of cap and no load?
That would be different from what Jim is assuming.

The principles are the same, though. When you load the piezo the voltage at your rectifier output will drop and you will see ripple.

6. Feb 18, 2016

### DrDu

7. Feb 18, 2016

### jim hardy

wow
piezo transducers are essentially a charge pump
if this piezo datasheet is typical
then even one milliamp is optimistic let alone 102 of them

http://www.piezo.com/prodproto4EHkit.html

i'm not sure what they mean by (microApeak/Hz)
....if that thing puts out 57 microamps
by i = c X Δv/Δt , Δv/Δt = i/c = 0.000057/ 0.001 = .057volts/sec
which between your 17 hz peaks is only .057v/s X .0294s = 1.68 millivolts

so, for that sensor, 1000 microfarads is more than large enough to accept all the charge the sensor can pump out with almost no change in voltage..

if your piezo generator is similar ,
there's a justification - 1000 uf is more than big enough to capture and smooth the feeble pulses of current it generates.

8. Feb 18, 2016

### DrDu

Yes, but probably it will take a year to charge up the capacitor.

9. Feb 18, 2016

### jim hardy

he needs to learn .

10. Feb 18, 2016

### grasscut

What i meant was, in order to obtain the best output result from the piezoelectric, the piezoelectric needs to vibrate at its resonance to produce the maximum output. 28 V is just a rectified voltage without any capacitor or loads.

thanks jim and dr du for the help. im trying to obtain the maximum output power from the piezoelectric and needed to explain why i chose the value of the smoothing capacitor. eventually, my resistance load will be an rechargeable battery where i will use the piezoelectric to charge it up.

11. Feb 18, 2016

### DrDu

With a battery, you won't need any condenser at all as it acts itself like a giant condenser.

12. Feb 18, 2016

### jim hardy

Thanks for clarifying that
piezo doesn't make significant power
i was afraid you were trying to reduce your electric bill or something

i hope you'll post photos, partslist and results
it might come in handy to know how to make one

old jim

13. Feb 18, 2016

### grasscut

Im doing it for my project, a piezoelectric energy harvesting project. Indeed the output power is not too significant. it can produce a good voltage, but not so much on the current.
I got my piezoelectric material, PZT from APC International.