formlabs fuse x1 z-axis lifetime testing

during my internship at formlabs, i owned lifetime testing of the z-axis on a next-generation sls printer, the fuse x1. i was responsible for proving the mechanical reliability of a new linear actuator design over the build chamber lifespan, including:

•defining lifetime parameters, load, thermal targets, and acceptance criteria

•full mechanical design and assembly of the high-temperature test enclosure

•heater sizing, pad heaters, and a power distribution box with pid control

•thermocouple instrumentation and data logging

•adapting and running the multi-axis fatigue cycling software

•pre/post precision measurement and failure diagnosis

the z-axis sits in the build unit and actuates the print bed. it has to hold 110 μm layer precision while moving significant mass through severe thermal gradients. the motivation for this test was a new linear actuator between Proto0 and Proto1 that eliminated active cooling. the reliability question was whether precision and motion would hold up over a full product lifetime at temperature.

the key failure modes i cared about were drift, loss of precision, and the actuator binding or stalling. three z-axes were mounted in a sheet-metal enclosure, loaded, heated, and cycled for a full lifetime of travel so we could compare commanded motion accuracy before and after.

first, i defined the test envelope from product life assumptions. i worked with the product team to define this, based on expected lifespan (years), and number of full builds per year, to derive the total lifetime travel. i determined the load corresponding to a worst-case packing density with nylon 12, our most popular sls powder.

this was a true worst-case test for several reasons. first,temperature was controlled by holding the top lead-screw bearing at 160°C (the manufacturer maximum) even though prior testing suggested ~120°C in real use. second, we assume that the printer is near 100% packing density. acceptance was based on deviation from a 50 μm commanded move: with a 110 μm layer height, anything under ~10 μm (1/10th of a layer) was considered acceptable.

from there, i designed and built the enclosure and heater system. a first-principles heater calc sized the power needed to bring the actuators up to the thermal target. i started with 180 W pad heaters. the first issue was that they couldn't reach a steady 160°C. the second issue is they didn't have good contatct (and conduction) with the enclosure roof and as a result, one of them caught fire. i resolved these problems by switching to 360W pad heaters and added a mounting plate, connected to the roof with thermal paste. the heaters were controlled with inkbird pid controllers and solid state relays, packaged in an acrylic box, separate from the motor drives.

the system was designed to cycle with powder exposure inside the enclosure, but testing was done without powder, since its contribution can be evaluated separately in a real build chamber and is largely independent of loaded cycle life under temperature.

fourteen thermocouples covered top bearings (control + logging), bottom bearings, motors, and enclosure air. these channels were logged continuously through phidgets while the pid loops closed on the top-bearing control thermocouples.

for motion, i adapted an existing fatigue script to cycle three axes in parallel with resume-from-log support across multi-day runs. precision was measured separately with a dial indicator on the rigid z-axis support: each axis was commanded through a down-up move and deviation from the 50 μm relative step was recorded before and after lifetime cycling.

after a ton of loaded high-temperature cycles, average deviation across the three axes actually reduced by about 1 μm on the 50 μm commanded move, meaning the z-axis repeatbility improved, and this was statistically significant when the axes were analyzed together. the root cause of the improvement is unclear and possibly within measurement noise. also, i observed grease running down the z-axis but there was no apparent impact. the takeaway was clear: very little risk of z-axis wear impacting print quality over the build chamber lifespan.

one issue observed during testing was that the motor <-> axis pulleys slipped and lost coplanarity, causing the belt to slip off and wear. the cause is that when i was installing the axes, i aligned the pulleys by eye and didn't precisely position them. i proposed a fixture to resolve this in manufacturing.

given the positive test outcome, my recommendations coming out of testing were simple: to use higher-temperature lubrication that holds viscosity over life, and a simple pulley alignment fixture.

fun fact: this was the highest potential-energy testing system in formlabs history (1.1 kJ!)