Isn’t that a polar opposite to an open source project though …?? We can all see different issues as we go through the build / config process but we are sharing those issues - we are trying to make it better for everyone by collaborating, rather than taking something someone has put thousands of hours into designing and going “but mine is better as I’ve tweaked xyz…”
This isn’t aimed at the AliExpress $100 buyer, and if I wanted commercial $10k+ CNC then there are plenty of decent options out there - if you think there is a market for your prebuilt large scale CNC then crack on, but please in the spirit of open source share what you think could be done as then everyone benefits.
1000% agree - if you are benefiting from all the hard work that Bar and team have put in to this point, as well as all the feedback coming from the community, karma dictates that you should share your feedback and suggested improvements. That doesn’t prevent you from trying to manufacture and sell prebuilt machines with those improvements. And maybe the community will find other issues with your improvements and give you feedback on those too.
I’m sure there would be a market for a pre-built option. I think the success of the two Kickstarter campaigns shows that there’s also a market for DIY kits.
@bar, sorry for sounding like a broken record. But I feel like history is setup to repeat itself… Consider submitting that patent, and non commercial license in short term…
Consider limited sharing design/specs of just the problematic parts, to open up opportunity for community to help mod and help scale improvements/progress. For fun, or contest or bounty?
Sorry for being an overly opinionated basement dwelling armchair community member.
That being said I don’t know that the calibration process will ever truly be safe to let run unattended. At that point the machine doesn’t know the locations of it’s anchor points so it can’t keep the proper tension in the belts which will always lead to some chance of them getting sucked into the gears. We can keep working to reduce the chances, but I think that I should also emphasize in the user guide that calibration isn’t really meant to be run unsupervised.
If anyone else wants to tackle it I’m happy to share.
Re: people making non-open source versions of Maslow4 - I am 100% in favor of that. I think that if we are serous about making large format CNC routing accessible to more people it’s going to be critical to have a bunch of different companies competing. I think that our role as an open source project is to keep innovating and pushing the technology forward. We saw in the past when we stopped doing that innovation stopped happening and prices went up. We’re here to drive innovation ever forward and prices ever down.
that view doesn’t show the belt path, but does look like it would do a much
better job of keeping the belt out of the gears.
I think that if we do a minimal calibration at the center, and then a manual
drive to the 4 corners, it will be safe to leave the machine unattended to fill
in the area in between.
it may even be possible to start with manually driving to the 4 corners, but I
think there will be fewer problems with slack if we start with a minimal
calibration in the center first.
This seems to work pretty well. I’m going to make four and do more testing.
I don’t think so. I’m open to being proved wrong, but I think that until we have good calibration and can rely on all four belts being under the correct tension we have to leave two belts slack while moving which will always lead to a chance of them getting tangled.
I think that the real issue is that the calibration process isn’t reliable enough and people are running it multiple times. If it’s a one time thing it’s not such a big ask to keep an eye on it during the calibration process, it’s only if you are doing it over and over that it gets really tedious to have to tend to it.
This seems to work pretty well. I’m going to make four and do more testing.
looking good. Question, since these are fitting so tightly, can this eliminate
the need have screws to hold it in place?
I don’t think so. I’m open to being proved wrong, but I think that until we
have good calibration and can rely on all four belts being under the correct
tension we have to leave two belts slack while moving which will always lead
to a chance of them getting tangled.
there’s “a little slack” vs “enough slack that they can tangle”
I don’t know how well it could work, but I think you should know if a belt
should be going out or not from even a minor calibration (you should know what
quadrent you are working in) and when you have a belt that’s going out, you have
the movement of the belt from the encoder, and how much power you are applying
to feed it out (and how much power you are applying on the other side to pull
them in). unless the spools just have too much friction, it seems like it
should be possible to try feeding just a little more or less power on feeding
out and measure the effect on the power/speed that you are feeding in.
If feeding out faster lets you feed in faster with the same power, then you have
too much tension and feeding out faster is good.
if feeding out faster makes no diffrence in feeding in with the same power, then
the feed out belts are slack and you should feed them out slower.
not a trivial PID loop, but seems like it should be doable.
I think that the real issue is that the calibration process isn’t reliable
enough and people are running it multiple times. If it’s a one time thing it’s
not such a big ask to keep an eye on it during the calibration process, it’s
only if you are doing it over and over that it gets really tedious to have to
tend to it.
I thought the design was that calibration would be done each time you hook up,
as you can’t be sure that you are on the same frame and that the frame hasn’t
shifted??
@bar, once you’ve tested enough that you think they are a good addition, let us know. I’d probably order a set for my machine from CraftCloud to minimize the risk of mangling another belt.
Good luck, seems you missed the point of an open source project where people contribute to make it better. I’d be sure never to buy your products and tell everyone else the same. That greed always getting in the way eh?
I saw you added the 200ms of feed/retract to the latest firmware. Is there going to be an option to “Retract All” but instead “Retract 1of4, 2of4…”? I am hesitant to put my new spool on if there isn’t a way for me to more closely monitor what it is doing 1 at a time when setting up.
I am just scrubbing around the firmware and I see retractAll is already broken out into retractTL, ect. Can we have webUI access to retractTL, TR, LL, LR directly?
Crap I don’t have the UI available as my machine is apart and index.html preview just gives me the “Connecting” popup. Somewhere in the config popup or maybe in the machine control area…
I also just saw the new belt guard which I will definitely try. Is this the kind of thing that would be in the not-store or is there a different location for cad drawings?
I can put it in the not-store once we’ve tested it a bit more, but right now the only place to get it is in this thread
It’s still not foolproof though, I did (with some concerted effort today) manage to get it to eat a belt while using. I was like literally pushing it into the gears though.
