An Improved Height Difference Based Model of Height Profile for Drop-on-Demand 3D Printing With UV Curable Ink

TL;DR

The paper addresses accurate height profile prediction for drop-on-demand UV-curable ink 3D printing, especially for 2D patterns. It extends prior height-profile models by deriving height from volume and area via $h = c \frac{v}{a}$ and by modeling $\Delta v$ and $\Delta a$ as piecewise functions of height differences within a 3×3 ROI, with empirically learned coefficients. The empirical determination of $m_v^+$, $m_v^-$, $m_a^+$, and $m_a^-$ from multiple patterns (with bootstrapping) demonstrates RMS errors consistently below 7% across six 2D patterns and shows improvement over graph-based models while matching prior 1D results. The approach promises real-time applicability for process control in UV-curable ink printing by providing accurate, volume-conserving height predictions for complex patterns.

Abstract

This paper proposes an improved height profile model for drop-on-demand 3D printing with UV curable ink. It is extended from a previously validated model and computes height profile indirectly from volume and area propagation to ensure volume conservation. To accommodate 2D patterns using multiple passes, volume change and area change within region of interest are modeled as a piecewise function of height difference before drop deposition. Model coefficients are experimentally obtained and validated with bootstrapping of experimental samples. Six different drop patterns are experimentally validated. The RMS height profile errors for 2D patterns from the proposed model are consistently smaller than existing models from literature and are on the same level as 1D patterns reported in our previous publication.

Figures (4)

• Figure 1: (a): Single drop with radius $r$ on $3 \times 3$ cell. (b): Zoomed-in section of cell (2,0) in (a), illustrating how to count area covered by inks. (c): Height profile comparison of center row. Proposed height profile is in purple, while spherical cap height profile is in light blue.
• Figure 2: The experimental setup. Gate closes when UV light is on to protect the dispenser head.
• Figure 3: Besides drop patterns with 0 and 8 prior drops within the region of interest, 6 more drop patterns are used to validate the model. From (a) to (f) are the drop patterns with 1 to 7 prior drops. The blue disk represents the new drop while the black circles represent the existing drop before the deposition.
• Figure 4: (a): One of the sample after a drop (3rd pass) is deposited at the center. Red lines are the cell boundary. The blue line indicates cross section measurements to be shown in (b). (b): The blue line shows the measured height along the blue line in (a). The red line shows the predicted cell height.