Suzie Model One - CNC Machine

Introduction

SUZIE MODEL ONE was made to cut PCBs. The objective was to create a rigid machine, accurate and as much automatic as possible. Right now the only manual process needed is changing cutting bits and placing the Z-probe in the right place.

This machine is the result of a trial and error process where several parts were redesigned and rebuilt to get the best result possible. The main problem was rigidity: when the machine is not hard enough it will bend as it is cutting and the result will be a failure. Jitter when cornering is the other main problem. It is possible to reduce speed to avoid that, but time is money, and some kind of cuts work better at the right speed, not slower.

To see more check #suziecnc on Instagram.

The next chapters will show and explain why soSUZIE MODEL ONE was designed to cut PCBs. The goal was to create a rigid, accurate, and as automatic a machine as possible. At the moment, the only manual tasks required are changing the cutting bits and placing the Z-probe in the correct position.

This machine is the result of a trial-and-error process, in which several parts were redesigned and rebuilt to achieve the best possible outcome. The main issue was rigidity: if the machine isn’t stiff enough, it bends during cutting, leading to failure. Another major problem is jitter when cornering. Reducing the cutting speed can help, but time is money, and certain cuts perform best at the right speed, not slower.

Dimensions

External size:

• 46cm width
• 62cm length (43cm not working)
• 58cm height (counting with the cable tube)

Working area is approximately:

• 20cm width (X)
• 20cm length (Y)
• 6cm height (Z)

Suzie's Views

XY Axis, Stepper Mounting and Backlash

Backlash "...is a clearance or lost motion in a mechanism caused by gaps between the parts..." - Backlash in Wikipedia

SUZIE has zero backlash on both the X and Y axes.

No compensation is applied in the firmware, and it always cuts with the correct dimensions. This is achieved by using belts instead of lead screws.

Lead screws require a small clearance between the screw and the nut to operate without friction, and that clearance must be compensated for. That’s not a problem by itself, but over time, as the nut wears out, constant backlash develops, which must be corrected in the firmware. If the lead screw itself wears, the issue becomes even more complex because a CNC machine uses different segments of the screw under varying speeds and loads. This causes uneven wear, making it impossible to compensate accurately.

The belts, on the other hand, are strong enough to apply the necessary force even for cutting metal, and when properly tensioned, they exhibit virtually no noticeable elasticity.

Both of SUZIE’s horizontal axes use the same method to transfer force from the stepper motor to the moving part, but differ in how the belt and stepper are mounted. In both cases, the belt does not form a closed loop like in a 3D printer; such a system would take up more space. Instead, the belt is open, fixed at both ends, and the stepper motor runs along it like a rack-and-pinion system.

The X car has the stepper attached and the belt is fixed on the machine. The stepper moves with the car.

Y Tray

The Y tray uses the same concept but with a fixed stepper and a moving belt; exactly the opposite of the X axis.

This design offers another advantage: the maximum length of the Y tray is virtually unlimited and can be adjusted as needed without altering the machine’s overall design. This is also possible because the rail is mounted on the moving tray, while the bearings are fixed to the base of the machine’s structure.

This picture shows the Y bearings support part. This one (on the left side), and the other one on the right (where only a part of the top bearing is visible) are aligned with the X carriage movement and positioned directly below and centered with the spindle.

This means that all downward and sideways forces are transmitted to the base in compression. This allows the support to remain very rigid and capable of withstanding significant force.

As we can see in this picture, the length of the tray is not mechanically critical and doesn’t require a strict limit (well, not entirely true). The limiting factors could be the software, tray fabrication, or the maximum belt length.

For example, it would be perfectly possible to install a one-meter-long tray for a special task.

Z Axis and Lead Crew

Using a belt configuration on this axis presents a major problem. The stepper motor needs to remain constantly under load, and in the event of a power failure, the spindle could plunge into the workpiece, potentially damaging the bit, the motor, or even the work surface and tray.

The solution here is obvious: use a lead screw.

The lead screw and nut were custom-built for this machine. The screw has the required length, space for bearings, a locking nut, and a connection to the stepper motor. It was made on an old manual imperial lathe configured for metric threading. The only issue was that the synchronization wheel doesn’t work with metric threads, so a very unconventional method was needed to machine this screw. The nut was 3D printed.

This system has no backlash because the 3D-printed nut was made with zero clearance (it only runs smoothly thanks to plenty of oil). However, the entire XZ carriage assembly is quite flexible. What actually keeps the spindle in position is its own weight. If pushed upward by hand, it can move about 2 to 3 millimeters due to slight bending of the rails.

The only mechanical way to fix this is by increasing the machine’s rigidity. Another solution is to reduce the plunge speed. Keep in mind that cutting primarily occurs laterally along the X and Y axes, so as long as the plunge speed isn’t too high, the weight of the Z carriage and spindle will naturally bring the cut back into alignment.

In SUZIE’s case, the maximum plunge speed works fine for cutting soft materials — no slowdown needed — and I only had to reduce the speed when cutting metal (brass).

3D Printed Parts

This machine was only possible thanks to 3D-printed components. The tools I had access to were not suitable for making complex parts by hand — such as the stepper mounts, for example. Every part was designed and printed multiple times to accommodate specific functions or to improve structural performance.

These parts slightly reduce the machine’s overall rigidity but allow it to be much more compact.

Arduino, GRBL, Shield, Spindle and Probe

SUZIE's brain is an Arduino Uno, running the GRBL firmware.

There are several commercial shields to GRBL but in this case was more a challenge case. The first shield I have built I used the L298N and L297 combination to drive the steppers. It was hot and loud. Later with the Pololu A4988 drivers, the system was redesign to be improved.

The circuit is built on a prototype board and provides nearly all the GRBL connections. Since adding this board, compared to the previous version using the L298N and L297 combination, the machine has been able to operate for much longer periods without overheating, and it runs more quietly thanks to the micro stepping drivers.

The firmware is currently updated to GRBL 1.1f and runs very smoothly.

Another recent addition was a proper brushless spindle motor (dmw57314). It is now controlled by the Arduino and is much quieter compared to the Dremel-like drill I used in the first iteration.

This probe was designed to measure only the Z axis. A probe for the X and Y axes is more complex to build by hand and will be a challenge for another time. With this probe, SUZIE can precisely locate the cutting bit’s tip, a crucial step in PCB milling.

It works as an always-ON button: when the bit presses it down, the two wires disconnect and signal GRBL of the position. Its repeatability is highly accurate.

Conclusion

SUZIE MODEL ONE was an excellent way to learn how to build a CNC machine. Many problems were solved, many features were improved, and I can say for sure: it’s done!

Any further improvements would no longer make sense within this structure.

SUZIE MODEL TWO will be designed to expand on the strengths of this model and radically improve on its weaknesses. One thing is certain — it will be metal!