Tiny Finshing

Getting the Job Done on Micro-Scale Parts

Most people who do fine finishing work have probably hit the wall where their usual techniques stop working entirely. A sanding block designed for automotive paint has no place near a 2mm brass filing, and a standard buffing wheel will just eat small parts alive. Tiny Finshing is what you do when the scale of the component makes conventional methods physically impossible, and it mostly comes down to patience, the right media, and not spending more time on finish than the part itself is worth. The term isn't formalized in any SAE standard or ISO document. It describes finishing operations on components smaller than roughly 15mm in their largest dimension. That includes things like precision bearings, watch gears, micro-machined aerospace brackets, and jewelry components. The defining characteristic isn't size alone. It's that standard finishing workflows can't be applied directly at that scale. You can't clamp a 3mm part into a vise and hit it with a rotary tool without it launching across the room. You can't spray coat it without losing the part inside the overspray catch. Every step requires a fundamental redesign of the tooling approach. I learned this the hard way about four years ago when a client sent me a batch of 8mm titanium aerospace clips that came out of CNC with visible tool marks. They needed a uniform surface finish of 0.4 micrometers Ra for a fretting resistance test. Standard vibratory finishing with steel media was out of the question. The clips would have just jammed between each other and tumbled as a single lump. I ended up building a custom nylon mesh bag system, separating each clip individually, and using a glass bead media in a low-rpm planetary barrel. It took nine hours of cycling to get the parts clean enough for secondary hand deburring. The whole job that should have taken a day and a half ended up taking three.

The Core Methods That Actually Work

There are three approaches you should know about, and they aren't interchangeable. Ultrasonic cleaning with finishing media is the first real option. You put the parts in a tank with a slurry of fine abrasive media suspended in water or a water-soluble fluid, then run ultrasonic waves through it. The cavitation bubbles collapse right against the part surfaces and micro-fracture the burrs and tool marks. It works well for complex geometries with blind holes and internal passages that no brush or wheel could ever reach. The trade-off is that you need to buy an ultrasonic generator that can actually deliver meaningful power at the frequency range these parts respond to. A lot of cheap units on the market are just novelty toys with transducers that can't couple properly through the tank walls. The second method is chemical or electrochemical micro-finishing. This is where you immerse the part in a controlled electrolyte and apply a precise current density. Material removal happens only at the microscopic asperities, which are the highest points on the surface. It smooths without changing dimensions appreciably, which matters when your part is already at tolerance. I used this on a batch of stainless steel surgical instrument pivots once. The specifications called for a mirror finish inside channels that were only 1.2mm wide. Mechanical access was impossible. The electrochemical process cleaned the channels completely and brought the Ra down to about 0.15 micrometers. It cost roughly $4 per part in consumables and took about twenty minutes per batch of twenty parts.

The third approach is abrasive flow machining, sometimes called flow forming. You force a viscous abrasive media through or across the part surfaces under hydraulic pressure. For tiny parts, the media extrusion version is what you want. The part gets sealed between two cylinders, and the media is pushed through the part's own geometry or across its external surfaces. This is excellent for deburring sharp micro-edges without rounding them over. The issue is that setup time scales poorly with part count. You're building custom seals and fixtures for each part geometry, and if you only have a dozen parts to finish, the fixture fabrication alone will eat your margin.

What Beginners Mess Up

The biggest mistake I see is underestimating how much thermal damage these processes generate. Ultrasonic finishing can heat a small batch of steel parts from room temperature to over 60°C in under ten minutes. That thermal expansion alone can change the dimensions of a precision part by several micrometers. If the part was already finished to tolerance, you've just moved it out of spec. The fix is to run the bath temperature below 25°C with a recirculating chiller, and to check the part dimensions immediately after the process before it warms back up to ambient. Don't let it sit on the bench for an hour and then measure it. Another common error is using media that's too aggressive for the scale. A standard 0.5mm ceramic media grain will cut into a 2mm brass pin like a chainsaw. You end up with pitting instead of smoothing. The grain size should be roughly one-tenth of the part dimension you're working on, which means you're often looking at 50 to 150 micron media for most tiny finishing work. That's nearly dust-like in comparison to what you'd use on anything larger. It also means your separation and cleaning steps afterward need to be thorough, because spent micro-media embedded in tiny surface pores is a nightmare to remove. Chemical finishing has its own trap. People assume it's just a matter of dipping the part in acid and waiting. The electrolyte composition, pH, current density, and immersion time all interact in ways that aren't intuitive. Get the pH wrong on a titanium part and you don't get a smooth finish, you get hydrogen embrittlement. Hydrogen atoms diffuse into the metal lattice during the electrolytic process and can cause delayed cracking under stress. I've seen it happen on precision springs that looked perfect coming out of the bath and cracked two weeks later during assembly. The workaround is post-process hydrogen baking, which means putting the parts in an oven at about 190°C for at least four hours after the electrochemical step to drive the trapped hydrogen out.

When Tiny Finshing Won't Save You

There are scenarios where none of these methods make sense. If you have a batch of plastic components under 5mm, abrasive and chemical finishing will either melt them or deform them. Plastic has a very low thermal conductivity and a low melting point, so even the mild heat from ultrasonic cavitation is enough to soften the surface and redeposit material in an uneven pattern. For plastics at that scale, the only reliable finishing route is mechanical hand work with progressively finer abrasives, or laser polishing if you have access to a femtosecond laser system, which most shops don't. Another case where tiny finishing fails is when the part geometry has undercuts or recessed features deeper than three times the feature width. Any media or fluid flow needs a clear path into and out of the area you're trying to finish. If the opening is too narrow relative to the depth, the media either won't reach the surface or will create a bottleneck where it accumulates and causes localized over-cutting. There's no workaround other than redesigning the part, which isn't always an option once the tooling is built.

A Practical Step-by-Step for a Typical Job

Here's how I normally approach a run of small stainless steel medical components that come off the CNC with visible grinding marks and need to go from about 1.6 micrometers Ra down to 0.4 micrometers Ra. First, I separate the parts by size and geometry. Mixed lots finish inconsistently because different shapes catch media differently and can interfere with each other. I put them in labeled mesh bags, no more than fifty parts per bag, loaded loosely so media can flow through freely. Second, I run an initial ultrasonic clean in warm water with a mild detergent to remove any residual coolant or grinding fluid. This takes about twelve minutes at 40kHz with a chiller running. Don't skip this. Leftover machining fluid interferes with the actual finishing media and can contaminate the electrolyte if you're switching to electrochemical finishing afterward.

Third, I load the media. For this particular part, I use a pre-rounded glass bead media at 100 microns. The rounded edges prevent new scratch patterns from forming while the hardness of the glass does the cutting. I fill the tank to about sixty percent volume with media and add deionized water to cover. The water-to-media ratio matters. Too much water and the parts just sink and don't interact with the media. Too little and the cavitation energy dissipates into air gaps rather than transferring to the part surfaces. Sixty forty works empirically. Fourth, I run the ultrasonic cycle at 40kHz for twenty minutes, then check a sample part under a microscope. If the tool marks are still visible, I run another ten-minute cycle. Usually two or three cycles get you to about 0.6 micrometers. The last bit, from 0.6 down to 0.4, is where chemical micro-finishing comes in. I rinse the parts thoroughly, then load them into the electrochemical bath at a current density of about 15 amps per square decimeter for eight minutes. The exact time depends on the part thickness and the electrolyte concentration, so you'll need to run test coupons first. Fifth, I neutralize the parts immediately after the electrochemical step. A quick dip in a sodium bicarbonate solution stops the reaction and prevents any residual electrolyte from sitting on the surface and causing staining. Then a deionized water rinse and an alcohol wash to displace the water and prevent spotting during drying.

The total process for a batch of fifty parts takes about two and a half hours from start to finish, including the intermediate cleaning and inspection steps. A single part done by hand with progressively finer abrasives would take roughly forty-five minutes per piece, which is not sustainable at any volume beyond a handful. The limiting factor in this workflow is almost always the cleaning and media separation step, not the actual finishing. Tiny parts and tiny media are a separation nightmare. You need a multi-stage filtration system with a mesh sequence starting at 150 microns and going down to 50 microns, followed by a sedimentation tank. If you try to reuse media without proper separation, the next batch will get contaminated with spent media and worn-out slurry, and the surface finish will degrade with every cycle. I check my media condition every three batches under a microscope. When the beads start looking angular instead of rounded, or when I see visible dark specks of removed material embedded in them, it's time to replace the media. That's usually after about twelve to fifteen batches depending on the material being finished.