Yes - using POE for power and data crossed my mind! But ultimately I ended up only towing fishing line which I then used to pull the cable, which was probably easier than directly towing shielded cat 6!
kizzard
joined 1 year ago
I unimaginatively called it Duct Rover 9000 but I like that, or Mclane :D
that's great information - thanks!
How long depends on your humidity. If you live somewhere with high humidity, PLA or PETG filament can go bad in a couple of weeks or less. In a dry location, maybe never for PLA. Nylon will go bad anywhere in a matter of hours. It's not easy to tell if filament is wet - weighing filament can tell you about the moisture gain/loss, but of course doesn't work when you are actively consuming it. When filament gets wet, prints start to get bad in various vague ways that can seem like printer tuning issues. Sometimes it's bad bed adhesion, sometimes poor first layer quality, stringing, or bad overall print quality.
If you want a good storage solution you need an airtight box. I use Iris Weathertight storage boxes with 1lb desiccant canisters and a humidity monitor. The 62qt box can fit 16 spools and room for the desiccant.
Dude, awesome post! In the last couple of weeks I too have switched to FreeCAD. The real turning point for me was finding RealThunder's fork which is infinitely more usable than the main fork due to stable geometry. At first I wasn't so sure I could hack it, but in recent days I've been getting really comfortable. I haven't opened Fusion in weeks for anything but to grab measurements from my old designs! Cheers.
Videos:
I threw together this robot in a couple of days to inspect my HVAC duct and pull a cable (side note: I’m aware it’s generally advised against to run a cable through a duct, but I live in LA and we don’t have heating in our HVAC system so I don’t see it as a problem)
It's a pretty bare-bones platform, designed to be as simple as possible for a quick design + build using parts I had on-hand.
The vehicle is powered by small geared 28BYJ-48 Stepper motors driven by ULN2003 drivers, controlled with an ESP32 Cam development board. It’s a decent board for the application as it has wifi, a camera and a bright forward facing LED. The board does not have the necessary 8 GPIO pins to drive the ULN chips, so I used a 8 bit shift register.
I originally intended it to be powered by an 18650 battery but I had problems using a boost converter to get the required current to drive the electronics at 5v. the board is pretty power hungry, and coupled with the steppers and high power LED, it wouldn't even boot. I resorted to using 2 lithium polymer batteries in series to achieve 7.2v with no boost converter, which worked well enough. Even then, use of the LED kept causing brown-outs and reboots so I resorted to lowering it to 1/4 duty cycle which fixed the problem.
I used Micropython and the Picoweb framework to serve a REST API for rover control and a web page to provide a UI:
The rover was designed in FreeCAD and printed on my Voron 0 using PLA. The tires were printed in NinjaFlex TPU which is a very flexible rubber like plastic, in order to provide more traction.
The rover suffered from the aforementioned brown-outs due to the LED power requirements and also struggled to turn in certain locations in the ducts due to traction limitations and as side effect of being 2WD, but ultimately succeeded at what I needed.
Awesome video, very cool to watch FDM 3D printing used in a modern manufacturing process. That said I don't think "Mass producing" is the correct term here. A boutique sign company offering custom, chiefly handmade signs (after the printing is complete) is the opposite of mass production: A lot of work is going in to each of these and they probably aren't churning them out.
FreeCAD cannot perform any operation to solidify a sketch that would result in more than one discontiguous solid.
Link can do this with a single sketch+pad (which is what I was referring to in the original post)
>
I agree! It does enforce being clean and thoughtful about your design. But the inability to use a single sketch for more than one operation still bugs me. I loved being able to plan out and see all (or at least multiple) features in a single sketch in Fusion. In FreeCAD I can only figure out how to do this by making a master sketch and then projecting single features out to multiple other sketches, which works, but like everything in FreeCAD, just takes more time...
I also used SuperSlicer until I recently tried OrcaSlicer and was very impressed. The developer of SuperSlicer recently quit his job to work on it full time though, so I imagine it will start to catch up in features soon.
Recently I've been having feelings about moving away from Fusion 360. The combination of cloud app / filesystem and their demonstrated willingness to remove features and add arbitrary limitations (eg. 10 editable model limit) makes me feel uneasy about using it. To be clear I'm grateful that AutoDesk provide a free license at all, and it's an incredible piece of software, but I have a sense of vulnerability while using and honing my skills in it. If you've ever rented a house you'll know the feeling - you quite don't feel like it's really your home, if the landlord wants to make renovate or redecorate you don't have any choice and you could be evicted at any moment.
So I tried FreeCAD. At first, I have to say that it felt a little like stepping out of a spaceship (Fusion) and banging rocks together like a caveman. It's not that you can't do (most) of the same things as an enterprise CAD package, but the killer feature of Fusion is the level of intuitiveness and "it just works" that makes FreeCAD seem like trying to write Latin.
After a week of on-and-off learning I was not sure I wanted to continue. Even after getting comfortable with the basics, frustration levels would spike to 11 sometimes. The main issue I kept running into was that altering a previous feature would break everything that came after, requiring a varying amount of work to fix. The FreeCAD wiki suggests ways to mitigate this but many of them are un-intuitive and/or inconvenient. After some googling this seems to be caused by a pretty difficult to solve issue called the "Topological Naming Problem" (where FreeCAD can't keep track of surfaces / edges / vertexes in a stable fashion when features are changed). Then I came across this blog post that pointed out a fix has actually been developed earlier this year. A developer by the name of RealThunder has created a fork of FreeCAD called "Link Branch" which can track topology in a (more) stable fashion.
I tried this branch and was blown away by how much more usable it is. Not only can it handle changes to past features almost perfectly, but I can create multiple bodies from a single sketch (not possible before) and there are other UI tweaks that make creating features easier such as the ability to preview fillets and chamfers at the same time as selecting their edges. I'm not totally sure which of these features are unique to Link branch vs which might be pre-release in the main branch, but certainly the topology naming fix is unique to Link.
So if you have tried FreeCAD in the past and been frustrated, or if Fusion's past free license changes or price increases are making you uneasy, give the Link Branch a try! Downloads are available in the releases page.
Recently I haven't been overly happy with the extrusion consistency of my Orbiter v1. After watching a particularly enlightening video by MirageC I decided to look into updating to an extruder using an IGDA (Integrated Drive Gear Assembly) which potentially reduces the problem that Mirage noticed of non-concentricity caused by gears attached to shafts using set screws. This led me to randomly browsing ali express for ideas and I came across a semi clone of the Bondtech LGX lite called the HGX-Lite. It's not an exact copy and uses a thumb screw to tension the gears instead of Bondtech's lever system. This seems like a more robust design in my opinion, although I haven't used the LGX. It seems to go by various brands. I bought a "Haldis".
I printed Mihai's extrusion test, and it shows an enormous improvement in consistency over the Orbiter. I've taken a photo with glancing lighting to exaggerate the problem - the orbiter is on the right and HGX on the left. Very happy with the result so far.
Only problem I currently see is the backlash might be worse. I'm not sure how to test this or what effect it might have on print quality - advice welcome.
Disclaimer: Not saying the orbiter is bad, I loved it and it's quite possible the V2 is better re: consistency than my original version, although it still does use a set screw in the design.
Here's a picture of the extruder (ignore the mess of wires - testing before committing to a rewire)
For anyone running into this, I figured out what the problem was. The android filesystem on the device seems blind to changes that have been made outside of android apps, so any changes to file structure made with the Chrome OS file app does not appear in android apps until device restart. To fix this, I started using an Android file app to copy the files.
Semi-related rant: if you are considering using Chrome OS to do "productivity" tasks, I would reconsider it. Android app integration is full of bugs and it's a nightmare to get anything done beyond browsing the web or consuming content. Additionally, android apps are evicted from memory all the time even with 8GB ram (which I paid more for specifically to avoid this problem), - anything you are working on is frequently lost when multi tasking (alt-tabbing). I should have bought a windows laptop instead.
view more: next ›
Yeah - I tried that technique to change the inherited profile setting on a new profile to an existing master printer profile that I have, but when you do that, the new profile just disappears in the UI. If you reset it back to a system profile, it reappears. I can't understand the utility of that.