Help Needed: Piezoelectric Micro-pump

Hi everyone, I’m new to the forum, I hope this is not breaking any rules.

I am trying to create a micro-pump using the vibrations of the underside of a piezoelectric buzzer as a diaphragm to move fluid. It’s based off of this post:

The pump does not use an active valve, instead it uses “bluff bodies”, which are hemispheres that provide more resistance to flow in one direction as opposed to the other, meaning theoretically that more fluid moves in one direction as opposed to the other.

Here’s the setup: The body of the pump is 3D printed, and the piezo disc is driven by a TI DRV8662 driver using an external PWM signal to control the frequency and amplitude of the driving signal. The driving signal is around 130V max p-p and resonance is achieved around 60-70 Hz as well as around 1.6 kHz.

The problem is that while I’m getting resonance of the fluid in the pump, the fluid is moving back and forth an equal amount, meaning no net displacement of the fluid. In other words, the bluff bodies are not working as a valve as intended.

I have attached a picture of the CAD model of the bluff bodies, and a video of achieved resonance . Please let me know if you have any ideas as to why the set up isn’t working as intended or if you have any alternative suggestions of valves etc.

Video: (

1 Like

Welcome to the forum, definitely not breaking any rules, this is a cool project!

I have a few thoughts that may help in debugging the pump:

You want to run it at the first resonance of the disk a so the higher order modes will not be the disk deforming up and down like in the diagram, it will instead be a much more interesting mode shape. The second eigenmode will probably be a symmetric and so there won’t be a change in volume in the chamber.

From the looks of the diagram the pumping mechanism seems to assume that the central chamber is full and that the vibrations are moving the water (air is compressible so ruins the efficiency). I cannot tell from your video but it looks like there is not much water in the system. (Of course filling it to the top with 130V going to the piezo is going to take some thinking about risk, some sort of dealing membrane may be worth considering)

I am not sure it will work pumping from the current containers, as unless the the output container is filled above the hole, air will seep into the system. What if you made the input chamber a bit larger, and gave it a lid. Then remove the output chamber so it is just a spout, and made it bit longer than in the current design. The aim would be to for the system to have the liquid in the input chamber not leaking out of the end by surface tension on the end of the spout, so it would probably require a fairly small hole. You would also need some clever way to tilt the system to purge all air from put pump container (hence the lid in the input chamber).

I assume the pumping force is pretty weak so you need to find another way to rely on surface tension. I Would not be surprised if it takes quite a lot of optimisation of the shapes and sizes of the components to get a noticeable force. The group that did this work seem to do a lot of simulations to get something with a pumping speed of a few uL/min.

Good luck, let us know how it goes.

Going to second Julian’s guess that there’s air in the system. And the volume of sound generated suggests the same conclusion. There may also be porosity in your print. The piezo disc may be leaking air where it attaches to the pump body.

Here’s a similar way to move water. Maybe less noisy.

If you’re going to pursue this angle, and you’re going to 3D print the device, you can also look into using a Tesla valve instead of “HSBB”.

On a pragmatic note, you can buy peristaltic pumps, some very small and suitable for microfluidics. These are metering pumps which may be useful for your application.

1 Like

Some very good points by both of you. Yeah there is definitely more movement at the first resonance frequency.

I will definitely have a look at the suggestion of air in the system. I have some silicon tubing that I will use in the next design instead of the input and output chambers. This way, I can have the tubing submerged in another reservoir, and prime the pump by manually pulling liquid through with a syringe through the output tube. This should eliminate air in the system. I’ll update if there is any success.

Also excellent idea about the Tesla Valve, that will be my next attempt, thanks!

Some more notes: there doesn’t seem to be significant porosity in the print as the vessel holds water without leaks for a long period of time. There is also a thin layer of silicon between the piezo disc and the pump body so I don’t believe there are any leaks there.

Thanks for the suggestion Harold, I actually already have a working peristaltic pump for use with this project. What attracts me about the piezo pump is the size and the way it can be integrated into the pump body.

Again, the help is much appreciated and if anyone has anymore ideas, please share!


Just a quick update on this post, and a BIG thank you to both of the responders! I finally got some pumping.

So I changed a couple of things: As was suggested, I replaced the input and output chambers with silicon tubing. As Harold suggested, I ditched the HSBB and replaced it with a Tesla Valve. I also put hot glue over the edges of the piezo disc to improve the seal and safety.

Even with a Tesla valve with only one baffle, there is some decent flow rate (relatively speaking). The noise is also greatly reduced with a triangle wave instead of the previous square wave. There’s about 30 mm head pressure developed.

Please see the Cad Model of the new valve and a video of the pump working.

Once again, thanks so much for the suggestions from both of you, I would’ve still been scratching my head if I hadn’t come here :D.

Uploading: IMG_20201010_153505.jpg…

1 Like

That’s great, I am glad it is working.

I would you be able to share the original CAD file or the STL so that others could print it?

Sure, here is the link:

1 Like

Well I’m back with another question for you guys. I expected the most pumping to occur at the resonant frequency of the Piezo disc, which is 3.6 kHz. The actual best pumping frequency is much much lower at around 80 Hz.

Does anyone have any ideas as to why there’s such a big difference?

Thanks again!

Quick 1am thoughts (may not be coherent or useful):

How are you measuring the resonance? It will probably drop loads when clamped and in contact with water.

You may be hitting a resonance of something else inside the system. Waterwaves in the cavity?

While you will get the largest piezo motion at resonance, but the water motion through the pump may be much lower as it cannot react fast enough.

1 Like

You said what I was thinking! it reflects what I’m seeing when the pump is in operation. At lower frequency (for example as low as 1 Hz), the fluid moves much more per stroke than at high frequencies.

But of course there is a balance between how many strokes per second and amount of fluid displaced per stroke to get the highest flow rate (which is what the 80 Hz seems to be). As the frequency goes much higher than 100 Hz, the water doesn’t seem to have enough time to react just like you said, before the next cycle begins.

I hadn’t thought of the fact that the resonant frequency of the disk changes when clamped, very good point. I was getting the resonance frequency of 3.6 kHz from the disc data sheet, but this is most likely when the disc is not fixed to anything, and in contact with air only…

Final note, that other papers that deal with piezo-pumps also mention optimum driving frequencies of around 100 Hz.

Thank you again you’re a great help!

1 Like