// Showcase

Measuring the axis and the portal – with the machine's own vision system

A machine that has to place to ±1 µm may measure cleanly, but it still moves the way its mechanics allow.

Photonics & Laser Sensors & Metrology

01Starting point

A machine that has to place to ±1 µm may measure cleanly, but it still moves the way its mechanics allow. And about those mechanics, nobody had anything solid. The datasheet for the micro stage that traverses the camera optics in z gives two numbers in the same table: 0.1 µm of repeatability, and ±150 µrad of pitch and yaw. The latter, at 175 mm from the focal plane, works out at ±26 µm. A factor of 260 between the two figures, and no way to tell which one the machine inherits.

02Approach

We used the measuring instrument that was already in the machine: the three calibrated cameras that find the alignment marks. Instead of pointing them at the product, we pointed them at the machine – mark in the image, defined move command, measured displacement. No extra interferometer, no external metrology, just a reversed line of sight.

03Result

Numbers instead of guesses, and numbers that appear in no supplier datasheet. Over its travel, the stage walks sideways by 3.8 to 10.6 µm – periodically, with a period of about 0.5 mm. That is the rolling motion of the guide, not the drive. Preloading bought a factor of three, a different guide type a factor of six. On the portal we found two micrometres of lost motion at reversal: twice the machine's entire placement budget. After stiffening, it was 0.3 to 0.6 µm.

What's inside

Two complete measurement campaigns: the lateral walk of the camera optics over the z travel at four camera positions and in three build states; and the portal, with commanded steps of 10, 20 and 30 µm in x and y, including return to origin, cross-axis coupling and drift at rest – each before and after stiffening, recorded through three cameras at once.

And what happened next

Stiffening improved everything at the same time: drift at rest from 2.6 to 0.3 µm, step fidelity from −2 to −0.4 µm, return error from 19 to 0.4 µm. That is what a stiffness problem looks like once it is finally treated as one. And the method stayed with us: where a machine already carries a subpixel-capable camera system, that is often the only instrument on site able to measure the machine itself.

Get in touch

We are ready for new challenges

You bring the challenge. We develop the solution. From the first idea through concept and development to the validated machine, we guide your project.