New Project - Universal Adapter

I am starting work on a new project creating a board that will act as an interface between the Glowforge’s proprietary hardware and common 3rd party CNC driver boards.

General idea is a replacement board for the existing control PCB to provide simple terminals to connect to the fans, steppers and laser controls. This would then be connected to any of the many available boards that run GRBL or Marlin or whatever. It will also have the option to install a small SoC board like a RPI Zero 2 W (or similar) to get access to the cameras and lights, etc.

I’m looking for input from the community to see what features and capabilities there is interest in.

Thoughts?

A cable, bus or some flavor of port that would allow connecting to the control boars EXTERNALLY so that cooling, debris built up, trouble shooting etc. would not require constantly digging into the enclosure would be nice :grinning_face: . Also, although you haven’t mentioned it, would be happy to donate $ to help offset costs…

My intent is to keep the ability to use the internal cameras, and those run on very high speed signals that are picky about how/where they are ran. So, for the adapter board at least, it will need to remain inside. The controller/driver board you connect to it could certainly go where ever you wish (inside, outside, in another room :grin: ).

To your point, a conformal coating would probably be a good idea on the finished product, as that is notably missing on the factory boards. At least on any of the units I have seen in the wild.

Just wanted to jump in here and say that as someone who ended up with a secondhand Glowforge and hate their business model/software, I strongly support this project and the work you’ve done. As more of these end up on the used market, go out of warranty, or are picked up by people who don’t know any better, having options to make them usable is huge. I feel like being able to offer a plug and play board that will link to lightburn or other available softwares would be extremely popular. I’d also be happy to donate a bit to help with some of the costs, or lend other assistance as needed.

STATUS UPDATE:

After much noodling, I am taking a slight detour.

The GF has a lot of bespoke hardware that complicates the design of a “simple” adapter that would let you connect it to one of the many generic controllers on the market. The main hurdles are the cameras and the laser head. There is just no easy way to expose these to a generic controller, as there isn’t a single one that supports either. This would result in some sort of hybrid arrangement where there is a controller of some sort on the adapter that provides an interface for these specific components. That ends up requiring a web panel you interact with to get access to them (and other device features). This quickly ceases to be a “simple” adapter.

Not being a simple adapter is fair enough, but it quickly leads me down the same road I went before—with a whole new control board. I learned from that effort that there just wasn’t any real interest in that. At least, not enough interest to make the effort worth it.

That led me right back to where I left off last time: custom firmware for the factory control board that provides a GRBL interface so you can run the GF with your tool of choice. Some people on other forums mistakenly suggest that the reason the project ended was because it couldn’t be done. That is dead wrong. It was just a lot of work, and I lost interest in continuing it. What I had accomplished was an initial port of GRBL to work with the SDMA hardware in the GF (the same hardware component GF uses for their motion output). That was somewhat functional, but it needed a lot of work to dial in the timings and get all the kinks worked out. It was tedious.

Times have changed since then. A well-established port of GRBL with a hardware abstraction layer has now emerged. It effectively modularized the architecture and made it much easier to adapt for different hardware. There is even a port to the venerable RP2040 that does a somewhat similar implementation to the iMX6 the GF uses. Also, the advances in AI coding agents (something unheard of just six years ago) have become a force multiplier for me. While far from perfect and still needing a lot of guidance, they have taken over the tedium and let me focus on what is important.

These developments have led me in a new direction: finishing the Forgefirm concept. A custom firmware image that will run on factory control boards and provide a GRBL-compatible interface with an additional instruction set to control GF-specific hardware. There will also be a web interface to see the status of the hardware, provide streams from the cameras, and allow the user to tweak the settings of the machine.

I have completed a significant amount of the needed groundwork. I have updated Forgefirm to work with the latest LTS kernel and Yocto release. The GF kernel module has been updated to handle the numerous changes between the old and new kernels. My Python libraries have been updated to work with the driver changes and the latest GF cloud API (you can still use their web app with the new firmware if you like). And most importantly, I have a grblHAL hardware component that drives the GF hardware.

The full solution is not yet ready for release, as I am still working through validation and tweaking. But it is coming along nicely. I can currently control and run the GF with LightBurn (motion only—I’ll light the laser when I have 100% confidence in the motion). When the firmware is ready for release, it will still require you to break out your soldering iron to install it, as you will need console access. Maybe someday in the not-so-distant future we can convince GF to sign third-party firmware images so they can be installed via their local recovery mechanism (it exists—it uses the same tooling as the initial setup process).

I know there are other issues with the GF that this direction seemingly will not solve, like dead power supplies and the lack of expansion for connecting things like rotary drives. Well, I have good news… I will be releasing an adapter that will allow you to connect third-party power supplies. This will only work with Forgefirm, as the factory firmware expects feedback from the bespoke power supply and won’t work without it. Additionally, I will be releasing an I/O expansion board that will enable driving of external devices like rotary drives and adding homing sensors.

I don’t have a solid timeline for the release of any of these things, as the time I can dedicate to this is hard to predict. Hopefully another month or so… Stay tuned.

I have all the basic functionality working.

Both cloud and grbl modes are working well.
A thorough run down of all the safety systems is needed, along with some basic usability testing, and it will be ready for a beta release.

I hope to get this all completed by the end of this weekend.

1 Like

Everything (that I can test with my basic non-HD model) is functioning, and well.

Cloud mode works as expected.

GRBL mode is working perfectly, but it needs a lot of tuning of the defaults. Things like max acceleration and velocity, variable power curve, etc. This will be tedious and time consuming, but necessary before a release. If it doesn’t “just work” out of the box, people get feisty.

Progressing…

1 Like

Can’t say it enough, but THANK YOU for all your time and effort.

1 Like