I'm trying to drive this piezoelectric air pump that requires 1.5W and runs on 120VAC 60Hz wall power.

This will be controlled by a MCU/CPU of some kind (Raspberry Pi in this instance), so the control signal is 3.3V. At the expense of wiring through some probably-already-present transistors or level-shifters, 5V and 12V are also available.

In the bigger picture, this is part of a closed-loop control system. There will be pressure sensors, and the controller will run the pump until desired pressures are obtained. That pressure will be consumed until some low-level threshold is hit, at which point the controller will run the pump again. This will repeat indefinitely.

Wall power (120VAC) is available, which may or may not be helpful, depending.

This will run while I'm sleeping. It has to be quiet. (So I suspect that electromechanical relays won't work.)

Also, it's going to be cycled on/off A LOT, like 10-1000 times/day as a ballpark. (This also seems damning for electromechanical relays.)


What I've researched so far

So far I've considered 3 ways to accomplish this, in broad terms:

  1. Use a solid-state relay to switch wall power (120VAC) on and off.
  2. Use a small power inverter to convert the 12VDC rail into 120VAC.
  3. Use a different piezo pump that runs on 3V-12V, because this 120VAC stuff is nuts. :/

Option (3) is what I really want to do. But it's impractical enough that I'm still ending up here, writing this post ;)

Ideally, the unit would push static pressure >=10kPa (>=100 mbar). These exist, but availability is problematic. The muRata MZB1001T02 seems to be the plentiful wholesale-ish option for this, but its static pressure is a lackluster 1.42 kPa. Thankfully they also make the MZB3004T04 which sports a generous 60kPA ceiling and looks like it will easily hit 10kPa while powered by a 12V rail. Problem: it can't be on my desk before 2021-04-30 (it's on order, and AFAIK, Digi-Key doesn't have it at all). Also, the quantities look low, so I worry about supply stability. I've searched other options too, like this 15V-35V (ugh) Maxclever offering on Alibaba for $50 (ouch). The CurieJet GS9S looks potentially good and is approximately $20/unit according to Alibaba, but I'm having trouble sourcing it. I'm really hoping for retail offerings, for the sake of anyone who follows in my footsteps. I think there are other possibilities, but they tend to get excluded quickly (ex: pumps only for liquids, pumps only for pulling vacuum, etc etc). But the 120V units, those have the aquarium market in their pocket! They are very plentiful. I'm probably going to build two variants of my thing, one (the optimal build) that uses the muRata MZB3004T04, and another (the fall-back build) that uses the 120V aquarium piezo pump. For now, it looks like I might be working on the fall-back version first. That leaves me still needing to consider options (1) or (2).

Option (1), using a solid-state relay, was my first consideration for providing 120VAC. Unfortunately, I'm finding that Solid State Relays (SSRs) tend to have a minimum load current. For example, this normal-looking Crydom ED24D3 ($13.67/unit) has a 0.15A "Minimum Load Current". If the pump uses 1.5 watts, then 1.5W/120V is 12.5mA, and 12.5mA < 150.0mA, which leads me to believe that this relay won't be guaranteed to remain in its "closed" state after the pump starts sipping the juice. Following that discovery, I searched around for a while and found the Crouzet SSRs, specifically the GN family. The Crouzet GN 84137000N seems to just barely fit the bill, with a "Minimum Load Current" of 5mA. It certainly seems like a nice piece of hardware. But at $31.18, it's more expensive than the dang pump! As far as I can tell, SSRs that switch 120VAC tend to have >100mA minimum load current, and anything with a lower load current is somewhat rare. Compounding my frustration is that no Etailers seem to be interested in making this a searchable parameter, so I can't exhaustively scan this market. Maybe there's a perfect fit out there. I wouldn't know! And that, in and of itself, is a substantial problem.

In a nutshell, SSRs that switch 120VAC seem to be designed for handling things that draw significant power, and my piddly 1.5W is not a design consideration for them.

So I thought, maybe I can reframe the problem and just not use the wall voltage at all! After all, it's just a 1.5W draw, and the DC power rails can handle that without issue.

Enter option (2): using a power inverter to convert the 12VDC rail into a dedicated 120VAC line. This had me searching power inverters for some time. And it's also a frustrating exercise, as the market seems saturated with devices that produce >=150W, which is waaay overkill for that piddly 1.5W piezo pump. I could presumably pick up a cheap (~$15) automotive 12VDC->120VAC converter and snip off the "cigarette lighter" plug to solder the 12V end to my 12V rail. But this has caveats. First, it might put out a square wave or some crappy not-really-sine-wave that might wear out or damage the piezo pump (maybe not, but I don't know). Secondly, power conversion devices tend to have a "power curve" and operate most efficiently under some ideal range of loading (say, 50%-80%), which means that the inverter might have a high quiescent load (usually not specified, unfortunately). So while the pump-only 1.5W load should be no problem for my PSU, the pump+inverter load of unknown magnitude might be no good at all. Assuming the inverter load isn't an issue, $30 can buy a modified sine-wave inverter (whatever they mean by that), and $40 can buy a pure sine-wave inverter. This puts us back into "it's more expensive than the dang pump" territory, except with additional caveats about quiescent load. Also they tend to be very bulky.

Now, it seems like the electronics component gods are almost aware of the problem, as they provide us with single-board options like this and this. But. They provide AC in higher frequencies, like 20kHz in these cases, which means they also seem to provide a small graveyard of DIY projects and sad reviews. Bummer! Oh, and they are square wave outputs. Maybe with the intended load being as small as it is, I could just put a capacitor and smooth that out, but it doesn't matter, because 20kHz != 60Hz. (And capacitor cost != $0.)


Other ideas I've considered

These tend to involve just physically converting the 120VAC pump into a 12VDC pump, but this is either impossible, or I don't currently have the requisite knowledge.

Along those lines, my first thought would be to buy the 120VAC piezo pump, take it apart, and remove some power supply circuitry to end up with a 12VDC piezo pump. This would assume that the manufacturers of the aquarium-style pumps are using a 12VDC pump under-the-hood and just convert 120VAC into that. I'm pretty sure that this assumption is false, unfortunately. I took to the internets and found a teardown video, and it would seem that the manufacturers are likely just putting the 120V directly into the piezo element. (For clarity: there is power supply circuitry in that teardown, but it just seems to convert 240VAC->120VAC, and in a very cheesy and inefficient way, perhaps just so that they could reuse the 120V piezo internationally.) So I don't think I can solve the problem by just subtracting components.

Past that, I've wondered about the physics of piezoelectric devices and if the 120V is truly necessary to drive the aquarium-style piezo pumps. Maybe 12V would still max its deflection. Or maybe, because higher frequency waves tend to transmit more power (the energy in a mechanical wave goes like the square of its frequency multiplied by the square of its amplitude), 12V at a higher frequency would end up providing the same air-flow + pressure as 120V at a lower frequency. Notably, the 120V versions seem to be using a 60Hz oscillation, while the more whole-sale-oriented devices running at lower voltages (like the MZB3004T04 at 8V-18V) seem to typically operate around 20-25kHz. Similar to the "just use a different piezo pump" option, such a workaround could potentially allow me to ignore the "how do I provide MCU-controlled 120VAC" problem entirely. But this is all speculation; I am not confident in these matters.


Now what?

All of that has me bothered. I only need 2W out of this thing. And 60Hz is a really low frequency; GPIO pins can do that (and a DAC pin would even yield a sine-wave!).

It's just that jump from 12V to 120V that seems to introduce a bunch of baggage. At the simplest, I could envision introducing a boost converter before switching to AC, or maybe switching to AC and then putting that through a transformer to level up the voltage. But those are also additional components and assembly steps. If I lean on Mouser, I get dunked on by availability issues or drown under the "Minimum Load Current" of SSRs. If I lean on Amazon, I end up adding USB-C ports for no reason. ARGGHHH!

Surely there has to be a better way!

Is there an easy+inexpensive way to build a quiet, reliable(ish), MCU-controlled, low-power/low-current 120VAC supply? If so, how?

(If the physical properties of these piezoelectric pumps can be exploited to turn a 120V pump into a 12V pump, I'd be interested in that too. That feels off-topic, but I'm just not optimistic enough about such things to spend time researching that and writing a question for it, at least not just yet. In other words: I currently expect this problem to actually require 120V conversion.)

I eagerly await your guidance, experienced ones.


The Project

For the curious, or in case it helps, this is what I'm working on:

I'm building a device that keeps my ears at pressure equilibrium during CPAP/AutoPAP so that I don't destroy my ears. In some patients, otic barotrauma is a possible outcome of CPAP, and I suspect that's me. (Yes, I've already talked to my doctor. The medical community doesn't have a good answer for this.) I won't be putting 100 mbar (about 100 cmH2O) into my ears, of course, but I suspect I'll need some headroom (and possibly a modest buffer cavity with motorized valves) to allow my control system to quickly adjust the pressure, because the CPAP/AutoPAP is constantly changing pressure in-sync with my breathing patterns. Other people get ear pain from CPAP too, and I've heard some sad stories. So I'm really hoping to design this in a way that's practical for others to build; which means it's highly desirable to reduce part count, use abundant components, minimize assembly steps, and avoid specialized tools (though basic soldering tools are a given at this tech level). Piezo pumps seem to have excellent characteristics for this (quiet + cheap + good-enough static pressure), but are proving to be a bit of a corner case, one that leaves me wrestling with a parts-availability vs build-complexity trade-off.

  • \$\begingroup\$ I see it has an AC adapter - maybe it already runs off DC. What does it say on the adapter? \$\endgroup\$ Apr 10 '21 at 1:46
  • \$\begingroup\$ The link shows a wall wart device. That looks like some sort of power supply. I'd be finding out what that is first. I'd also suggest you look at a simpler means of regulating the pressure - using a micro is not the most reliable course of action. \$\endgroup\$
    – Kartman
    Apr 10 '21 at 1:48
  • 1
    \$\begingroup\$ At such low power, a 60Hz audio oscillator into an audio amp to a 12->230V transformer. You can try it if you have some amplified speakers for your PC. Get a small 12V->230V transformer (from an old wall wart) and connect the low voltage winding across the speaker. Get an audio oscillator app and play around with the volume until you get 230VAC on the output of the transformer. Concept proven/disproven. \$\endgroup\$
    – Kartman
    Apr 10 '21 at 1:59
  • \$\begingroup\$ @Kevin_White A reasonable question, but I can't check. I don't even know if this is practical, so I haven't bought the thing. This might suggest that it's different from what I saw in the teardown video. Documentation is scarce. I did find this question+answer on an identical-or-related product: amazon.com/ask/questions/Tx2F3HV2XLDP3PG/ref=ask_dp_dpmw_al_hza The seller was inconsistent about their description, but it does make me wonder if the wart gives low volt 20kHz or something. It's starting to sound like it might be worth the risk to buy-and-try. \$\endgroup\$
    – chadjoan
    Apr 10 '21 at 2:31
  • 1
    \$\begingroup\$ Mechanical failure is way easier to predict than a complex hardware/software solution. How does a microcontroller fail? Very unpredictably - what line of code was it executing when it got upset? What happens when ram is corrupted? There’s lot of design gone into your cpap machine -,with a price tag to match. \$\endgroup\$
    – Kartman
    Apr 10 '21 at 8:12

At such low power, a 60Hz audio oscillator into an audio amp to a 12->230V transformer. You can try it if you have some amplified speakers for your PC. Get a small 12V->230V transformer (from an old wall wart) and connect the low voltage winding across the speaker. Get an audio oscillator app and play around with the volume until you get 230VAC on the output of the transformer. For a standalone circuit, there's single chip audio amps with boards available online and the addition of some R & C would make a power oscillator at any frequency in the audio range.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.