Machine from Bing Luo

[IMN]

Greetings from Brazil,

Have anyone paid attention to the model developed by a person named "Bing Luo" ? He has a channel at youtube showing a machine made by him.

With two heads, I found it particularly fast for what seems to be a simple machine. And aparently precise.

Has anyone acquired such model or know something about it? 

The links is the following:

https://www.youtube.com/watch?v=VB8zvFTmsEo&ab_channel=bingluo

Iran

[serge]

Nothing spécial in thé vidéo. Classic openpnp machine. You have a link in thé description. Available on Aliexpress 

[IMN]

Hi Serge,

How is wheather in France today?

Yes, seems to be a normal version based on what I saw already. The fact is that at internet 90% or more of "projects" are abandoned, or are just a plain copy of someone else true work. But I guess it is not an internet thing, but a representation of life. 9 failures for each success story.

What troubles me is the possibility of embarking in one of those 90%, and I have no longer age to spare time or money.

But it is being a nice experience to learn about this tinny part of the world of technology. 

Thanks,

[Niclas Hedhman]

Bing Luo posts on this forum quite regularly[1], updating about his
progress and answering question.

https://groups.google.com/g/openpnp/search?q=bing%20luo

Niclas

> --
> You received this message because you are subscribed to the Google
> Groups "OpenPnP" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to openpnp+u...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/openpnp/78054ecd-1c95-4433-aff6-61654a65b699n%40googlegroups.com
> [1].
>
>
> Links:
> ------
> [1]
> https://groups.google.com/d/msgid/openpnp/78054ecd-1c95-4433-aff6-61654a65b699n%40googlegroups.com?utm_medium=email&utm_source=footer

[Mike Menci]

Look closer to the video - one nozzle is only in production - the 2nd one is for the show!

[IMN]

You are right, Mike. I did not noticed that.

[bing luo]

All parts can be purchased  in  many   online store  . and parts that cannot be purchased can be 3D printed. The core of Openpnp is its open source software and firmware，Hardware actually has little value.   If you want to save costs, the best way is 3D printing, only the running speed of 3D printing will be slower. but  this can save a lot of logistics costs and customs duties.   then  you can  build a machine for yourself with just  $200.

[IMN]

Hi Bing,

As I agree that certainly the most important part is the brain of the system (software) I dare to confront (just a mere opinion) the second part of the sentence in the meaning that hardware maybe should advance also. Since by one hand a "stablished/stabilized" hardware standard could be benefical from the point of view of software development (not having to "reinvent the wheel" all the time), it does not necessarily mean hardware should not advance at all. Or even be totally reconfigured at some point.

As for the use of cheap 3d printed parts, I guess that (from I read here), it could be a fine solution for those very limited in funds at the same time it degrades strongly the velocity/stability of the whole system. So it is up to each one what is best for ones pockets and objectives. The only thing I am against (not talking about you), is people promissing 'miracles' in terms of performance with 3d printed pnp machines. 

I saw recently a model made with whole parts (very little to assemble) made in 3d printing by a chinese engineer (using a very big 3d printer). But despite it made a precise printer, notwithstanding made a slow one, since it still lacks stability. Maybe for the sake of rigidity of a light hardware, the use of cables could possibly atenuate the inherent problem of the design. Something one may try, if not already done. 

My current opinion, as an "outsider", is that the use of lasers for distance measurement of all axis would free a lot the inventivity of hardware makers, and even cut costs, since no matter what, the software would find exactly the position in space. And such sensors are presently not pricey. 

Have you tried such approachs ?

[serge a]

hi,

Weather is nice today,  thank,

If you need a working system, You may also give a look to pandaplacer. However the project needs documentation. It looks great  but I can't find any review about it, except the video on youtube.

I Am also on the learning curve of building a simple machine using widely available parts and only 2 specifics openbuild compatible y motor plates.

Regards    

--
You received this message because you are subscribed to a topic in the Google Groups "OpenPnP" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/openpnp/CewSgbRGClI/unsubscribe.
To unsubscribe from this group and all its topics, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/66fa301b-f736-4048-aa31-c354a0458444n%40googlegroups.com.

[bing luo]

I don't plan to try any new solutions because the  mounting   machine is a very mature product, and the best DIYer  mounting   machine cannot compare in speed and accuracy with the JUKI2050 produced 40 years ago.  The value of OpenPNP lies in making expensive software open source, making   mounting  machines  cheap as  a  soldering irons.

If you need to use a mounting machine, then you can make it according to the recommended plan on the official website, which is definitely the lowest cost. Because the vast majority of innovation and creation have been proven to be meaningless. They are only suitable for enthusiasts with time and money. The strength and speed of 3D printing are sufficient for laboratory use, unless you plan to use it for production, there is no need to use metal.

[mark maker]

> at the same time it degrades strongly the velocity/stability of the whole system.

Are you sure?

I am not saying you are not right. Just asking whether you based this opinion on concrete facts/information available to you?

If I understand your example right (not sure), it is not the same.

We all agree, that you cannot 3D-print the rails and extrusions themselves. But the corner joints, the head chassis, and other such parts, I believe you can.

Especially, if you really design them right, i.e. do not try to mimic the typically slim metal parts, but design them much more boxy, to be stiff, while still being much lighter than metal (no need for a very high fill factor). You can also make fantastically 3D-shaped objects, where, for instance, the extrusions fits right in precisely, which nicely aligns them in all three dimensions. If designed right, you can let the elasticity of the plastic work for you and make the fit real tight (might need to print a few times, to get it right).

Metal parts are often just plates and angles with holes, and a lot of play in all but one degree of freedom, simply because anything more complex and 3D-ish becomes unaffordable, and/or unmakable in DIY. A 5/6-axis mill would be required, and could still not do the same stuff as a 3D printer. In reality, these rather simple plates are then often just "air-adjusted", while tightening a screw (I'm not saying it could not be done better, just saying this is the "makable" reality).

The low weight of 3D printed parts gives you less flexing in belts etc., less energy to dissipate when breaking, i.e. less vibration in a pragmatically dimensioned frame, so in the end, I would not a priori exclude the possibility, that a 3D printed design might be more precise, not less. With a lighter head, you can also let the machine run faster.

> My current opinion, as an "outsider", is that the use of lasers for distance measurement of all axis

Last time I looked, the lasers that have the large distance range we need (>500mm), are not nearly precise enough.

But people do use linear encoders.

Having said that: even a simple DIY "plastic wheels on extrusion and belt" machine has enough precision for PnP if done well, IMHO.

_Mark

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/ab346603-25bc-43a1-bd15-28a2a31d65fen%40googlegroups.com.

[Roland Exler]

I have bought such a hardware kit last summer directly from the microsmt website https://www.microsmt.com.cn. It looks like now there is a somewhat modified version available in between. I also bought all the remaining parts required (vacuum kit, dual nozzle head, nozzles, power supply), some of them no longer show up at the microsmt website.

Ordering has been easy, came well packed within short time (to Austria). My son has assembled and configured using a smoothieboard as controller. Assembly has been easy, instructions came with the kit. We added some 3D printed parts to mount the cameras

Never using OpenPNP before all calibration steps worked well and accuracy look fine. At least for the 0603 components which are my target (I would have to dig out the numbers if anyone is interested in details). Larger components like SOIC8, SOIC16 are fed as loose parts nice found by the vision system and picked up without any problems.

I also would like to note we are running the software on a notebook (i7-8550U, Windows 10) where we had to use an USB hub (USBC to 4x USB 3.0) and faced no problems with both cameras and the smoothieboard connected to this USB hub.

The most challenging thing to be noted has been the cameras, as both cameras show up in windows device manager with the same name, also in openpnp software. That would not be a problem, but on every start of the computer and OpenPNP the cameras are assigned in a random fashion as top and bottom camera. Finally we solved this by deleting all related configuration from the device manager, then connected the first camera and renamed all entries in the registry referencing to that name to something like "openpnp - top". Afterwards we connected the second one and did the same registry changes for the bottom camera, renaming that to "openpnp - bottom". From that point the cameras always have been assigned correct in OpenPNP.

We did a table with a wooden plate and a metallic sheet on top to be able to use magnet based holders for the PCB and strip feeders (as shown in the video). That's the weak point at the moment, as the (quite strong) magnets used there look like have magnetized my nozzle(s) and 0603 components stick on the nozzle for about 1 out of 10 components placed. I had no time to continue with this (hope to get back to this soon), but I will go on with either

a) Demagnetize the nozzles and use another design to mount the strip feeders and PCB on the table without the need for magnets

b) and/or combine with blow off.

Except for this magnetism issue so far all placements worked very well.

Roland

--

You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/78054ecd-1c95-4433-aff6-61654a65b699n%40googlegroups.com.

[IMN]

Hi Bing,

>> "Because the vast majority of innovation and creation have been proven to be meaningless." 

I understand the angle by which you consider it, but I have the terrible tradition on doubting even about my existence, so to speak. Therefore questioning about the fundamentals of everything I see is part of me. And I agree it is possibly unpleasant and somewhat the explanation why I tend to create so disconfort, not to say hate by merelly expressing my allways not standardized point of view.

But your comparison is correct about "JUKI2050 produced 40 years ago" being better than DIY models of today. Nuclear bombs where invented earlier than that and just a few countries have access to the technology.

By other hand, If the past is fix, the future is in constant mutation. Maybe one day every country may have access to nuclear technology (I hope not bombs), as well cheap PNP machines able to some level of production. When that day arrives, no country will be in darkness, and many small technicians will be able to fight his/her way out of poverty.

[IMN]

Hi Mark,

I was talking specifically about completelly 3D made PnP machines. As the one I mentioned about being made by a specific chinese engineer who are selling it at Taobao. To have plastic in many portions of a project cannot be deemed as incorrect or we would not being driving any modern car.

But a heavier body is a today common sense in the industry as a manner to counter the energy generated by mass x velocity of the head (linear momentum equation). What explains the weight of current professional fastest machines. But, more weight = higher price (structure + shipping).

Of course your idea of a lighter head is to be considered ideally in any DIY project to solve the equation above. But there is a limit to this, at the same time it makes difficult the use of multiple heads.

The idea of cables was to transfer the momentum of the head movement to a cheap absorbing platform, making it possible the use of more heads/weight. That's why I asked if someone has tried it before.

[IMN]

I was talking with this guy from PandaPlacer. And this is exactly the problem. He aparently sold a lot. But no reviews at the moment. 

Of course, someone has to buy first, or there would never be new suppliers/vendors. 

But the problem is the DIY area itself. Not everyone knows exactly what they are doing. So the risks are higher for the stand point of the buyer.

So it is your call. To risk or wait.

[Nimish Telang]

Emailed the MicroSMT person but haven't had a response, would be interested in people's reviews of it. It looks like a pretty standard (i.e. well done) XYZ+R gantry+good custom head. Which is really all OpenPNP needs. I have been toying around with a modified Cartesian design for a bit.

I don't really get the reason people are making their own kinematics. Tons of Chinese suppliers will sell you belt drive/ball screw drive with 50 micron repeatability XYZ gantries for < $1000. Then you can spend the $ on a good placement head, while reusing a 3d printer board for the motion axes and vacuum actuation, with camera input via USB. 3d print heads already have CAN boards for multi-sensor expansion which can be adapted pretty easily for the motion axes, vacuum or other GPIO.

Using 3d printed parts in the frame is, IMO, a waste. They are more expensive to make than gussets and brackets or blind joints for aluminum extrusions and are less stiff. Much easier to buy the brackets and endplates, then mount the gantry on that. As for the head, that's where 3d printed parts would make the most sense as you would want the flexibility in designing good mounts. Machined Acetal looks stiff enough for mount plates and that's not super expensive.

Overall the MicroSMT approach looks superior to the LumenPNP approach for frame+kinematics since it needs very little custom engineering so the $ can go into the head and other stuff.

Nimish

[justin White]

Lots of "idealism" in this thread.

>>"the best DIYer  mounting   machine cannot compare in speed and accuracy with the JUKI2050 produced 40 years ago. "

Yeah but do you want to put that thing in your house?

>>"We all agree, that you cannot 3D-print the rails and extrusions themselves. But the corner joints, the head chassis, and other such parts, I believe you can."

You can 3D print half of your machine, but then you wind up with a LumenPnP. Way overdesigned 3D printed joints, it looks flimsy and the biggest complaint from people who use it is that it is flimsy. You can get away with alot with fiducial alignment and all of the vision stuff but flimsy machines can only lean on that crutch so far.

>>"Especially, if you really design them right, i.e. do not try to mimic the typically slim metal parts, but design them much more boxy, to be stiff, while still being much lighter than metal (no need for a very high fill factor). You can also make fantastically 3D-shaped objects, where, for instance, the extrusions fits right in precisely, which nicely aligns them in all three dimensions. If designed right, you can let the elasticity of the plastic work for you and make the fit real tight (might need to print a few times, to get it right).

I've seen lots of DIY PnP designs, but can you show me one that has 3D printed plastic in it's structure that is comparable to just a reasonably designed extrusion/metal connector type machine? The problem with say a "boxy" large 3D printed joint is that it's exactly that, a large plastic joint. Whereas if you just use a simple metal plate for a joint, the joint can be almost non-existent, which is what you want. Same is true with plastic but then it's just a thin piece of plastic and well.

>>"Metal parts are often just plates and angles with holes, and a lot of play in all but one degree of freedom, simply because anything more complex and 3D-ish becomes unaffordable, and/or unmakable in DIY. A 5/6-axis mill would be required, and could still not do the same stuff as a 3D printer. In reality, these rather simple plates are then often just "air-adjusted", while tightening a screw (I'm not saying it could not be done better, just saying this is the "makable" reality)."

That's just not true. Something that's poorly designed with " lot of play in all but one degree of freedom," is just poorly designed but it's simple for that not to be the case. It absolutely does not require "a 5/6-axis mill", even super complicated parts done on a CNC mill are generally only done in 3 Axes, just multiple operations. The 3D printing fad opened up the DIY world to cheap and easy complicated structures, but somehow people think that that is what they need. A sheet of 1/4" (6mm) aluminum goes a very long way with an extrusion based machine.

In the machine tool world it's well known that 2 things can never be substituted. A) is mass and B) is rigidity. Anything that deals with precision needs some combination of these 2 things. I'm not even talking about milling machines and VMCs, just look at metrology devices, generally made from granite and cast iron, and all they do is measure stuff. An extrusion based machine with some cheap linear rails and some metal connectors is a compromise. To further compromise with 3D printed plastic is really going down hill. You could obviously make a plastic PnP machine with plastic wheels and it will work, but it won't be good. You can lean on a vision system for alot but turn it off any you'll see what kind of machine you have.

If you look at the lower connector on my machine....

You can't make that connection stiffer, lighter or simpler with a 3D printed part....I can guarantee it. Most of my non-extrusion parts are 1/4" aluminum that can easily be made by sending some DXFs to a waterjet shop. I also hear good things about an online place called "sendcutsend" but I've never used them. I have a CNC mill with a 4th axis so that's what I used but if I wanted to make more than just the 1 machine I would just have the stuff waterjetted except for the few parts that need more machining.

Not saying it's the grand wizard of DIY type machines, just saying that it's very stiff and accurate for a small extrusion machine.

Still working on electronics here...

[Zdenko Stanec]

>>"the best DIYer  mounting   machine cannot compare in speed and accuracy with the JUKI2050 produced 40 years ago. "

Challenge accepted! 😊

https://www.youtube.com/watch?v=XpU1aV3POuQ&lc=Ugx1-eEhNc1Xk4xPx_N4AaABAg.9pSFMHr_vBa9pSLydbn2_n&ab_channel=ZdenkoStanec

Zdenko,

[justin White]

Definitely fast but ya gotta put something on a PCB! I don't think he was counting on someone to show up with a granite base when he said that.

Can OpenPnP be setup for fly-by vision? looks like you'd waste alot of time on part alignment.

[Zdenko Stanec]

Will do, still in the building process, need to finish the conveyor.

As far as I know, OpenPnP still does not support fly-by, and for that, you would need a global shutter camera too, but I think not lowering Z to the Bottom cam would help to speed things up a bit, we will see when I finish it.

Apologies for off-topic,

Zdenko,

[bing luo]

Zdenko ,

           JUKI2050 looks outdated. It is still driven by a belt。 but it can still be seen in  many  small SMT factories。 Since you are willing to spend so much money on manufacturing the SMT machine, have you used a grating ruler closed-loop drive?

[bing luo]

this  is  JUKI 2050

[justin White]

I'm still working on mine, wont be anything like yours though. I was planning on using a global shutter camera. I do some other high speed machine vision stuff so I know a global shutter is a must. 

I found these but didn't really look into the lens focal length yet. From the specs I did see it looks pretty good for a cheap camera. https://www.arducam.com/product/120fps-global-shutter-color-usb-camera-board-1mp-ov9782-uvc-webcam-module-with-low-distortion-m12-lens-without-microphones-for-computer-laptop-android-device-and-raspberry-pi-arducam/

The cameras I use for my other projects are over $1k each, not really the type of thing I wanna use here

[justin White]

>>"grating ruler closed-loop drive"

Must be a bad translation or something.....Do you mean linear scales or something?

[Zdenko Stanec]

@bing No, I have my limits in building the machine and I am happy with AC servos and 17bit encoders with C5 grade ballscrews precision.

I am not spending... I am investing :D...

Zdenko,

[Nimish Telang]

No he's correct. Optical gratings are the scales used for linear encoders, alongside magnetic ones.

Ic haus has a few linear encoder dev kits that are cheaper than buying your own encoder 

---

From: ope...@googlegroups.com <ope...@googlegroups.com> on behalf of justin White <blaz...@gmail.com>
Sent: Wednesday, May 10, 2023 7:35:40 PM
To: OpenPnP <ope...@googlegroups.com>
Subject: Re: [OpenPnP] Re: Machine from Bing Luo

 

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/64a7ca68-987c-45e6-8382-c9f363ab6cd3n%40googlegroups.com.

[bing luo]

Nimish Telang

         It is best not to use 3D printing on the mount. For sufficient strength, your 3D printing needs to be very thick, and the mount will become very large. Moreover, due to the Z, C1, and C2 stepper motors in the mount, they will generate heat during operation. Both PLA and ABS materials will deform due to the heat, making your Z-axis unable to be horizontal.  

         For PNP, 3D printing is definitely not a good choice, it is just very  cheap, the cost is almost just electricity. but  it won't take long  time ,  the machine will  need to reprint the accessories due to deformation, Reconditioning the machine will make you feel very troublesome.       

         No matter what accessories you purchase from China, you will have to bear shipping costs, customs duties, and value-added tax.The cost is still much higher than 3D printing, although the prices of these accessories in China are very low.

[Ben]

Hi

I assume that you saw that the machine is with basic performance, works nice, but If you want to know if you can relay on Bing Luo, I bought from him several products too, quality is good, communication and support is good too, and all products came well packed in a very good condition. I can recommend buying from him.

Mike, in the video only 1 nozzle is working, but as openpnp support this twin nozzle format, no problem to work with 2 nozzles. machine has a valve support for the 2nd nozzle.

Ben

[justin White]

No he's correct. Optical gratings are the scales used for linear encoders, alongside magnetic ones.

Ic haus has a few linear encoder dev kits that are cheaper than buying your own encoder

I'd never heard that term, but it sounds like we're talking about "linear scales" which is what we call em around here

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/BL3PR04MB79318F0A3DE6BB9A9EB23A3BA7779%40BL3PR04MB7931.namprd04.prod.outlook.com.

[mark maker]

> Whereas if you just use a simple metal plate for a joint, the joint can be almost non-existent, which is what you want.

Why would I want that, specifically? 😉

> You can 3D print half of your machine, but then you wind up with a LumenPnP.

I beg to differ. Just because one 3D printed design sucks (if you say so), you can't conclude it must be always so.

> That's just not true. Something that's poorly designed with " lot of play in all but one degree of freedom," is just poorly designed but it's simple for that not to be the case.

OK, taking rotation into account, it should say " lot of play in three out of six degrees of freedom,"😅

What I actually meant is this: If you put plates on the extrusions, they can swim in any direction (e.g. X/Y) except against the extrusion (e.g. Z). They can slightly rotate  (e.g. C, but not A, B). There are usually wiggly extrusion nuts inserted, the holes in the plate have play against the screws, some washers are - "washing around?" - and then you press everything together, hoping it is aligned and squared and tighten the screw ... and often curse because the whole sandwich is shifting on the final turn of the screw, because the gripping head/washer unevenly drives it to one side. Like I said, that's just the practical take, I know mechanical geniuses always have some rig or other to align stuff perfectly.

> can you show me one that has 3D printed plastic in it's structure that is comparable to just a reasonably designed extrusion/metal connector type machine?

Unfortunately, I know no existing machine to show exactly what I mean, really.

This guy's machine has some elements that go in the right direction... but then other elements, I'm not so sure.

https://youtu.be/Q3Tuo8eVnu4

> In the machine tool world it's well known that 2 things can never be substituted. A) is mass and B) is rigidity.

About A) Agree. The question is how much rigidity is needed for PnP, and how much you are prepared to pay or sweat for it.

About B) I believe one needs to differentiate. Mass is good for the machine base, sure. But it is actually unwanted for moving stuff. Often you get more and more weight, that requires more and more stiffness, that requires more and more weight to fix, ...

> parts are 1/4" aluminum that can easily be made by sending some DXFs to a waterjet shop

Stuff like this is very expensive here in Switzerland, shops usually wrinkle their noses at such small orders and set prices accordingly.

> I have a CNC mill with a 4th axis so that's what I used

Good for you. Most people don't have a mill. Or maybe we should all make this one? 😎

https://youtu.be/ctyLjOHg7Ag

All I'm saying is that, in the PnP use case, for the same amount of cost and effort, the 3D printing option should not be dismissed all too quickly.

_Mark

You received this message because you are subscribed to the Google Groups "OpenPnP" group.

To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/d73e12d7-fed6-42e6-8a4d-3922ddb91ffbn%40googlegroups.com.

[justin White]

Why would I want that, specifically? 😉

Moar better. If the flattest, squarest, most rigid pieces involved in the joint are the extrusion why would you want to introduce 3D printed plastic or an air gap between them?

I beg to differ. Just because one 3D printed design sucks (if you say so), you can't conclude it must be always so.

Well I don't have one, it's my opinion that it looks like it sucks, but that's subjective to what it costs. If it were ~$400 it'd be great. Huge effort was put into complicated shapes to make something as cheap as possible but it's not sturdy and it's not sold cheap. That PnP in the video actually looks pretty good. That's not to say that that same machine couldn't have been made better with simpler metal shapes but I have to admit that's pretty well designed and I really can't find alot of fault with it for what it is other than the fact that it's only 3 pieces of extrusion and using 2040 would have made it an awful lot better at very little cost. I think the router is just silly. Not for lack of effort but there's no good reason to make it like that with what's readily available, but it wouldn't have been a good video.

OK, taking rotation into account, it should say " lot of play in three out of six degrees of freedom,"😅

What I actually meant is this: If you put plates on the extrusions, they can swim in any direction (e.g. X/Y) except against the extrusion (e.g. Z). They can slightly rotate  (e.g. C, but not A, B). There are usually wiggly extrusion nuts inserted, the holes in the plate have play against the screws, some washers are - "washing around?" - and then you press everything together, hoping it is aligned and squared and tighten the screw ... and often curse because the whole sandwich is shifting on the final turn of the screw, because the gripping head/washer unevenly drives it to one side. Like I said, that's just the practical take, I know mechanical geniuses always have some rig or other to align stuff perfectly.

I pretty much knew what you meant, slightly overblown though. all of those concerns are really set aside by A) making sure the ends of the extrusion a square and B) choosing the right type or number of fasteners. That lower corner connector is a good example. It only has 10 M5 fasteners in it, not because it needs the clamping force, but because A) if you lightly snug 3 that hold 1 piece of extrusion it's not going anywhere when you crank down on the 4th. And B) when all 4 are tight it's just not going anywhere after that. Flat washers are not good to use on things that load from the side and it wasn't necessary but it would have been better to use countersunk screws on the 2 screws that tap into the end of the 1 extrusion. The 2 tapered heads opposing each other would lock any movement. Some places shoulder screws work better just to take slop out of 1 or 2 holes and align the part, while cap screws lock the part down from the other holes. All this really amounts to is "what screw do I use here and how good did I drill the hole" It's not rocket surgery. Extrusion is generally pretty straight along it's length so if you cut the ends square there's no fighting.

About B) I believe one needs to differentiate. Mass is good for the machine base, sure. But it is actually unwanted for moving stuff. Often you get more and more weight, that requires more and more stiffness, that requires more and more weight to fix, ...

Well....you're right there. For a PnP with no real benefit to mass in the moving components, it is a major hindrance to speed when acceleration has the whole rest of the machine swinging all over the place. This is 1 of the reasons the 3D printed router you showed is so bad. Cutting tools need mass to dampen cutting forces. Routers get away with low rigidity/mass compared to milling machines because using a very high speed spindle like a router in soft materials means the chip load for every cut is much smaller. That only goes so far though, that thing had a tough time with aluminum and mild steel would be completely out of the question.

Stuff like this is very expensive here in Switzerland, shops usually wrinkle their noses at such small orders and set prices accordingly.

I'd think Switzerland would be the land of opportunity for stuff like this. Is an internet based order out of the question?

Good for you. Most people don't have a mill. Or maybe we should all make this one? 😎

That's not the point, and that's not a milling machine, My mill is just my metal cutting machine. Maybe my metal cutting machine is better than yours. It's not a Haas, it's a small manual mill converted to CNC by me but it is cast iron with balll screws....It costs 2 or 3 LumenPnPs to build lol. Point I was making is that I used it because I have it, not because I needed to. I actually would not use it for most parts if I made many because It's better worth my time to just have a specialized outfit handle it.  Your location and access to services aren't an excuse why 3D printed parts would be better than metal, it's just an excuse as to why you would use 3D printed parts over metal. If I knew I couldn't get access to reasonably priced CNC milled parts, I would have designed those 8 or so parts that absolutely need it differently so it could all be water jetted overseas somewhere. Shipping cost is negligible because the smaller aluminum parts probably actually weigh the same or less than bulkier plastic parts.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/e757255a-352f-a5f6-bf87-0c339365ced0%40makr.zone.

[mark maker]

> If the flattest, squarest, most rigid pieces involved in the joint are the extrusion why would you want to introduce 3D printed plastic or an air gap between them?

Good point, see below.

> That lower corner connector is a good example. ...

Sorry, I don't know enough "Mechanics English" to understand all the specific terms in that whole section, so it is likely I'm missing something. 😁 Therefore take the following with a grain of salt:

From the image you posted, I get the impression that the plates are just "slapped on" and the squareness of the extrusions against each other is not enforced in any way other than you desperately trying to press them together laterally while tightening the screws.

From my gut feeling, the usual "cheap" way of tapping the hole(s) in the first extrusion profile, and, through a hole across the second extrusion, screw them together, gives you more assured squareness (assuming the tapped extrusion was cut with a precision tool). I've seen solutions going all the way through the second extrusion, others use special screw heads matching the slot of the second extrusion, just clamping it from there.

(Image: Liteplacer)

Such a connection could easily be combined with 3D-printed "shoe" that gives the precise lateral alignment (eliminating the last degree of freedom, indicated by the green line of the image). If the extrusion was boxy enough, such a through-connection would also be quite stiff, as the screw is directly load-bearing. Otherwise the stiffness could be improved by the 3D-printed shoe being boxy, and shaped all around (like two halves with the slots counter-moulded, and then screwed together).

In contrast, if you think about it, plates actually work through the lateral surface friction between plate and extrusion, i.e. the actual plate screws are 90° to the planar forces effective for the stiffness we are talking about, i.e. the screws are not load-bearing, they are merely upholding the friction. So I'm not so sure how stiff such a plate construction actually is, especially in the long term, once a machine is subject to constant load cycling. It is likely okay for our light machines, but could be a concern for heavier and "crackier" loads.

> This is 1 of the reas