# What should I do to control the impedance in piezoelectric energy harvester

I am trying to design an piezoelectric energy harvester circuit and using voltage multiplier rectifier to build it now.

I am using a piezo film, LDT4-028k as a source and it has 11nF cap inside the film. Based on the datasheet, what should I do if I need to make the input impedance as 10Meg ohm?

What is the reason that my circuit it so inefficient and just can charge a few microvolts at a time?

What should I change to enlarge the charge rate and how to determine the actual value of the components I need to use?

• You have a 142.69 henry inductor - why? – Andy aka Apr 10 '20 at 14:06
• just because my friend told me to use the v/I against time to calculate the value of the inductor. But after I calculate it and use the value of inductor, it still cannot solve the efficiency problem. – Anson Apr 10 '20 at 14:13

Your piezo sheet is a sensor.

It isn't intended to collect large amounts of energy, and cannot do it.

The datasheet says the load impedance should be kept above 1 million ohms, and preferably above 10 million ohms.

The maximum voltage it can produce is 100V, and that would be into a 10Mohm load. That would be a current of 10 microamperes. At 100V, that would be about 1 milliwatt of power.

The leakage current through your electrolytic capacitors is probaly higher than the current the piezo sensor can provide.

Throw out all the voltage multiplier stuff.

Use a single diode and capacitor to collect the energy.

Use a ceramic capacitor (or several in parallel) to collect the energy. They leak less than electrolytics.

Use a capacitor rated for 100V. Use maybe 4.7uF as a collector.

Use a low leakage current diode.

I'll sketch a circuit for you later when I'm at my desktop computer.

Keep in mind that you will never collect large amounts of energy with that sensor.

No amount of efficiency will allow a butterfly to carry an elephant.

As promised, a simpler version of your collector circuit:

simulate this circuit – Schematic created using CircuitLab

There is still a (Greinacher) voltage doubler in there, but I used it more because it'll help you catch all the AC from the piezo as DC.

The capacitors are the main problem you are going to have. Your piezo can't deliver much energy, and it can't do it quickly. You need good, low leakage capacitors that won't lose the collected charge.

Electrolytic capacitors are out.

Film capacitors would be better, but it is unlikely you will have any at hand.

Ceramic capacitors are your best bet, but common ceramic capacitors that you will find don't have large values - you can't store much energy in them.

Working just with common parts that you can scavenge, you'll want to look for the highest value ceramic capacitors you can find. 100nF is the largest common value you'll find.

Put as many of them in parallel as you can.

You may also need to put several in parallel for C1.

The whole thing needs to be well insulated. At the very low current levels you are dealing with, the charge can leak off even through things that you wouldn't expect to conduct. I mean, you can normally ignore the leakage in a capacitor but for what you are doing it is a major factor - and that's measured in microamperes. Resistances in the mega ohm range are a loss of current that you can't afford.

The source impedance isn't really a problem, and I doubt that was what was causing you problems before. The source doesn't see a load until its voltage has risen above the voltage already on the storage capacitor, so the voltage from the piezo can rise with hindrance until it is high enough to pass the diodes and charge the collector capacitor.

This will only do as a demo circuit. Put the piezo in your shoe, and run around. When you stop, reach down and push the button. The LED should flash briefly. That's about the best you can expect.

Good luck. You're going to need it.

I had another go around with the circuit and a simulator. I think you would do well to simplify it further, and to drop any ideas of using a voltage multiplier.

Look at the circuit above. C1 will charge up to the same voltage as the collecting capacitors. That's energy that you can't access, but have to collect. In every doubler stage, you will have one capacitor that you charge but for which you can't access the stored energy. I did a bunch of simulations, and that circuit actually works best when C1 has the same capacitance as the total of the storage capacitors. That means you have the same amount of energy stored in C1 as in the storage capacitors - but you can't use the energy in C1.

AndyAka suggested using a full wave rectifier bridge, so I ran some simulations on a simple bridge rectifier and capacitor collector.

That works much better, and you get access to all the stored energy.

The circuit looks like this:

simulate this circuit

The simulated piezo is just an AC voltage source in series with a resistor to limit the current.

That circuit charges up quickly, and follows the RC time constant for the capacitor and the internal resistance of the piezo. All of the collected energy is available from the output.

I couldn't find any reasonable description of the impedance. When the capacitor is completely discharged, then the piezo will see a low impedance over the duration of every pulse it generates. As the capacitor charges, the piezo will see longer periods where there is no load because the diodes aren't conducting with ever shorter periods where it sees a low impedance. Over time, the average changes from very low to very high impedance.

I tried using a transformer to raise the impedance of the collector. All it did was slow down the charge time for the capacitor.

I looked up a couple of papers (paper 1, paper2) on harvesting energy from piezos. One suggests using a series inductor or alternatively a series switch between the piezo and the rectifier. The other suggests using a buck converter. Both will control the load that the piezo sees.

I tried a series inductor with values from 1µH up to 10kH. There's no noticeable change in the current drawn from the piezo until you get about 100H. All it does is to slow the charge time on the capacitors.

I didn't try the switching setup on the charge side. I don't think you'll be doing that, any way. You'd have to drive an oscillator from the collected energy as you are collecting it.

The same goes for the buck converter on the output. You need an oscillator that can run from the collected energy to drive the switch in the buck converter.

In any case, those things all relate to trying to control the impedance of the load that the piezo sees - that's including the charge load for the capacitor and the output load.

You are intending to collect energy and dump it out after it is collected, not use it as it is being collected.

Given that, I don't see that impedance matching does you any good anywhere.

• Thank you so much! I am trying to understand your point now. Since this is my first time to build a piezoelectric harvester circuit, lots of the things I need to learn. – Anson Apr 10 '20 at 15:00
• Yeah, there's a lot of concepts in there. I tried to use terms that you can look up. "Leakage current" is a thing you can find specified for capacitors and diodes. Good datasheets will include those terms. – JRE Apr 10 '20 at 15:04
• I need to use some time to understand the concept... Before this circuit, I used my piezo film to connect a full-wave rectifier and via the rectifier to charge a 4.7uF capacitor. Why I can charge up the capacitor at that time? Although it only can charge a few micro Volt at a time. Since I just want to charge about 5V and use it to light LED or some small energy device. – Anson Apr 10 '20 at 15:29
• Oh, thank you so much!! This is very helpful for my project! – Anson Apr 11 '20 at 8:45
• Hello JRE, I tried to use your suggested circuit and applied a voltage doubler to make it different. But my teacher always requires me to need to do the impedance matching...Can you tell me how to do this matching part? Since my teacher already requires me again and again... – Anson Apr 15 '20 at 5:32

According to the data sheet this is a sensor and cannot be used to generate/harvest any usable energy.

If you want to just detect motion or vibration you can of course amplify its signal, but this requires an external power source to actually light up your LED or whatever indicator you wish to use.

• Is it means I cannot use LDT1-028k to being a source of my harvester circuit? Then what source should I use if I need to build a harvesting circuit? – Anson Apr 10 '20 at 14:41
• Correct. You will need to specify how much energy and/or power you wish to generate and what kind of motion that will be used to actuate the harvester – Jakob Halskov Apr 10 '20 at 14:47
• I just want to charge around 5V and light a LED or something that only needs a small voltage. I want to put it in the shoes and charge the energy when I am moving. – Anson Apr 10 '20 at 15:03

I'm not so gloomy about the sensors inability to produce some usable power especially when the "right" harvesting circuit is employed such as this: -

In any case, if it proves to be useless at least you will have a test circuit that is more likely than any other offerings to yield results with the "right" sort of sensor.

The LTC3588 data sheet is here at Mouser: -

Good luck and come back and tell how you got on. Be gentle i.e. don't go hitting the sensor with a hammer - it can produce 100 volts and the chip above is only rated for 25 volts so, take it easy and gently get to the point where you are satisfied with performance.

• Is it have any method to only use a circuit and no need to use a chip? Maybe just keep storing the energy in a capacitor and use the voltage in the capacitor to light the LED or other devices? Since I need to submit to my school and the time is not enough to buy a chip on the internet now... – Anson Apr 10 '20 at 17:41
• Use low leakage diodes in a bridge rectifier and use a low leakage 1 to 10 uF capacitor to receive charge. If you have an oscilloscope, monitor the voltage gradually rising on the capacitor. you should be able to "acquire" a few tens of volts so, if the capacitance is 1 uF and you acquire 20 volts, that's an energy acquired of 0.2 mJ. The more voltage you can accumulate then the better because energy is proportional to voltage squared. If the rate at which voltage is obtained can be reduced, do so by using more capacitance because energy is proportional to capacitance ($W = CV^2/2$). – Andy aka Apr 10 '20 at 17:47
• Try 1N4148 diodes for the bridge as these are reasonably low leakage but, if there are signs of the bridge discharging the acquired voltage then look for lower leakage types. You might also want to do experiments with a charged capacitor to see how long it might light an LED - use a current limiting resistor of course. You'll get a short flash and this shows why the chip solution is much better - it takes the relative high volts from the sensor and "bucks" it down to a much larger storage capacitor that would sustain charge for much longer. – Andy aka Apr 10 '20 at 17:49
• Thank you so much! I will write the chip method in my report later. But for the circuit, Is it means I need to use my sensor to connect a full-wave rectifier and via the rectifier to store the voltage in a 1uF capacitor? Is it have any method I can speed up the charging time? Since I am using 1N5819 now and the charging time is too long. After changing the diode to 1N4148, what another method should I use to speed up the time and enlarge the charging voltage? Is it helpful for using voltage multiplier after the capacitor? Or it is already enough for me to charge 5V at a short time? – Anson Apr 10 '20 at 18:44
• You can only speed the process up by mechanically moving the piezo more often. – Andy aka Apr 10 '20 at 20:17