This is the video I have been looking forward to seeing. Speed is not an important factor for me, as it will always be faster and more accurate than my hand placement will ever be. More importantly it also allows me the time to do other tasks while the PNP is working. Thanks again for sharing your journey with us.
Agree, speed is not the highest priority requirement for personal desktop use. Yet it is faster than hand placement, in the video the placement speed is around 220 cph
Good job! I think the glitchy button presses are happening when the pump is running, so either EMI from the unshielded wires or maybe the pump is pushing electrical noise into the power supply. Probably nothing to worry about, but it could result in random glitches later.
I'd be interested to see speed tests. Maybe larger components can go fast without placement problems, but slow for 0402 and fine-pitch chips that need maximum accuracy?
Will get back to that, currently I have checked several speed faster than what I have in video, which looks good, in general, if we tuned the acceleration and jerk well, even if it is higher speed, it should be good enough, it also depend on the machine as well. The other thing is when moving fast, some machine come with higher vibration, it will cause the component to change position during movement, that will also be one limiting factor, I observed this in Ender CR10.
Is there any benefit in having printed barcodes/QR code labels on the feeders, so the machine could scan the build plate with the camera and identify which components were mounted at each location? Or is it quick to just enter that information manually?
Yes, definitively, I haven't got time to get to this point, setting up the feeder is a tedious job but with a lot of effort, so anyway to reduce the effort will make it valuable optimization.
The button press issue seems to correlate with the air pump being on, maybe put a ferrite core around its power line or shield/bypass some of your electronics?
Thanks, I actually thought about it before, but I find it easy to use stencil, therefore didn't go that direction, it is definitively possible, can use the rotation stepper for injection control. OpenPnP has that function before, but not with 2.0 version, so some effort needed as well for software.
Main challenge is when component is attached to the nozzle, the acceleration cannot be too high, or the component might move. This can be improved by setting carefully acceleration, jerk,.etc, and also by adjusting suction pression. The speed in the video is set to be slow, I think it can be much faster, but not as fast as 3d printer.
For trimmer cassette feeder, you can rotate the trimmer, then the tape will move forward. For the normal feeder, the tape need to be moved manually forward. For the camera spec, you can have a look here: github.com/xpDIY/3DPlacer/blob/main/designs/BOM/BOM.csv
I would recommend using resistor as some printer come with mosfet supporting current lower than 1A, so when component is attached, the current will go higher than mosfet can stand, so it is good to have a protection . You can refer to my earlier videos that I fixed the mosfet issue.
In relation to placement of the components, my opinion is that it looks really good, almost like its been done by and Ikea Engineer, as compared to my hand placement which looks like its been done by a Koala bear on crack cocaine.
Not really necessary for fiducial, you can select multiple components in OpenPnP and calibrate by manually center to each component. Fiducial is a easy way to automate the board location calibration.
This is the video I have been looking forward to seeing. Speed is not an important factor for me, as it will always be faster and more accurate than my hand placement will ever be. More importantly it also allows me the time to do other tasks while the PNP is working. Thanks again for sharing your journey with us.
Agree, speed is not the highest priority requirement for personal desktop use. Yet it is faster than hand placement, in the video the placement speed is around 220 cph
This is awesome. Thanks so much for sharing!
very cool to see 0201 placements
this is awesome! i appreciate you documenting your hard work!
Thanks, more will come, stay tuned!
This is great, keep up the great work.
Thanks, more improvement coming up!
Good job! I think the glitchy button presses are happening when the pump is running, so either EMI from the unshielded wires or maybe the pump is pushing electrical noise into the power supply. Probably nothing to worry about, but it could result in random glitches later.
Right, it is not happening now, but I will check when it is coming back and observe the signal. Maybe some shield will help.
I'd be interested to see speed tests. Maybe larger components can go fast without placement problems, but slow for 0402 and fine-pitch chips that need maximum accuracy?
Will get back to that, currently I have checked several speed faster than what I have in video, which looks good, in general, if we tuned the acceleration and jerk well, even if it is higher speed, it should be good enough, it also depend on the machine as well. The other thing is when moving fast, some machine come with higher vibration, it will cause the component to change position during movement, that will also be one limiting factor, I observed this in Ender CR10.
Is there any benefit in having printed barcodes/QR code labels on the feeders, so the machine could scan the build plate with the camera and identify which components were mounted at each location? Or is it quick to just enter that information manually?
Yes, definitively, I haven't got time to get to this point, setting up the feeder is a tedious job but with a lot of effort, so anyway to reduce the effort will make it valuable optimization.
The button press issue seems to correlate with the air pump being on, maybe put a ferrite core around its power line or shield/bypass some of your electronics?
Indeed it is related to air pump, and thanks for the hint, I will need to look at it a bit closer..
Good work, Good Video. Have you thought about making a paste injector. I single prototypes and I don't want to mess around with screens.
Thanks, I actually thought about it before, but I find it easy to use stencil, therefore didn't go that direction, it is definitively possible, can use the rotation stepper for injection control. OpenPnP has that function before, but not with 2.0 version, so some effort needed as well for software.
What’s the main speed limitation? I bet when it works as 3D printer is moves much faster
Main challenge is when component is attached to the nozzle, the acceleration cannot be too high, or the component might move. This can be improved by setting carefully acceleration, jerk,.etc, and also by adjusting suction pression. The speed in the video is set to be slow, I think it can be much faster, but not as fast as 3d printer.
Thanks for sharing, how the feeders feed components when the top row is empty? Also whats the camera spec and model?
For trimmer cassette feeder, you can rotate the trimmer, then the tape will move forward. For the normal feeder, the tape need to be moved manually forward. For the camera spec, you can have a look here: github.com/xpDIY/3DPlacer/blob/main/designs/BOM/BOM.csv
If I use a 12v pump do I need the Keramik Resistor ? If the output voltage is 12V of my printer ?
I would recommend using resistor as some printer come with mosfet supporting current lower than 1A, so when component is attached, the current will go higher than mosfet can stand, so it is good to have a protection . You can refer to my earlier videos that I fixed the mosfet issue.
In relation to placement of the components, my opinion is that it looks really good, almost like its been done by and Ikea Engineer, as compared to my hand placement which looks like its been done by a Koala bear on crack cocaine.
Hand placement is quite difficult, especially for 0201, a small shaking of the hand will make the component dancing with other component near by, lol!
do you have to have the calibration points on the pcB or is another calibration already possible or in planning
Not really necessary for fiducial, you can select multiple components in OpenPnP and calibrate by manually center to each component. Fiducial is a easy way to automate the board location calibration.
thank you very much for your answer will you create a complete guide including setup with openPNP?@@xpdiy
@@CranckRa1nyes, that is in my plan actually, stay tuned!
@@xpdiy perfekt just ordered a replacement for my endet 3 v2 can’t wait for the videos