Ask three suppliers how ultrasonic cleaning works and you’ll get three versions of the same sentence about sound waves and tiny bubbles. Technically correct, practically useless. It doesn’t tell you why the machine cleans one part beautifully and leaves the next one dirty, which is the question that actually matters once you own one.
Here’s the mechanism, and then the four things that decide whether it works on your part.
Short version, for anyone who wants the answer to how does ultrasonic cleaning work before the detail: a generator drives transducers, the transducers vibrate the tank, the vibration tears microscopic voids in the liquid, and those voids implode against the part. The implosions do the work. Everything else in this article is about when that mechanism reaches your part and when it doesn’t.
The actual sequence, step by step
Every ultrasonic cleaner does the same four things:
- A generator converts mains electricity into a high-frequency electrical signal — 20,000 to 120,000 cycles per second in industrial work.
- Transducers bonded to the tank wall (or mounted on a plate inside it) turn that signal into mechanical vibration. Piezoelectric ceramic expands and contracts with the applied voltage; that’s the conversion.
- The tank wall vibrates and pushes pressure waves through the liquid — alternating compression and rarefaction, thousands of times a second.
- Cavitation happens during rarefaction: the liquid is pulled apart faster than it can stay together, and microscopic voids form. On the next compression they implode.
That implosion is the cleaning. Each collapsing bubble releases a tiny, extremely localised burst of energy against whatever surface it’s touching. There are millions of them per second across every surface the liquid can reach.
Why it beats scrubbing — and where it doesn’t
Two properties make cavitation useful in ways a brush never will be.
First, it reaches wherever the liquid reaches. A cavitation bubble doesn’t care about thread roots, mesh, hinges or the underside of a burr. That’s the whole reason the technology exists: it cleans geometry a person can’t physically touch.
Second, it works without abrasion. Nothing rubs the surface, so dimensions and finishes survive — as long as the frequency and power are right for the material.
The flip side matters just as much. Cavitation only happens where there’s liquid, and only where sound energy arrives. That’s why the same machine that cleans a mesh basket of fasteners perfectly can leave a blind hole untouched: air is trapped there, so no liquid, so no cleaning. It’s also why a part resting flat on the tank floor comes out dirty exactly where it touched.
The four variables that decide the outcome
The mechanism is fixed. Everything you can change lives in these four, and they interact — change one and the others need to move.
Frequency
Lower frequency makes fewer, larger, more energetic bubbles. Higher frequency makes many more, much smaller, gentler ones. So 28 kHz strips machining oil off castings and would pit an optical coating; 80–120 kHz cleans that coating safely and won’t touch heavy oil. Our range spans 20 kHz on immersible transducer rods to 950 kHz on megasonic units for exactly this reason.
Chemistry
Cavitation breaks the bond between soil and surface; the solution dissolves what’s been released and holds it in suspension so it doesn’t land on the next part. Plain water cavitates fine and cleans badly.
Temperature
Warmer liquid thins oil and activates detergent, but past a point — roughly 70 °C for most aqueous work — bubbles fill with vapour and collapse softly instead of violently. Cleaning gets worse as the bath gets hotter. Most industrial work sits at 50–65 °C.
Time
Cleaning is fast at first, then flattens. If five minutes doesn’t do it, twenty usually won’t either — that’s a signal to change frequency, chemistry or fixturing rather than to run longer.
Degassing: the step that explains most disappointing demos
Fresh liquid holds dissolved air. Those air pockets cushion cavitation — bubbles collapse into gas instead of imploding cleanly — so a newly filled tank can run at a fraction of its real capability for the first several minutes.
Every machine we ship has a degas function, and it exists because this is the single most common reason a new owner reports that cleaning “isn’t as good as it was at the demo”. Run it for 10–20 minutes after filling and after any large top-up.
Batch tanks, immersibles and inline systems
The mechanism is identical across formats; what changes is how the energy is delivered. A benchtop or batch tank has transducers bonded to the base and sides, which suits mixed workloads and low volume. Immersible plates and rods drop into an existing vessel — useful when the tank already exists, or is too large to retrofit, or holds a process liquid you can’t move. Inline and conveyor systems put the part through fixed stages on a belt, which is what you need when the same part runs all day and cycle time is the constraint.
People ask which one cleans best. None of them: they cavitate the same way. The choice is about throughput, tank size and whether the part mix changes — not about cleaning power.
Why identical machines give different results
Two customers, same model, same chemistry, completely different outcomes. So how do ultrasonic cleaners work so differently in two plants running the same model? It’s almost always one of these:
- Loading. Parts touching each other, stacked, or resting on the tank floor are not cleaned at the contact points. Overloaded baskets also absorb energy before it reaches the middle of the load.
- Orientation. Blind holes pointing down trap air permanently. Turn the part over and the same cycle works.
- Bath condition. A loaded bath redeposits soil. Results decay gradually, so nobody notices the moment it stopped working.
- Fill level. Below the marked level the standing wave pattern changes and cleaning becomes uneven across the tank.
None of these are machine faults, and all four are free to fix.
Does more power always clean better?
No — and this is where money gets wasted. Beyond the level needed for stable cavitation, extra power mostly increases the risk of surface erosion on soft metals and thin sections. Aluminium is the material that most often comes out of an over-powered tank looking worse than it went in.
Power matters for tank volume and load: a large tank needs enough transducer area to cavitate throughout, not just near the walls. That’s a coverage question, not an intensity question, and it’s why immersible transducer plates are often the right answer for a big vessel instead of a bigger generator.
Where ultrasonics fits in a real process
Ultrasonic cleaning is rarely the whole answer on production work. On the lines we build, it’s one stage: wash → rinse → DI rinse → dry, with agitation, spray or vacuum drying added wherever the geometry demands it. Our multi-tank systems exist because a single tank leaves detergent on the part, and detergent residue is contamination just like oil is.
The parts that need more than immersion are predictable: deep bores, long internal passages, assembled modules with gaps, and anything with a written particle-count specification.
The honest test
Everything above is mechanism. Whether it works on your part depends on your soil, your alloy, your geometry and how dry the residue is when it reaches the tank — variables no article can resolve.
Jietai Ultrasonic has built ultrasonic cleaning equipment since 2003, from our own factory in Dongguan, and of the forty-odd custom systems we’ve delivered, most exist because a standard tank couldn’t handle one specific feature of one specific part. Send us the part with its real soil. We run it on our own machines and send back the frequency, chemistry, temperature and cycle time that worked — and tell you if immersion alone won’t do it. The cleaning test is free, and it settles in a day what a specification argument can drag out for weeks.