Technical Guide · Diode Laser · LaserBase · Mechanics Series

Raster Engraving Speed

Theoretical vs. real maximums — why does it take longer than you'd expect?

The basic question

A common experience: on paper the machine reaches 10,800 mm/min, yet a 30×30 cm raster engraving at 318 DPI unexpectedly takes a long time. The configured speed isn't equal to the net engraving speed.

Configured engraving speed

What you set in the software. The firmware's $110 (X Max rate) is its upper limit — the machine won't move faster than this on the X axis.

Sustained net engraving speed

What the machine can actually, sustainably engrave at a given DPI and workpiece size. This is always lower than the configured value.

The bottleneck is acceleration ($120 / $121), not the max rate. It doesn't matter how high $110 is — if acceleration is low, on short raster lines the machine physically can't reach the configured speed.

What reduces net speed?

In raster engraving, the controller works line by line. At the end of every line it reverses direction — this means acceleration, deceleration, and an overscan zone. Beyond this, several other factors play a role:

FactorEffect
Acceleration / deceleration ($120/$121)A ramp up-down cycle on every line; the target speed has to be reached within the line's length
Overscan zoneThe head overshoots past the end of the line, decelerates, reverses — this costs time and area
PWM update limitAt high DPI + high speed, the firmware can't generate PWM fast enough
Planner buffer limitGRBL only buffers a limited number of moves ahead
Dithering processingJarvis/Stucki/Floyd need more computation — the software feeds lines more slowly
Serial communication (baud rate)An independent bottleneck at high DPI and with complex dithering

Short scanline effect: on a small workpiece, the head in many cases never even reaches the configured speed, because it has to decelerate immediately. With a small object + high DPI + many direction changes, the gap between configured and net speed is largest.

Key GRBL parameters

Error 12: if the $100 step/mm × speed combination exceeds the controller's maximum step rate, GRBL returns Error 12: "Step rate exceeds the maximum step rate". This is a controller limit, not a mechanical one.

Mechanical setup — step rate limits

Step rate (steps/second) is the product of $100 step/mm and speed. On classic 8-bit GRBL controllers, the practical limit is around 30,000–40,000 step/s.

MotorPulley$100 step/mmStep rate @ 10,800 mm/min
1.8°GT2 20T8014,400 step/s
1.8°GT2 16T10018,000 step/s
0.9°GT2 20T16028,800 step/s
0.9°GT2 16T20036,000 step/s

The 0.9° + GT2 16T combination can approach the step-rate limit around 10,800 mm/min on classic GRBL controllers. On grblHAL (ESP32/STM32/Teensy) systems, this limit is much higher.

Acceleration values by category

Typical $120 (X axis) ranges for diode laser systems:

CategoryAcceleration ($120)Typical effect
Weak / basic setup300–800 mm/s²Slow, long overscan, never reaches target speed on short lines
Normal raster1,500–3,000 mm/s²Average performance, most diode laser machines run stably in this range
Aggressive / fast5,000+ mm/s²Close to the mechanical limit — needs belt tension, frame rigidity, and motor torque to support it

High acceleration only makes sense if the mechanics can handle it stably: taut belt, rigid frame, sufficient motor torque. Otherwise the result is skipped steps and banding.

Baud rate — the lesser-known limit

At high DPI and with complex dithering, serial communication speed can become an independent bottleneck. The controller has to process many small PWM changes per line — if the serial channel can't keep up with the data-transfer demand, the machine slows down not because of the mechanics, but because of the data transfer.

115,200 baud (default)

The default value on most boards. Above 254–318 DPI, with dithering, it can become throughput-limited.

250,000 baud (fast)

A noticeable difference at 318+ DPI, with grayscale and dithering. Not every board supports this stably — check the documentation.

Warning: with unstable communication, a higher baud rate can make things worse. Test with a simple engraving before switching.

Realistic net speed — by baud rate

Estimated, typical ranges. These can vary machine to machine — acceleration settings, image content, and the dithering method all have an effect.

DPI115,200 baud250,000 baud
254~8,400 mm/min~10,200 mm/min
318~7,800 mm/min~9,600 mm/min
400~6,900 mm/min~8,700 mm/min
508~6,000 mm/min~7,800 mm/min
635~4,800 mm/min~6,600 mm/min

Highlighted: 318 DPI — the range most commonly used with diode lasers. A gap of nearly 10–30% between the configured speed (e.g. 10,800 mm/min) and the net productive speed is typical.

10,800 mm/min configured speed ≠ ~7,800–9,600 mm/min net productive speed / 318 DPI

Two users' machines — one at 6,000 mm/min with lower power, the other at 10,800 mm/min with higher power — can give nearly identical engraving times. This isn't a bug: it's the combined, system-level consequence of acceleration-deceleration cycles, overscan, baud rate, and planner limits.