Its only a matter of time surely, its such fantastic content as soon as the algorithm associates it with something like DIYPerks itll take off like mad?
You go through so much effort to make sure everything is as easy as possible to integrate into each other. Thank you for all the RnD you do. I'm sure you don't get enough credit from UA-cam or its algorithm. A lot of the stuff you've done is very inspiring to me, hopefully more people see your stuff soon.
If you are using the embedded STM32 USB HID (USBDP + USBDM pins), you have to programm your own driver... In this case it`s easier to use an FTDI chip and connect it via UART
Another way to fix the panel rigidity would be to add 2 extrusions in direct contact with the pannel going through lenghtwise and attach them to the existing frame, more expensive than the 3D printed foot though
@@stephen_hawes just moving it away may not be enough since the frame will still vibrate, I recommend trying to come up with a mounting system which will minimize vibrations as well like foam/rubber padding (assuming you still want to mount everything to the frame). Also other ppl have suggested migrating from Marlin to Klipper.
Another possibility for making a fiberboard or acrylic plate more rigid would be to add two lengths of extrusion on the long dimension for it to rest on, or even in the slots.
Stephen, consider running a pair of extruded aluminum profiles under the fiberboard. This would add extra cross support and hold up the fiber board. I would suggest doing this for the steel plate too, it won't do a ton for vertical support but it still has value as a cross support.
Wow that laser cut steel was super expensive for what it is! I had a local place cut mine - more like 2x3 feet with similar hole pattern and some corner jigs for $80.
Next stop: interchangeable attachments: A drill; vacuum pump; solder dispenser; track printer etc. A locking mechanism holds the tool in place and pins provide both identity to the controller and power to the attachment. That way you could load a blank PCB, the software drills the holes, prints the tracks and artwork, dispenses the solder then places the components ready to be re-flowed or with a heat gun attachment, soldered in situ to the board.
I was wondering how well the push fit of the nozzles would work, but that looks pretty slick! Doesn't seem to bind up at all. I suppose if you do start to have troubles, you could make them attach with ring magnets? Kinda like those phone charger cables. That'll give you a highly repeatable and accurate mating, without worrying about parts wearing
Hum for the STM, Prusa just moved to STM for prusa mini if I remeber well. IDK if their firmware for prusa mini is open source but maybe you can search this way ?
Not entirely trivial but github.com/hathach/tinyusb . Setup a USB CDC device and setup a service call in the main loop. Haven't looked, but suspect someone already has already done this as serial over actual USB is definitely becoming a more common thing in embedded space.
Hey you can try somthing, is not verry efficient but you can have both microcontrollers and ARM to control every thing and a AVR to use marlin, and you can comunicate the chips with SPI
How to make it not wobble? Board, meet beam. Stiffness is overwhelmingly defined by the third power of beam width. So add a rib to the board and your problem is solved. PS: This is an awesome project, thanks for the video! Also, is someone working on adding a pick'n'place tool to the E3D motion platform toolchanger?
Stephen: I am gonna redesign the mount for the camera and the ringlight so that it actually has a foot Me: Oh cool, so it'll have a foot............. like a snail
my pump motor really work like vibrator and and I am not sure if its a good idea to put it near camera . also marlin easily has pre configuration for some of STM32's and also LPC series and you can easily add more chip's to it
STM32F407 Issues Are you attempting to flash a simplified minimum development board first? This is what I would do. Remove all of the unknown factors of a design first by using a proven toolchain. I am only getting started with the STM32F411 and micropython on a containerized linux machine. I'm not much, if any, help. Micropython migrated its way onto my board pretty easily. I only had issues due to permissions in linux. I had to use the dfu-util method instead of the default utility shipped with micropython's github repo. The process I used is described in ../micropython/ports/stm32/README.md. Obviously, you're not trying to run micropython here, but many times in the past I have figured out toolchain issues after looking at how less esoteric projects work with similar hardware. Micropython, supports the F407. GL -Jake
Peek array: Bunnie's tribute to Nadya Peek, a badass who researches "Machines that make machines": theamphour.com/208-an-interview-with-nadya-peek-gallant-gcode-gerontology/
Admittedly less about tool changing then everything else in the video, a little bit of click bait I think, having said that I do enjoy watching your videos
I don't quite understand why you spend so much time trying to compile marlin for this machine. Use klipper and a raspberry pi, should work out of the box, and be significantly more flexible AND easy to configure. Connect the pi via network and make the port available with ser2net and you're all set.
I would imagine he either has experience with marlin, or it was the first thing that popped up when he tried to find a solution. My bet is he hasn't heard of your solution yet.
he's trying to run a custom board that replaces the pi and all the different interfaces necessary to drive the machine. while it could be done on the pi, the custom board will make the final product more reliable and simpler. also, since this isn't a 3d printer marlin/klipper/etc , as he mentions, needs to be modified.
@@theJonnymac I understand his intentions. I would still recommend klipper over marlin any time. I recommended klipper because I have successfully used it with openpnp and it has an official STM32F4 port and thus would not require any modification at all, it can also be very easily configured for pnp. Yes the pi is an annoyance in this setup, but at least it would not take him weeks to compile. And it's not like there is the one perfect firmware for openpnp out there, but there is more choices than marlin.
@@wolfganglienbacher you bring up an interesting point, just in general. he is trying to build 2 major components here, one a device that auto builds circuit boards, and a complex controller for it. one way to develop it would be to build an interface board that handles all the device controls and interfaces to something like a pi. and once the physical aspects are done and you know what all the components are needed. you would then build your fancy custom board that will take several iterations and lots of software time, but at a risk of lots of rework. Another, is you design and build both at the same time, which might require more work at different times and progress seems slower, but you can minimize rework/complete redesign at the end. I've never built a device this complex myself, I've helped build/repair complex industrial machines using both methods. It wasn't a clear win using either method.
@@xxportalxx. but marlin actually isn't recommended for OpenPNP, so I dont really understand why he insists on using marlin instead of switching to the recommended software.
I love your channel man, I hope the algorithm picks up on it soon
Its only a matter of time surely, its such fantastic content as soon as the algorithm associates it with something like DIYPerks itll take off like mad?
It did for me hi
You go through so much effort to make sure everything is as easy as possible to integrate into each other. Thank you for all the RnD you do. I'm sure you don't get enough credit from UA-cam or its algorithm. A lot of the stuff you've done is very inspiring to me, hopefully more people see your stuff soon.
If you are using the embedded STM32 USB HID (USBDP + USBDM pins), you have to programm your own driver... In this case it`s easier to use an FTDI chip and connect it via UART
Another way to fix the panel rigidity would be to add 2 extrusions in direct contact with the pannel going through lenghtwise and attach them to the existing frame, more expensive than the 3D printed foot though
Indeed it is unfortunate that T or U beams made from tree carcases wasn't an option :)
Just use your same idea but fix it with right angle v-rail adaptors (used in the old machine) and it's done
You put the pump, which vibrates, next to the camera, which needs to be stable.
OH. YO. that's a super good point. thanks for the note! i should move it super far away
@@stephen_hawes just moving it away may not be enough since the frame will still vibrate, I recommend trying to come up with a mounting system which will minimize vibrations as well like foam/rubber padding (assuming you still want to mount everything to the frame). Also other ppl have suggested migrating from Marlin to Klipper.
@@stephen_hawes Nah, you will need to decouple it from the frame.
@@stephen_hawes Look for some camera isolators used in drone industry and mount the PUMP with them. ;-)
Or let it keep swinging with a spring based holder 👍
Another possibility for making a fiberboard or acrylic plate more rigid would be to add two lengths of extrusion on the long dimension for it to rest on, or even in the slots.
Stephen, consider running a pair of extruded aluminum profiles under the fiberboard. This would add extra cross support and hold up the fiber board. I would suggest doing this for the steel plate too, it won't do a ton for vertical support but it still has value as a cross support.
Here I was just chilling and being bored out of my mind. Boom! New video! Thanks!
Nice idea is a very good upgrade for the Index
Wow that laser cut steel was super expensive for what it is! I had a local place cut mine - more like 2x3 feet with similar hole pattern and some corner jigs for $80.
Next stop: interchangeable attachments: A drill; vacuum pump; solder dispenser; track printer etc. A locking mechanism holds the tool in place and pins provide both identity to the controller and power to the attachment. That way you could load a blank PCB, the software drills the holes, prints the tracks and artwork, dispenses the solder then places the components ready to be re-flowed or with a heat gun attachment, soldered in situ to the board.
This is slowly getting too awesome to ignore any longer
You damn freaking genius :D
I saw the preview on IG and I'm happy it's on YT this evening (EU time).
Great jog as usual, always entertaining
I love your energy and enthusiasm!!
I was wondering how well the push fit of the nozzles would work, but that looks pretty slick! Doesn't seem to bind up at all.
I suppose if you do start to have troubles, you could make them attach with ring magnets? Kinda like those phone charger cables. That'll give you a highly repeatable and accurate mating, without worrying about parts wearing
I’m not sure but I think these might be Juki. Web search juki pnp nozzles and atc
Hum for the STM, Prusa just moved to STM for prusa mini if I remeber well. IDK if their firmware for prusa mini is open source but maybe you can search this way ?
You really deserve at least 1m subs
Needs to move his face 1m further away from the camera.
Not entirely trivial but github.com/hathach/tinyusb . Setup a USB CDC device and setup a service call in the main loop. Haven't looked, but suspect someone already has already done this as serial over actual USB is definitely becoming a more common thing in embedded space.
2:28 "This thing does not bend"
*literally shows how much it bends* lmao
The plate should have a U or T profile over the length on the long length sides. This is still bending a lot.
Dude where do you get all this amazing music for your videos what the hell.
Where did you get the nozzles and the mount from?
Question: Do you have a st-link to flash the STM?
Your prints are always like glass, what are the settings you use for your 3d printer?
I like the music (and the whole video), but cannot find it!
J'ai rien compris mais sa a l'air super
Hey you can try somthing, is not verry efficient but you can have both microcontrollers and ARM to control every thing and a AVR to use marlin, and you can comunicate the chips with SPI
I would check out local sheet metalworking shops for lower prices on laser cut metal
How to make it not wobble? Board, meet beam.
Stiffness is overwhelmingly defined by the third power of beam width. So add a rib to the board and your problem is solved.
PS: This is an awesome project, thanks for the video!
Also, is someone working on adding a pick'n'place tool to the E3D motion platform toolchanger?
Neato, entertaining as always! Gotta wonder if you'll ever actually be "done" with this project :D
Stephen: I am gonna redesign the mount for the camera and the ringlight so that it actually has a foot
Me: Oh cool, so it'll have a foot............. like a snail
You could use garolite vs laser cut steel, wouldn't be a lot cheaper however.
my pump motor really work like vibrator and and I am not sure if its a good idea to put it near camera . also marlin easily has pre configuration for some of STM32's and also LPC series and you can easily add more chip's to it
bends... why didnt you make it like an actual sheetmetal part like a u section
Could you describe exactly what you're trying to do with the stm32f4xx? Sorry, I think it was not clear enough... Let me know, maybe I could help you
Try replacing the micro, ESD damage is insidious and can manifest through many weird errors.
Loving this project, Just wish it was a video per day 🤣
STM32F407 Issues
Are you attempting to flash a simplified minimum development board first?
This is what I would do. Remove all of the unknown factors of a design first by using a proven toolchain.
I am only getting started with the STM32F411 and micropython on a containerized linux machine. I'm not much, if any, help. Micropython migrated its way onto my board pretty easily. I only had issues due to permissions in linux. I had to use the dfu-util method instead of the default utility shipped with micropython's github repo. The process I used is described in ../micropython/ports/stm32/README.md.
Obviously, you're not trying to run micropython here, but many times in the past I have figured out toolchain issues after looking at how less esoteric projects work with similar hardware. Micropython, supports the F407.
GL
-Jake
Can you post your used pin_map.c? Then I can compile it for you
Nice om tattoo 🤩 on left wrist 🙃
Peek array: Bunnie's tribute to Nadya Peek, a badass who researches "Machines that make machines": theamphour.com/208-an-interview-with-nadya-peek-gallant-gcode-gerontology/
What about watch
He’s so cute ☺️
Admittedly less about tool changing then everything else in the video, a little bit of click bait I think, having said that I do enjoy watching your videos
You are such a cutie..
I don't quite understand why you spend so much time trying to compile marlin for this machine. Use klipper and a raspberry pi, should work out of the box, and be significantly more flexible AND easy to configure. Connect the pi via network and make the port available with ser2net and you're all set.
I would imagine he either has experience with marlin, or it was the first thing that popped up when he tried to find a solution. My bet is he hasn't heard of your solution yet.
he's trying to run a custom board that replaces the pi and all the different interfaces necessary to drive the machine. while it could be done on the pi, the custom board will make the final product more reliable and simpler. also, since this isn't a 3d printer marlin/klipper/etc , as he mentions, needs to be modified.
@@theJonnymac I understand his intentions. I would still recommend klipper over marlin any time. I recommended klipper because I have successfully used it with openpnp and it has an official STM32F4 port and thus would not require any modification at all, it can also be very easily configured for pnp. Yes the pi is an annoyance in this setup, but at least it would not take him weeks to compile. And it's not like there is the one perfect firmware for openpnp out there, but there is more choices than marlin.
@@wolfganglienbacher you bring up an interesting point, just in general. he is trying to build 2 major components here, one a device that auto builds circuit boards, and a complex controller for it. one way to develop it would be to build an interface board that handles all the device controls and interfaces to something like a pi. and once the physical aspects are done and you know what all the components are needed. you would then build your fancy custom board that will take several iterations and lots of software time, but at a risk of lots of rework. Another, is you design and build both at the same time, which might require more work at different times and progress seems slower, but you can minimize rework/complete redesign at the end. I've never built a device this complex myself, I've helped build/repair complex industrial machines using both methods. It wasn't a clear win using either method.
@@xxportalxx. but marlin actually isn't recommended for OpenPNP, so I dont really understand why he insists on using marlin instead of switching to the recommended software.