I’ve been using the UTM tool for a while now, and frankly, it’s time for an upgrade. While it has its merits, there are a few issues that I’ve run into that I think can be improved. Just a heads-up, my DIY solution might not align perfectly with the original UTM tool’s design, but I believe it could work better in the long run.
Issues I’ve encountered:
Weak Magnets: The 1kg max weight isn’t cutting it for future projects like lens-based weeding. (Wait, isn’t this Z max force, not the tool itself?)
Pogo Pins: While they’re gold-plated, they wear down quickly due to water exposure and mechanical stress.
Pin-Screw Interaction: When the pogo pins connect to the tool’s screws, they scratch easily, and the gold plating wears off fast.
Watering Tool Pressure: The water pressure seems to require an O-ring on the watering tool, but the magnets’ force is too weak to manage this.
Pogo Pin Contact: The pogo pins only connect to the tool via a tiny surface area, which is problematic in terms of power transfer and durability.
Data Speed Limitations: The UTM isn’t reliable at higher data speeds. RS485 might work, but the data rate is underwhelming.
Tool Detection: The current mechanical solution for tool detection is unreliable, and this needs to be foolproof.
Tool Identification Issue: While the tool bay positions should cover this, what happens if you’re off-site for weeks or months? Power loss or an accidental reboot could leave you wondering what tool is mounted. How can this be detected?
The UTM Cable: It’s bulky, unused, overpriced, and not very practical (outdoor, cable chain-compatible, UV-resistant).
My Proposed Solutions:
Wireless Power: A fully waterproof, wireless power system.
Wireless Communication: Using NRF24L01 modules + Arduino Pro Mini + a switched buck converter for tool communication.
Faster Data Transfer: Replace current data communication with a faster interface, like optics or CAN over optical, which could support more sophisticated tools (e.g., 5-axis arms, interchangeable cameras such as IR/PIR/daylight/3D).
Tool Locking System: Incorporate a locking mechanism similar to the Jubilee CoreXY design. Jubilee3D Tools
I’m looking for collaborators who:
Want to share their skills
Prefer working with OpenSCAD
Are familiar with Git
Have time to cooperate on this project
Have strong electrical engineering knowledge
Are willing to invest in the development
Or have better ideas to improve this
Let’s see if we can make this project bigger and better by working together!
I too have been looking at the very same thing.
I have had reliability issues with head changes and have lost the soil sensor in the ground many time as the magnets are weak (or so I thought)
I looked at the complex methods for addressing this issue and made some changes which addressed all of the issues I had and I successfully perform 70 head changes in a soak test before my X2 motor started to fail. (New one on its way).
The changes also solved the soil sensor getting stuck issue and the head is very tight in the UTM.
Believe it or not, I used the existing hardware from the other sensors to make the changes.
The issue is not that the magnets are weak. Far from it. It is the configuration of spacing caused by the choice of cheese head screws in the UTM and distance. By changing the two parameters, it fixed the detection issues and improved the strength.
Please don’t think I am against you ideas. Far from it. Reliability is my number one parameter and the UTM is a weakness.
I agree using pogo pins against a rough screw head is less than ideal. I might suggest that magnet in my new arrangement might solve the strength issue rather than forcing mechanical such as servo driven.
I also recently found out the connection for BC is not buffered with a transistor but instead connects direct to the chip. If we are going to revise things, we should certainly look at the eletronics to ensure a solid detection mechanism
I will be publishing all the details once the testing is complete, along with photos.
The short version is the issue is not the strength of the magnets - just the spacing method. Neodymium magnets are VERY strong, but not in the way they are inserted and secured in the UTM head at the moment.
What I have done is not replicatable at the moment. I have magnets in upside down with weird spacers. I have just received some new magnets from FB and will be rebuilding using standard stock. The magnets were part 1 of the solution.
Part 2 of the solution is a robust mount_tool_safely() LUA code function that checks to see if the head mounted correctly, and remounts is safely if it is not detected in the way it is expected. This stops FB getting confused whether it has something mounted - or not. I have published that code in the forums and it made a difference as well.
I had 70 successful changes then a wheel broke and my X2 motor failed. I have new wheels and am waiting for a replacement X2 motor. Without it, I cannot complete the soak test as X stalls on a regular basis.
I will be publishing the loading code, the soak test code and photos hopefully next week.
It is a bit messy as I mostly used the parts I had lying around, but you will get the idea of how I have solved the issue. I intend to do a cleaned up version but I think I will want to source the right parts first.