My friend Nate got an XYZprinting DaVinci 1.0A 3D printer for free. Of course, it was free for a reason, having been passed between several friends and acquaintances before ending up in Nate's hands. Judging by the liquid stains and corrosion on some components, it may have been used as a drink table at one point, or left out in the rain. Unsurprisingly, some work was required to make it functional again.

Among other things, the cooling capabilities of the printer were inadequate for printing with PLA. The only fan it came with was mounted on the hotend's heatsink; there was no layer cooling at all.

Adding a layer fan aimed at the nozzle and another, larger fan to push air through the enclosure drastically improved PLA print quality. These new fans were put in strictly as a test, and were powered externally using a USB power supply and PWM controller. Ideally, all fans in the printer would be powered and controlled by the printer itself.

Using the open-source Repetier firmware and unpopulated components on the printer's controller board, I added two additional fan drivers to the board, both of which could be controlled through the M106 G-command.

Overview

The DaVinci's controller board is based around an Atmel ATSAM3X8E ARM microcontroller. When the system is reset by jumping the J37 header when running the stock firmware, it presents itself to a USB host as an Arduino DUE. The Arduino IDE can then be used to flash its firmware. (A reset is not required to accomplish this if running the Repetier firmware).

The board appears to be used across different models of printers. In the case of this 1.0A board, with silkscreen text F1.x F2.x Engine board F1, 3DP03-ES0007, 2014/06/09, roughly half of the component footprints are actually populated. The unpopulated components provide features such as a second extruder and a 3D scanner.

The second extruder's unused fan is a pretty straightforward choice for adding another controllable fan. The circuit is more likely than not identical to the first extruder's; the components just need to be identified and populated.

To add a third controllable fan to the printer, a different feature of the board would have to be commandeered. Looking at the unpopulated headers, the 'Extruder 2' heater connector caught my eye. The circuit that controls the heater already switches 12V at a relatively high current, so driving a 12V fan should be well within its capabilities. Of course, the complete circuit would have to be traced to confirm this theory.

Controller Board Photos

Circuit Traces

The first step of the modifications was to trace the relevant portions of the controller board and figure out how the existing hotend fan and heater drivers worked. Some things to note about the process:

• Capacitors cannot be measured accurately in-circuit. The values I picked were educated guesses driven by a bit of electrical design knowledge:

  • A multimeter was used to probe capacitance at either node after tracing the remainder of the circuit, so I had an idea of what other components may influence the reading. I generally assumed that the measured capacitance would be higher than the component value, and component values were members of the E24 series.

  • In cases of two parallel decoupling capacitors, I assumed that one acted as a bulk capacitor and provided the majority of the total capacitance, while the other was bypass capacitor and an order of magnitude smaller.

• SMD fuses on this board use "R" as their reference designator instead of "F" for some reason. All fuses appear to be from the Littelfuse 437 series. R47, R57 have a "K" marking, indicating a 1.5A rating. R56, R271, R272 have a "T" marking, indicating a 5A rating.

• There are networks of a resistor (R22, R198), capacitor (C24, C132), and diode (D4, D6) that appear to be part of each MOSFET driving circuit for the extruder 1 and 2 heaters. I wasn't confident that I had traced them correctly enough to understand their purpose. Protecting a MOSFET from back EMF using a diode is a common practice, but the circuit I came up with made no sense to me. They've been left out of the schematic and I didn't populate them for the Fan 3 modification. If Q105 releases magic smoke during normal operation, these components may become relevant.

Fan, Thermistor, and Sensor Circuits (fan-thermistor.sch)

Extruder 1 Circuits

Extruder 2 Circuits

Common

Extruder Heater Driver Circuits (heater.sch)

Extruder 1 Heater Circuit

Extruder 2 Heater Circuit

Buzzer (buzzer.sch)

I traced the buzzer circuit because I wanted a quick way to disable the buzzer; the "chime" the firmware plays during startup quickly got old during testing. It was disabled by shorting the base (pin 1) of Q3 to GND. I used a pin of the J32 header (either pin 7 or 12, depending on how you count) as the GND signal. I undid this mod towards the end of the project.

Extruder (Remote PCBA) (extruder.sch)

The extruder assembly contains its own PCBA, featuring an optical sensor for filament detection. It also breaks out the signals for the hotend fan and thermistor. This board wasn't modified for this project, but it was necessary to trace where the fan and thermistor signals went.

Resources - Hardware

Type	Name + Description
Gitlab Project	davinci-mods/davinci-circuit-traces davinci-circuit-traces Git repository

Downloads - Hardware

Type	Filename + Description	Date	Size	SHA256
ZIP Archive	davinci-circuit-traces-1.0.0.zip KiCAD project files	2022-06-12	238.1 KiB
PDF Document	davinci-circuit-traces_1.0.0.pdf Schematic PDF	2022-06-12	303.1 KiB

Hardware Modifications

Fake Thermistors

If the controller board is powered up without anything connected (i.e., only power and USB for bench testing), the Repetier firmware asserts a temperature fault and refuses to process any G-commands.

To appease the firmware with fake thermistor readings while testing software modifications, I created short cable assemblies to put 120kΩ resistors in place of thermistors.

For J10, the heated bed sensor, the resistor was across pins 1 and 2. For J31, the Extruder 1 hotend sensor, it was across pins 3 and 6.

The readings are scaled differently for each of these sensors, resulting in this resistance being interpreted as substantially different temperature values, but it's enough to fool the firmware.

Fan 2 Modification

To enable the second extruder fan, components were simply populated to match the existing fan circuit for the first extruder.

Fan 3 Modification

To turn Extruder 2's heater into a fan driver, all components to drive Q105 directly were populated. Instead of matching the existing extruder circuit, which appears to use the LM2903 comparator and inverter to convert the 3.3V signal to 5V, this is bypassed with a 0Ω resistor.

Fan Connections

Fan 2 Connection

For the Fan 2 connection, I cobbled together a pigtail to adapt a common Molex 4-pin PC fan connector to the JST PH header on the board. This used a PC fan adapter from my junk bin, which I believe was included with one of the numerous Noctua fans I've bought over the years.

Fan 3 Connection

For the Fan 3 connection, I would've made a similar pigtail to adapt a standard PC fan connector to the JST VH header on the board. However, due to the Great Part Shortage of 2021/2022, the crimp terminals for VH connectors were nowhere to be found.

As a stopgap, I butchered another PC fan adapter and soldered its leads directly to the board header. I'm not proud of it, but this will be fixed when the terminals are available again, I promise.

Completed Modifications

Software Modifications

The software side of this project consisted of two distinct changes:

1. Enable existing Fan 2 control support
2. Create another instance of a fan controller for Fan 3

Most of the work to enable Fan 2 was already done for us - it was simply a matter of setting the right compile-time options in Repetier's Configuration.h.

#define FEATURE_FAN2_CONTROL 1
#define FAN2_PIN ORIG_FAN2_PIN

Adding third fan controller instance was a bit more involved. After tracing the connection between Q105's gate and the microcontroller, I had to cross-reference the physical LQFP-144 microcontroller pin with the Arduino DUE pin the software uses.

Once the pin for Extruder Heater 2's driver was established, I proved I could control it by "twiddling" it in the software's main loop. After that, I duplicated every instance of the management and control code for Fan 2, creating an identical instance for Fan 3 with this pin as its output. Just like the others, Fan 3 could be enabled in the configuration:

#define FEATURE_FAN3_CONTROL 1
#define FAN3_PIN ORIG_FAN3_PIN

All three fans could then be controlled using the M106 G-command,

M106 P[0-2] S[0-255]

where P0, P1, and P2 correspond to Fan 1, Fan 2, and Fan 3. This can be tested using Repetier Host:

Resources - Software

Type	Name + Description
Gitlab Project	davinci-mods/davinci-repetier-firmware davinci-repetier-firmware Git repository

Downloads - Software

Type	Filename + Description	Date	Size	SHA256
Binary File	Repetier.ino.arduino_due_x.bin Firmware image	2022-07-08	210.3 KiB
ZIP Archive	davinci-repetier-firmware-mod_fan3_1.0.zip Source code archive, with Arduino project	2022-07-08	5.4 MiB

Closing Notes

These modifications almost work as intended. Before I call this project done, there are some major things that remain to be addressed.

Failed Fan 2 BJT

The Fan 2 driver didn't last very long. After being used to power a 5V Noctua fan for a short period, it failed to pass any current to ground.

The transistor was most likely damaged by the back-EMF created when the fan's power signal is chopped by the pseudo-PWM driver (see below), rather than modulating its voltage, like it was designed for. A possible solution is to add a flyback diode to the circuit to clamp the induced voltage spikes.

Interestingly, the Fan 1 driver has seen plenty of runtime and hasn't suffered a similar fate. I suspect some fans are more tolerant of being driven this way.

Limited Fan 3 Current

The Fan 3 driver struggles to drive some larger 12V fans at less than 100% "duty cycle" (again, see below), and even then doesn't seem to deliver enough power to run them at full speed.

My theory is Q105 isn't fully turning on when driven by the microcontroller directly. The gate-source threshold voltage (\(V_{GS(th)}\)) of the replacement³ MOSFET is 2.5V⁷, and 2.3V⁸ for the original, so one would think a 3.3V CMOS signal would be more than enough to push it over the threshold and turn the transistor on.

Looking at the schematic of the original fan driver, however, a comparator serves as a level converter to control the MOSFET's gate with 5V. Replicating this circuit in its entirety, rather than bypassing it with the strapping resistor, is probably the next thing to try.

Not Quite PWM

Finally, the printer's fans aren't driven with PWM like one might expect, because it isn't available on this board without a major refactoring of the Repetier firmware. Hardware PWM is available on the ATSAM3X microcontroller, just not on the pins used to drive Fan 1 (PD0) and Fan 2 (PC26) ⁹. PWM is more feasible for the Extruder Heater 1 (PA13) and 2 (PB12) drivers, whose pads can be assigned to PWM peripherals ¹⁰.

So instead of PWM waveforms, the signals driving the fans are a bastardization of pulse width and frequency modulation, bit-banged in a rather lengthy timer ISR, resulting in frequency changes staggered throughout the 0-255 speed range.

This isn't necessarily a problem on its own, but the code that implements this is more complicated than just setting up a few peripheral control registers and adjusting them as needed. In its current form, this algorithm cannot drive any fan at 100% "duty cycle".

Unless I'm struck by some unprecedented motivation to fix what'll probably become an entire project's worth of code, it's unlikely I'll attempt to address this.