So then you’d be asking the user to configure a static network interface and look at the interface to get the UP address of the interface and type that into LightBurn. Is /dev/ttyACM0 as shown in dmesg output so hard?
I’ll give it a shot one of these days but it would seem using USB would be the easiest.
USB is slow, is the reason I would go ethernet.
Configuring an address isn’t hard.
In my case the Cohesion3D rendition doesn’t have ethernet port. I can get the USB port to work, but only if I change the permissions of the device at each startup.
The thing that’s odd is that I’ve supposedly added the atomicpi to dialout group (the owner of the /dev/ttyACM0 port) AND if I ssh into the ATpi and do groups, it shows all the groups I’ve added, including dialout. However, if I open up a terminal from the lubuntu deaktop and do groups, it only responds with atomicpi. AND if I do sudo adduser atomicpi dialout, it says that atomicpi is already a member of that group! Very peculiar.
sudo chmod 666 /dev/ttyACM0 works if I do it after turning on the laser and before starting LightBurn.
I must be mixing up discussions, I didn’t think this was about a slow interface. If you want a fast interface I’m working on a project which puts MBed on the laser cutter controller so that LinuxCNC on an rPi4 can control the machine in realtime over a SPI interface. THAT would be fast.
Interesting, but what cable length can you get on a SPI interface? (unless the RPi is mounted IN the laser)
BTW guys, as a point of order, since I ‘own’ this thread, I am perfectly happy to let it wander in these very interesting pathways. I just wanted to let you know, it’s ok by me 
I smell a SPI spy! Yes, only 100mm for now. Going to research how far I can get with grounds between signals wires and then with pullups on the ends. Later though and might even look at using ethernet transport but too few controllers have ethernet ports but the current prototypes are working using MKS Sbase v1.3 boards and they do have ethernet. Going to worry about those details after we get some software images up and ready for people to try.
I’m inclined to think that USB3, and I really hate that I’m saying this, may be a better choice for a high speed, point to point connection.
I hate it because, from day one, I was always opposed to USB. Didn’t like it, didn’t like how it was being thrust down the industry’s throats, and didn’t like that it was actually delivered 2-3 years after all the major PC mfrs included it on their platforms (and were sued for false promises). However, I’ve been told, forgiveness is a virtue 
In this case, if you have any length at all on an interface that wants to be fast, differential, self-clocking is the only way to go… Lots of technology development in this. Just sayin…
I like SPI, for local connections. Easy and reliable. Certainly not fast by USB3 standards, but sufficient for most IO type devices.
Sweet enclosure! Most I’m looking at aren’t for the larger board.
Is it your design? Did you publish it somewhere?
John
only problem is we’re trying to work with what’s on the market for existing 3DP controller boards and coms is pretty poor. Ethernet is an option on some, almost all have USB but it’s not USB3. SPI with 100mm of ribbon cable seems to work and might work better with some grounding/shielding between sig lines. Most all boards have SPI. Just trying to work with what’s there and show that LinuxCNC can be a player at the low and middle end too.
I can post the SVGs (will do later today), but it’s easy to make the general box pattern if you go to:
https://www.festi.info/boxes.py
This enclosure was under Boxes->ElectronicsBox Enter the dimensions you want and it will spit out an svg with all the pieces.
However, because every laser cuts a slightly different kerf, you should go to Parts and Samples->Burn Test to see what finger adjustment value to use. Then make the basic enclosure and you can fit the ‘insides’ (all the openings I have) into that.
Update: @Driversaurus Tried to upload SVG files in a ZIP file and was told, no can do… Sorry. Not sure how to upload these on this forum.
Thanks, this helps clarify the scope of what you’re doing.
With SPI being completely synchronous, you can tweak the clock down to compensate for poor link quality. And, since all of the signals are unidirectional, you can use differential driver/receiver pairs to get very good signal quality over long distances as long as you’re willing to slow down the clock to compensate for the round trip delay (only critical if reading a slave).
A pair of LTC4332s will give you 10M at 2MHz over CAT5.
Yep. I’ve used that to do interesting things like my exhaust filter setup, which apparently I re-invented based on YouTube videos I found later. 
I also made some interesting box-based Christmas gifts this year that I’m thinking about putting out on Etsy just to get things rolling out there.
John
in connection with your universal plausibility, all systems mac, pc wether running windows, ios, linux have the same vulnerability in trying to reset after such a power supply issue is that the capacitors hold enough of a charge in them to hold the initial problem. Thats why they advise not to open a PS, a jolt from the capacitors hurts like a b%^*&% lol. cycling the power switch drains the capacitors allowing for a reset. May never happen again.
@TomWS So I built up a simple inline header, soldered all pins together and used on a 10 wire ribbon cable with a short jumper from this inline-header to ground. I did this on each end of a 200mm ribbon and it worked fine. The developer of this spiPRU/Remora did the same to a 600mm ribbon and also said it worked fine. He had originally had 100mm of ribbon, no transmission line considerations and he had to declock the SPI bus. So SPI is working great so far.
So, does this mean that every other wire in the ribbon is ground with signals in between?
If you’re only doing Master Writes you can probably go a pretty long distance. The problem is reading data from the slave. In this case propagation delays will shift the incoming data with respect to the clock at the Master.
However, the fact that its doing what you want is the important thing! 
As an aside, I’ve shelved my Atomic Pi project. The Mini PC is working, now that I’ve disabled the auto power on, and I had too many things not quite right on the ATPi. I’ll PM the list to you if you’re interested, it’s nothing major, just a bunch of ‘oh, didn’t see that coming’… kind of things.
Yes it’s getting data back from destination device( hotend/bed temps, endstops, etc ) but as you said, it’s working so far so good.
I’d be interested in seeing your list of issues you’ve run into. I’d not integrated the ATPi into my workflow as I have a laptop already at my machine but I’m soon to be replacing all control interfaces wtih LinuxCNC so seeing what you were attempting might give me hints on ways to improve my setup.
Well, with the fantastic help from @DougL I have been able to replace the MiniPC with an AtomicPi and will, therefore, declare this topic complete and ‘solved’.
You might want to check out the thread:
It describes the AtomicPi and it wouldn’t surprise me if Doug posts his recipe for setting it up to run LightBurn there.
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.