That’s a solid idea, and we’ve talked about doing it. It’s a great trick to keep in the back pocket if it ever proves necessary! Thanks for the suggestion.
@@lucianchapar just mounting parts on it can flex it at bit. These bends might help or perhaps just an alu beam beneath would counteract it and make it more stiff. 🤓
We could! It's a good idea. Part of the reasoning was to have a nice surface to mount a board or other things to magnetically, plus having a plate gives us space to easily mount the up camera. If you build one with extrusion instead I'd love to see!
The magnetic surface and the regular pattern of holes provides a great utility for dealing with wire management and having a place to mount the electronics in a predictable spot for kit assembly. As a part for your business to manage the outsourcing of, and as you show quality check, it seems as if this part has created too much complexity and cost for the utility it provides as a solution. Your suggestion of a boot is a better, simpler and cheaper method to get the same repeatable flat surface to support an MDF, plastic, melamine, plywood or aluminum plate with similar results. A couple of extruded aluminum rails could also span between the left and right sides with corner brackets. Some people could easily source a piece of material cut to the right size then cut the holes and a correctly sized hunk of wood as the boot to prop up the center span. @Stephen Hawes - you have already done an amazing job with this project and we see how excited you were to be able to outsource successfully from an external vendor, but it seems as though the plate part doesn't make much business sense for a DIY project except for those who would rather pay and extra $80 instead of cutting a piece of plastic or wood from a local big box store. I like the strategy of getting product in the customer's hands as fast as possible, but it seems to me that finding a reliably sources and fairly priced solder paste nozzle/syringe head kit and nozzles for the pick place head are the parts I am having the most difficulty with sourcing. Of course, the motherboard PCB, cameras and feeders (in that order) are the parts that allow someone to get most of the platform started.
You changed the manufacturer's part when you mounted your boards and other parts to it and then mounted it into the frame. Those will apply some unknown stresses and torques to the plate. So, what you're doing is in-situ testing, not acceptance testing. But, in the end that's what you want to test in your integrated environment, if the focus of the camera is sensitive at those error levels. Also as noted below, powder coating is not a precision consistent thickness operation, consider anodizing. I'm enjoying the series and might end up building one for my own PCB projects. Keep up the great work and I love your enthusiasm!
Four screws? Wouldn't you be better of mounting it on kinematic 3-balls if you want to measure the inherent flatness of the plate? Now you have copied any torsional inaccuracies of the aluminium profile into the plate, albeit reversibly (elastically). 6:05
Hey! Just FYI that dial indicators like the one you used usually have a serial port where you can read the measurement. I dont know if your particular one has one, buuuuut usually on amazon there are ones that do have a serial port and they are really cheap. Integrating the indicator in the system would be SO COOL! As for everything else.. keep it up!
How does it compare with a much cheaper option? Like 6mm acrylic with a support foot in the middle? Easier to source (eg, lasercut at most makerspaces) and lightweight means reduced shipping costs too
Could the bend in the plate be influenced by gravity? Usually, strings that are laid across a distance will bend according to a hyperbolic cosine once they come to rest. Why not flip the plate and run the same test on the other side- if gravity is the factor that is responsible for the bend, you should see a nearly identical curve on the other side. Measuring discrepancies between the curves you get for each side would be an interesting way to evaluate the net curvature.
Flip the board downside up and measure again. Also the spring of your measurement device adds to the bow deflection. I could see it in the film, specially in the middle. And .8 of a mm is about 3 times the thikness of a resistor. Rather big
He could flip the plate upside down and redo the measurements. Subtract one set's xy location from the other set's. Zero means the gantry was bowed. Big difference means plate was bowed. Could figure a formula that gives error of each.
@@Ziraya0 yeap I think the mistake, if you can call it that, is to talk about the plate flatness, instead of the ability for the end of the nozel to track the surface of the PCB, or for the webcam to track a part held over it etc.
@@TheRainHarvester True. I also suspect that he would see an apparent bend in the other direction once he flips the plate, just due to gravitational pull.
Just a heads up your 3D printer style rail might actually be a slightly bowed. I've had this issue with sourcing linear rails also. The only way to verify your measurements is with an precision flat plate. Then compare those results to your part.
I vote for showing off more of the index validation testing !! What are you trying to measure, and How you are measuring it ? Also what is the pass criteria ?
@Stephen I'm eagerly waiting for main Controller board and Feeders, can you please release beta versions? that's how you will get good feedback regarding issues and improvements.
+1 to FreeCad, would love to see a tutorial on how your transition away from Fusion 360 went, because I found FreeCad pretty difficult to go to after getting used to Fusion 360. It still takes me about 2x the time to make the part on FreeCAD :(, but I really want to support the project and make it better, so I'm pusshing through.
Maybe check out the OpenSCAD workbench for it. I *LOVE* how easy it is to create foundational geometries (or even the whole model) parametrically with just a simple script.
If the flatness of the steel plate is being driven by the camera's ability to focus, why not increase the the depth of field of the camera. You have plenty of light, so you could go to a smaller aperture thereby increasing your depth of field. See the online Depth of Field calculator for more information. Money spent on decreasing the camera's aperture with a special lens might be less than on the steel part (and you could use right angle stock to support the long stretch and stabilize it enough). Food for thought. You could also program for the components height and adjust your Z to the part to bring its bottom face within your field of focus.
for negating the bow of the plate, why don't you use 1 or 2 corner "L" profiles(like the ones used for the rails) on both sides of the plate, to add some rigidity?
Really cool ! There is something I don't understand, though : why not use a less expensive but more flexible plate (wood ?) supported by aluminium extrusions ? would'nt that be less expensibe and as good in terms of rigidity and precision ?
Wait, but if it's just a powder-coated laser cut piece of cold rolled steel, then the "flatness" can vary from spot to spot across the whole sheet, which the parts were cut out of. Isn't it OK to rely just on flatness of a single sheet of metal without any strengthening ribs? I think you have to put at least some 20x20 or even 20x40 extrusion across the whole sheet, to keep it from flexing and reduce any potential curvature. Edit: watched the remaining of the video, extrusion across the whole plate would still be way better than a foot in the middle. Less machine setup time would be required. More plug-n-play experience.
Stephen - I have 4 csm7100 essemtec machines which are functional. I am wondering if openpnp and your project could be used to bypass the main pc and incorporate a vision system. It uses the laser system right now. They easily place 0402 size. Would give these machines more capability. They are 4500cph.
What OS are you using @6.37 ? Is it Linux and I3? Or Windows with some window tiling tool? I guess the question is how did you get three windows next to eachother and re-sizable?
Not only machine imperfection, it would also remove gravitational influence. Everyone knows that a bridge will dip down under gravitational pull. The same thing happens to anything suspended by two points.
Have you factored in the bulge in your x-axis beam (that holds the z-carriage) while making measurements? I'm pretty sure you will have a sag on those aluminium extrusions for such a long span.
i would have gone with an extra member in the frame, and a big pcb. Maybe even integrating some of the wiring in it... But the steel plates look sexy! congratulations! :)
Shipping those are going to be the real test. It looks very bendable when thrown around, as packages often are. Or put into a bin with 50 pounds of weight on top of it.
You can get rid of the foot and most of the bowing by having a bend on the longitudinal edges… this must not add much to the fabrication cost… nor a big design change
and they don’t need to be 90 degree angles so you can get the parts stacked to save on shipping by having them a slightly less than 90 degree (85 would be fine I guess but you can do figure this out in CAD)
Yeah this flat steel part doesn't make much sense to me. Stephen chose steel so that things could be mounted magnetically to it and would be easy to reposition I guess. I'm not convinced the magnetic mount is particularly useful, and it makes this piece unnecessarily heavy to ship and to handle in general. Without one or two longitudinal bend it's going to get bent up in shipping, especially with that large hole in the middle creating a weak point. I think one or more readily-available aluminum extrusion rails would be as effective, using T-nuts to hold down whatever needs mounting. Or simply add a thin sheet of steel where needed on top of the aluminum rails, with the aluminum extrusion enforcing flatness.
the only issue with this type of "validation" is you can't take into consideration how straight your aluminum extrusion is what you should do is surface a spoil board on a CNC that will give you a parallel surface to your other axises Then mount the plate to be measured onto that it won't be as good as a surface plate but would be slightly better than your current approach and would be very accurate if you could get it done on something like a tormach HASS etc using the surface probe no need to make a spoil board.
but how about 1.5mm thick laser-cut and bent steel sheet for 20 bucks? I live in Ukraine where this kind of materials and labour is dirt cheap, so I would probably get such bent sheet for 10 bucks or something like that
Pretty much a waste of time and effort on Z, waiting with baited breath for repeatability of X and Y axes. That would have been my first priority, there is far less forgiveness on QFP 0.5 pitch in reflow than one would believe. Then you have to be careful the part does not move on the nozzle after camera alignment.
Am i the only one who wants him to go servo on index? Servo motors are fast,quite and accurate than stepper motors . The biggest reason i hate steppers is their creepy sound and missiong steps at high speed (with no feedback to the controller)
I like this series and your enthusiasm about the project. But I doubt this plate is the most logical solution for what it does. And more important: your measurements made absolutely no sense: you have absolutely no details on the straightness of the gantry. Keep in mind this mistake also invalidates any measurement you discuss in the future. In my opinion it’s better to keep it simple: “hey here’s a nice part that works for me. Keep in mind it isn’t 100% flat, so you need to do some checking and calibration yourself.”
See how much it would be to bend a 1/2" lip on the front and back. That would do huge things for the stiffness.
+1 on adding folded edges. Then it may not need to be as thick (save weight -> save cost of shipping)
That’s a solid idea, and we’ve talked about doing it. It’s a great trick to keep in the back pocket if it ever proves necessary! Thanks for the suggestion.
@@byoung255 good idea but unfortunately no. It would take up a lot more space, and space usually is what costs not the weight.
@@mikkelkirketerp4884 Good point. Shipping from supplier would be more tricky. They could be stacked in am alternating fashion.
@@lucianchapar just mounting parts on it can flex it at bit. These bends might help or perhaps just an alu beam beneath would counteract it and make it more stiff. 🤓
Still I don't understand why you have to use a giant metal plate to mount the electronics and pump and not just more aluminum profiles.
The "up" webcam goes there too. It does kind of make sense. Probably doesn't have to be the full length though.
We could! It's a good idea. Part of the reasoning was to have a nice surface to mount a board or other things to magnetically, plus having a plate gives us space to easily mount the up camera. If you build one with extrusion instead I'd love to see!
The magnetic surface and the regular pattern of holes provides a great utility for dealing with wire management and having a place to mount the electronics in a predictable spot for kit assembly.
As a part for your business to manage the outsourcing of, and as you show quality check, it seems as if this part has created too much complexity and cost for the utility it provides as a solution.
Your suggestion of a boot is a better, simpler and cheaper method to get the same repeatable flat surface to support an MDF, plastic, melamine, plywood or aluminum plate with similar results.
A couple of extruded aluminum rails could also span between the left and right sides with corner brackets.
Some people could easily source a piece of material cut to the right size then cut the holes and a correctly sized hunk of wood as the boot to prop up the center span.
@Stephen Hawes - you have already done an amazing job with this project and we see how excited you were to be able to outsource successfully from an external vendor, but it seems as though the plate part doesn't make much business sense for a DIY project except for those who would rather pay and extra $80 instead of cutting a piece of plastic or wood from a local big box store.
I like the strategy of getting product in the customer's hands as fast as possible, but it seems to me that finding a reliably sources and fairly priced solder paste nozzle/syringe head kit and nozzles for the pick place head are the parts I am having the most difficulty with sourcing.
Of course, the motherboard PCB, cameras and feeders (in that order) are the parts that allow someone to get most of the platform started.
You changed the manufacturer's part when you mounted your boards and other parts to it and then mounted it into the frame. Those will apply some unknown stresses and torques to the plate. So, what you're doing is in-situ testing, not acceptance testing. But, in the end that's what you want to test in your integrated environment, if the focus of the camera is sensitive at those error levels. Also as noted below, powder coating is not a precision consistent thickness operation, consider anodizing. I'm enjoying the series and might end up building one for my own PCB projects. Keep up the great work and I love your enthusiasm!
Four screws? Wouldn't you be better of mounting it on kinematic 3-balls if you want to measure the inherent flatness of the plate? Now you have copied any torsional inaccuracies of the aluminium profile into the plate, albeit reversibly (elastically). 6:05
Hey! Just FYI that dial indicators like the one you used usually have a serial port where you can read the measurement. I dont know if your particular one has one, buuuuut usually on amazon there are ones that do have a serial port and they are really cheap. Integrating the indicator in the system would be SO COOL! As for everything else.. keep it up!
0.8mm isn't "small". Consider anodizing instead to get more consistent finish thickness, and a edge flanges or mount to extrusion?
it's completely expected given steel's tensile strength. it's not the finish thickness that's an issue.
On my 1500x1000 CNC with 1100x800mm work space I have 0.05-0.1mm deviation and I think it is "big",
How does it compare with a much cheaper option? Like 6mm acrylic with a support foot in the middle? Easier to source (eg, lasercut at most makerspaces) and lightweight means reduced shipping costs too
Could the bend in the plate be influenced by gravity? Usually, strings that are laid across a distance will bend according to a hyperbolic cosine once they come to rest. Why not flip the plate and run the same test on the other side- if gravity is the factor that is responsible for the bend, you should see a nearly identical curve on the other side. Measuring discrepancies between the curves you get for each side would be an interesting way to evaluate the net curvature.
Flip the board downside up and measure again. Also the spring of your measurement device adds to the bow deflection. I could see it in the film, specially in the middle. And .8 of a mm is about 3 times the thikness of a resistor. Rather big
Nice, but a tiny bit disappointed you didn't interface to the SPC port on the Chinese dial indicator.
New to this Project, and loving what I see! Subscribed. Keep up the great work!!
Stupid question, but how do you know it's not the gantry which is distorted?
He could flip the plate upside down and redo the measurements. Subtract one set's xy location from the other set's. Zero means the gantry was bowed. Big difference means plate was bowed. Could figure a formula that gives error of each.
@@TheRainHarvester good point. Hopefully this will be covered in a future video, cos without it I dont really see the value in this data.
@@Ziraya0 yeap I think the mistake, if you can call it that, is to talk about the plate flatness, instead of the ability for the end of the nozel to track the surface of the PCB, or for the webcam to track a part held over it etc.
@@TheRainHarvester True. I also suspect that he would see an apparent bend in the other direction once he flips the plate, just due to gravitational pull.
Just a heads up your 3D printer style rail might actually be a slightly bowed. I've had this issue with sourcing linear rails also.
The only way to verify your measurements is with an precision flat plate. Then compare those results to your part.
Looking good! Loving this project! Have a good trip!
Proud of your work, man. Keep going.
I vote for showing off more of the index validation testing !! What are you trying to measure, and How you are measuring it ? Also what is the pass criteria ?
WTF are people downvoting, content on this channel is awesome....Keep doin what ur doing....I have received alot of value in ur work man....
So lucky to find you ✨
@Stephen I'm eagerly waiting for main Controller board and Feeders, can you please release beta versions? that's how you will get good feedback regarding issues and improvements.
+1 to FreeCad, would love to see a tutorial on how your transition away from Fusion 360 went, because I found FreeCad pretty difficult to go to after getting used to Fusion 360. It still takes me about 2x the time to make the part on FreeCAD :(, but I really want to support the project and make it better, so I'm pusshing through.
Maybe check out the OpenSCAD workbench for it. I *LOVE* how easy it is to create foundational geometries (or even the whole model) parametrically with just a simple script.
If the flatness of the steel plate is being driven by the camera's ability to focus, why not increase the the depth of field of the camera. You have plenty of light, so you could go to a smaller aperture thereby increasing your depth of field. See the online Depth of Field calculator for more information. Money spent on decreasing the camera's aperture with a special lens might be less than on the steel part (and you could use right angle stock to support the long stretch and stabilize it enough). Food for thought. You could also program for the components height and adjust your Z to the part to bring its bottom face within your field of focus.
for negating the bow of the plate, why don't you use 1 or 2 corner "L" profiles(like the ones used for the rails) on both sides of the plate, to add some rigidity?
Really cool ! There is something I don't understand, though : why not use a less expensive but more flexible plate (wood ?) supported by aluminium extrusions ? would'nt that be less expensibe and as good in terms of rigidity and precision ?
Wait, but if it's just a powder-coated laser cut piece of cold rolled steel, then the "flatness" can vary from spot to spot across the whole sheet, which the parts were cut out of. Isn't it OK to rely just on flatness of a single sheet of metal without any strengthening ribs?
I think you have to put at least some 20x20 or even 20x40 extrusion across the whole sheet, to keep it from flexing and reduce any potential curvature.
Edit: watched the remaining of the video, extrusion across the whole plate would still be way better than a foot in the middle. Less machine setup time would be required. More plug-n-play experience.
Stephen - I have 4 csm7100 essemtec machines which are functional. I am wondering if openpnp and your project could be used to bypass the main pc and incorporate a vision system. It uses the laser system right now. They easily place 0402 size. Would give these machines more capability. They are 4500cph.
What OS are you using @6.37 ? Is it Linux and I3? Or Windows with some window tiling tool?
I guess the question is how did you get three windows next to eachother and re-sizable?
I literally just found your channel and finished watching all the PnP videos and now there's a new one😂
Measure, flip top-bottom, measure, subtract. Should remove machine imperfection.
Not only machine imperfection, it would also remove gravitational influence. Everyone knows that a bridge will dip down under gravitational pull. The same thing happens to anything suspended by two points.
Cant wait for your feeders to become available. Looking forward to use them with my Liteplacer.
marlin does very well with UBL and this kind of distortion
I wonder how the plate would measure up if you flipped it bottom side topwise and rescanned it? I'd be interested :)
(Also gesund heit!)
Breaks are there for beeing taken. Awesome work ;-)
Have you factored in the bulge in your x-axis beam (that holds the z-carriage) while making measurements? I'm pretty sure you will have a sag on those aluminium extrusions for such a long span.
I have a question would this machine work with reprap and using the openpnp
Hello stephen i am also designing a pnp machine inspired by you but i can't get the blow off actuator to work ?
What kind of steel is it 1.4016?
Is it Tektronix tda220 oscilloscope behind you. I have same.
Hey man.. which tool are you using for data visualization?
Guess - python notebook
Yes - Jupyter Notebook - link in the description and here as I am just that service minded 😁:
github.com/sphawes/validation
@@TheStuartstardust thanks dear.
i would have gone with an extra member in the frame, and a big pcb. Maybe even integrating some of the wiring in it... But the steel plates look sexy! congratulations! :)
That crate is soooo cool
Nice video again :.-) Do you have the spec for that plate, like to make mine in aluminium and cnc it flat. :-)
Shipping those are going to be the real test. It looks very bendable when thrown around, as packages often are. Or put into a bin with 50 pounds of weight on top of it.
Hyped :D
I can *never* remember the name of the music at 5:00
You are the best.
I don't get it, why is a flexible plate used in a location where rigidity is needed?
You can get rid of the foot and most of the bowing by having a bend on the longitudinal edges… this must not add much to the fabrication cost… nor a big design change
and they don’t need to be 90 degree angles so you can get the parts stacked to save on shipping by having them a slightly less than 90 degree (85 would be fine I guess but you can do figure this out in CAD)
Yeah this flat steel part doesn't make much sense to me. Stephen chose steel so that things could be mounted magnetically to it and would be easy to reposition I guess. I'm not convinced the magnetic mount is particularly useful, and it makes this piece unnecessarily heavy to ship and to handle in general. Without one or two longitudinal bend it's going to get bent up in shipping, especially with that large hole in the middle creating a weak point. I think one or more readily-available aluminum extrusion rails would be as effective, using T-nuts to hold down whatever needs mounting. Or simply add a thin sheet of steel where needed on top of the aluminum rails, with the aluminum extrusion enforcing flatness.
the only issue with this type of "validation" is you can't take into consideration how straight your aluminum extrusion is what you should do is surface a spoil board on a CNC that will give you a parallel surface to your other axises Then mount the plate to be measured onto that it won't be as good as a surface plate but would be slightly better than your current approach and would be very accurate if you could get it done on something like a tormach HASS etc using the surface probe no need to make a spoil board.
Time for me to start learning how to make a pcb so I'm in time for the kit :)
Yay!
Congrats!
I’m still waiting for the next video
You should use your index machine...
to make boards for the index machine to sell
hell yeah!!!
but how about 1.5mm thick laser-cut and bent steel sheet for 20 bucks? I live in Ukraine where this kind of materials and labour is dirt cheap, so I would probably get such bent sheet for 10 bucks or something like that
Here to feed the algorithm
WHY NOT ALUMINUM??
Pretty much a waste of time and effort on Z, waiting with baited breath for repeatability of X and Y axes. That would have been my first priority, there is far less forgiveness on QFP 0.5 pitch in reflow than one would believe. Then you have to be careful the part does not move on the nozzle after camera alignment.
0.8mm 800 microns is huge
I want 5
Shiny
First😜
Am i the only one who wants him to go servo on index? Servo motors are fast,quite and accurate than stepper motors . The biggest reason i hate steppers is their creepy sound and missiong steps at high speed (with no feedback to the controller)
Terrified lol
AND ZERO FUCKS ARE GIVEN,...
He always looks like he’s pushing through a migraine.
😂
it's right but WAY way way over the top. with everything actually.
I like this series and your enthusiasm about the project. But I doubt this plate is the most logical solution for what it does. And more important: your measurements made absolutely no sense: you have absolutely no details on the straightness of the gantry. Keep in mind this mistake also invalidates any measurement you discuss in the future. In my opinion it’s better to keep it simple: “hey here’s a nice part that works for me. Keep in mind it isn’t 100% flat, so you need to do some checking and calibration yourself.”
Can someone tell me what drugs he's taking. He's wayyyy too excited over a piece of sheet metal.
Everything is Awesome,