Build part parameters · Hatching › Volume / hull hatching · group grp_power_opt
Power optimization
On a short hatch vector the laser comes back to its last line after a fraction of a millisecond,
while the metal there is still hot. With full power the heat builds up: the melt pool grows, and pores, swollen
edges and spatter follow. The power optimization gives every hatch vector its own power: less on short vectors,
full power on long ones. Speed and hatch distance stay the same.
In Advanced Toolpath Utility
Try it yourself
Open the parameter profile of your part (Open Profile in the Exposure Preview), expand
Hatching › Volume / hull hatching and set Power optimization to on. Rebuild and look at the
short vectors of your part.
To see the power, the demo uses its own colour mode of the Exposure Preview: Configure… next to
Color mode, a colormap of the point attribute power with the fixed range 150 … 280 W.
Before the change none of the hatch vectors of layer 50 shows the low-power colours; after the rebuild the
vectors near the tips do. The demo part parts/power_wedge.stl is generated by
tools/make_parts.py.
Deep dive
A closer look at how the power optimization works, from the heat build-up on short vectors to the power it
gives every hatch vector.
Try it
laser power, 150 W … 280 W (like the colour mode in the video)
Short vectors, short return time
The power wedge is a right triangle, 36 × 18 mm. Its hatch vectors get shorter towards the tips, down to a
few hundredths of a millimetre. What matters for the heat is the return time: how long the laser
needs until it is back next to the start of a vector. The buildstyle counts marking time (length ÷ speed),
the mark delay, one jump of a hatch distance and the jump delay. At 1200 mm/s a 0.5 mm vector returns after
0.49 ms, a 10 mm vector after 8.4 ms.
From time to power
The dialog asks for two vector lengths, because lengths are easier to imagine than times. The code turns them
into the two return times and interpolates the power linearly in between. Up to 0.5 mm the vectors get the
minimum power, from 10 mm on the maximum. Because the return time grows linearly with the length, the
Power curve is a straight ramp between the two lengths.
Segment splitting
Without splitting a vector gets one power from its own length. That is an estimate: in a meander the laser
comes back right next to the end where it turned, but only after two vector lengths at the far end. Switch
on Segment splitting: every vector is cut into segments of at most the segment length, and each
segment gets the power of its real return time, from the moment the previous line passed the closest point
until this segment is exposed. In the wedge the colour now changes along every line: the turning ends get
less power, the far ends more. If the previous line is more than 1.5 hatch distances away (no hot
neighbour), the segment gets the maximum power.
Merge segments with equal power joins neighbouring segments of the same power back into one vector.
The powers are rounded to 0.1 W, so long stretches with the minimum or maximum power become single vectors
again. Switch it off and compare the number of segments below the wedge. A shorter segment length follows
the return time more closely but gives more data.
One timeline per laser
The power is set after the build order and the
laser assignment, so it follows the vectors in the order the lasers really
scan them. All hatch blocks of one vector type on the same laser form one timeline, across islands and,
with the Global build order, across parts. With segment splitting a segment can therefore find its hot
neighbour in the block the laser exposed just before, for example where the laser assignment cut one block
into runs on lasers 1, 2 and 1 again. Without splitting the timeline changes nothing: every vector keeps the
power of its own length.
The splitting in this view treats the wedge as one meander block in line order, at the volume
hatch speed of 1200 mm/s. If you use jump delay scaling, the
short jump of one hatch distance also gets the shorter jump delay, so the return times shrink a little.
Set the maximum power to the power of the vector type in the Scanning group (280 W for the
volume hatch). The buildstyle does not check this; a higher maximum would raise the power of long vectors.
Parameters
Parameter
Default
Meaning
Minimum power
150 W
Power of the shortest vectors.
Maximum power
280 W
Power of long vectors.
Vector length for minimum power
0.5 mm
Up to this length: minimum power.
Vector length for maximum power
10 mm
From this length on: maximum power.
Segment splitting
off
Cut vectors into segments, each with its own power.
Segment length
0.5 mm
Length of these segments.
Merge segments with equal power
on
Join neighbouring segments of the same power again.
The group exists under Volume, Core, Downskin and Upskin hatching, each with its own settings.
In the code
Follow the settings from the dialog to the power attribute of every vector in How the code
works, and change the defaults yourself with the recipes below.
How the code works
Eight steps from the dialog to the power of every vector. Scroll through them: the drawing
on the left shows what each piece of code does. Highlighted lines are the ones the text talks about; the file
names link to the source.
laser power, 150 W … 280 W
Change it yourself
How to try a code change
Work on a copy or a git branch of the buildstyle project.
Open the project folder in an editor such as VS Code. ATU writes jsconfig.json and type
declarations into the folder, so the editor can offer code completion.
Make the edit, save, compile the project in ATU and rebuild the exposures.
Check the result in the Exposure Preview or the 3D View, like in the video above.