I'm looking for a good tool to make the 3D scanning of my mini-pc using the photogrammetry and my Kinect 2.

[ziomario]

Hello to everyone.

I've assembled all the hardware components to create a mini-pc / tablet based on FreeBSD. I found every piece and now I'm planning to make the case. I need to create it using a CAD tool,but I haven't any knowledge about CAD. But I've got an idea. I think to be able to make the 3D scanning of the device using what I have available here : the Kinect 2. I could learn how to make the 3D scanning using the photogrammetry. Once the model will be ready I will give it to someone who knows how to use CAD, to help him / her understand how to draw the case. I would like to have some suggestions from you about what could be the best software that I can use to achieve my goal. It's a very good idea if the tool is open source,for sure. I can't spend a lot of money in licenses. Furthermore,I should convert the model produced by the photogrammetry tool in DXF or DWG format,because these are the format requested by the 3D company that will print my case.

Please help me,thanks.

 

[S-TKUC]

To avoid misunderstanding, we will refer to the case as a part as the housing.

As a premise, the only file you need to pass on to the housing manufacturer (3D printing company) is the CAD data of the housing.

The only data you capture with the 3D scanner is the shape of the existing part.

In other words, you do not need to pass on any of the data you captured with the 3D scanner to the company.
There is no need to convert the data captured with the 3D scanner into DXF or DWG.

You only submit the CAD file of the housing that you designed in CAD to the 3D printing company.

It is definitely best not to send the internal parts.
Unless you want a plastic mock-up of the RockPro64 or power bank...



Are DXF and DWG really the only file formats?
For 3D printing, the most common file format is STL.
Recently, STEP has become more widely available.

DXF and DWG 3D data have compatibility issues, so detailed confirmation with the contractor is required, which is beyond the scope of a CAD beginner.

In my experience, DXF or DWG are only requested in cases where the housing is processed by cutting or injection molding.
If a company mainly handles cutting processing and also provides 3D printing as a secondary service, they may instruct you to prepare a DXF or DWG file on their website.



3D scanned data is very large, so you may need a powerful CAD workstation PC.
I would imagine it would be difficult to process even on the workstation I use for work.



The part doesn't need to be in perfect 3D.
You only need to simplify and model the necessary shapes, so it shouldn't be that difficult.

In fact, this process will give you the experience you need before designing the housing in CAD.
I recommend starting with manual modeling rather than 3D scanning.

I think you should only use 3D scanning once you understand how it works.

Here is an example of a part I created for 3D analysis.

Compared to the original part, it has been simplified to just the shape of the board to be screwed in and the position of the connector to connect the wiring.

This was enough for the eGPU BOX housing design I am working on.

Basically, 3D modeling of a circuit board can be done by making it a simple rectangle, ignoring some details, or just pasting in a photo.

Here is an example of a RaspberryPi Pico 3D model I made.

This is the 3D version of the product W6100-EVB-Pico.

 

[ziomario]

But your simplified 3D models do they contain the measurements of the various parts ? AFAIK only the models created using CAD software contains them. So,I feel obligated to learn how a CAD tool works. I'm not sure if I can use and learn different 3D tools,easier to learn than CAD tools, to produce the simplified models that you are talking about. Actually I have only a decent knowledge of Blender,anyway. So,to make things easier,I think that the most obvious thing to do is to use Blender. But AFAIK Blender is not a CAD tool. I know that I can install some addons and it can act as a CAD tool,but then,I should learn how to use Blender as a CAD tool.

 

[lilhasselhoffer]

So...you seem to be missing a few steps along the way...and I'm not sure if it's the rough english or a lack of understanding. As such, if I say something that you already know please forgive the repeat because I don't want to make bad assumptions.

Your goal is to take an object, scan with a Kinect, import data into a CAD program, generate a rough 3d object from the data, and then use this 3d object to model into an assembly which will have a housing. That's what I'm getting from your request...filtered into a usable set of steps.

So...if this is your goal, you are going to need some tools. I'm going to pretend that you are a professional, and move backwards from that. Your tool/software list is:
1) A scanner
2) A software to translate the scan data into a usable matrix
3) A software to translate the matrix information into a usable CAD file

Professionally, this might be:
1) Creoform for the scanner
2) Polyworks for the scan data pull
3) Polyworks for the translation

Personally this might be:
1) Kinect+SBC (single board computer) to pull data
2) Skanect to transfer the point cloud into something usable
3) One of dozens of CAD programs to transfer it into something usable...and not Blender. Blender is a graphics utility, and it's not going to output something usable for engineering without some significant plugin work


So, my recommendation to you would be to start with a tutorial:
Scanning tutorial
From there, understand your tool and the output.
Librekinect on github
I'd suggest the Kinect is not a tool designed for high precision. You are asking for it to scan in what looks like the millimeter accuracy range...and it was a thing designed to scan for gestures in a space the size of a living room...so maybe you should be wary about using the right tools for the right job. Otherwise, best of luck. There were people who loved the Kinect 5+ years ago. I'd suggest that you instead consider buying a tool designed for what you want...as the Kinect is literally dead tech (Amazon listing for open source scanner). What you are asking to do is delve into decade old articles, to resurrect decade old tech, to maybe save a bit of money over a modern and more tailored execution.

 

[S-TKUC]

As lilhasselhoffer points out, Kinect V2 would not be able to 3D scan with the precision you need.

But your simplified 3D models do they contain the measurements of the various parts ? AFAIK only the models created using CAD software contains them. So,I feel obligated to learn how a CAD tool works. I'm not sure if I can use and learn different 3D tools,easier to learn than CAD tools, to produce the simplified models that you are talking about. Actually I have only a decent knowledge of Blender,anyway. So,to make things easier,I think that the most obvious thing to do is to use Blender. But AFAIK Blender is not a CAD tool. I know that I can install some addons and it can act as a CAD tool,but then,I should learn how to use Blender as a CAD tool.

The important thing is to enter the necessary dimensions only for what is necessary.

For MLCCs, ICs, chip resistors, etc., in most cases, just the "height of the tallest item" and the "area where the components are located" will be sufficient.

You only need to measure the dimensions of the elements necessary for the enclosure design.


It's a shame that RockPro64's 3D files aren't publicly available.

The precision required for 3D printed chassis design is less than 0.2mm.
A vernier caliper can measure to 0.05mm.
An example of a caliper you can purchase.
Or a digital caliper.

When designing like this, I prefer plastic calipers.
Although their accuracy is inferior to metal ones, it is important that they do not damage the object being measured.
An example of the caliper I use when designing housings.
It is true that changing the batteries in digital calipers is a hassle...


I always hope that manufacturers of embedded SBCs release 3D models of their boards.
Some manufacturers do, so if I have a choice, I choose that (Connect Tech Inc, for example).

When using AVerMedia's EN715, 3D models were not distributed, so I will share an example of what I did to convert it into 3D.

So...you seem to be missing a few steps along the way...and I'm not sure if it's the rough english or a lack of understanding. As such, if I say something that you already know please forgive the repeat because I don't want to make bad assumptions.

Your goal is to take an object, scan with a Kinect, import data into a CAD program, generate a rough 3d object from the data, and then use this 3d object to model into an assembly which will have a housing. That's what I'm getting from your request...filtered into a usable set of steps.

So...if this is your goal, you are going to need some tools. I'm going to pretend that you are a professional, and move backwards from that. Your tool/software list is:
1) A scanner
2) A software to translate the scan data into a usable matrix
3) A software to translate the matrix information into a usable CAD file

Professionally, this might be:
1) Creoform for the scanner
2) Polyworks for the scan data pull
3) Polyworks for the translation

Personally this might be:
1) Kinect+SBC (single board computer) to pull data
2) Skanect to transfer the point cloud into something usable
3) One of dozens of CAD programs to transfer it into something usable...and not Blender. Blender is a graphics utility, and it's not going to output something usable for engineering without some significant plugin work


So, my recommendation to you would be to start with a tutorial:
Scanning tutorial
From there, understand your tool and the output.
Librekinect on github
I'd suggest the Kinect is not a tool designed for high precision. You are asking for it to scan in what looks like the millimeter accuracy range...and it was a thing designed to scan for gestures in a space the size of a living room...so maybe you should be wary about using the right tools for the right job. Otherwise, best of luck. There were people who loved the Kinect 5+ years ago. I'd suggest that you instead consider buying a tool designed for what you want...as the Kinect is literally dead tech (Amazon listing for open source scanner). What you are asking to do is delve into decade old articles, to resurrect decade old tech, to maybe save a bit of money over a modern and more tailored execution.

The Open Source Ciclop DIY 3D Systems Scanner Kit is very interesting.
Unfortunately, the Github repository that this product references hasn't been updated in 10 years... I'm not sure if it still works.

As is clear from the product video, it seems that holes appear in the 3D model.
I recognize that such holes can occur even when using expensive 3D scanners, due to factors such as reflections from the object and ambient light.
I have also tried a 3D scanning app on my iPhone, and had a similar experience.


For ziomario, I will list some of the issues that arise when using 3D scanned 3D data to design an assembly on CAD.

- 3D scanned 3D models have no reference planes.
In 3D CAD, assembly design is done by specifying the distance and angle between reference planes.
In other words, 3D scanned 3D models are not suitable for assembly design.
Correcting this requires intermediate or advanced 3D CAD skills.

- You need to select a 3D CAD system that is compatible with the file format of the 3D scanned model.
When using 3D scanning with older technology such as Kinect, it is expected that the file formats that can be output are limited.
When using with 3D CAD, you will need the know-how to select and set up a file format that can correctly handle dimensional units, etc.
This can be difficult for 3D CAD beginners.

- It is difficult to deal with the accuracy of 3D scanned 3D models and issues specific to 3D scanning.
Designing a housing for a 3D printer requires an accuracy of ±0.1mm in small details.
Only very expensive commercial equipment can achieve high accuracy with 3D scanning.
I don't think this is suitable for your purpose.

 

[ziomario]

That's too much for me. I can't afford that level of knowledge and experience. But I've got an idea to be able to accomplish my project with a moderate level of effort. I want to create a temporary model of the case (or enclosure ?) for my mini-pc / tablet using a box of polystyrene that has more or less the correct same shape and size of the right one. It should not be perfect. I can refine it by myself using a rasp. I can make the holes exactly on the right places for example. Once the model in polystyrene is ready,I can send it to a 3D CAD artist / modeller by regular mail. So that he / she can make the digital 3D model having the physical object under her / his eyes. I've started looking for a 3D printer company that can use the polystirene as material. I've chosen this material because it is easy to work with it. Maybe there is some different material that's easy,too. Anyway,I'm having a lot of difficulty finding a company that is able of creating a box like the one I need. They say they do not have the right mold. Strange, because on their official sites it is written that they can work and create any shape using the polystyrene. What do you think about my idea ? Can you reccomend some company that can ?

 

[S-TKUC]

This involves making a mock-up of the case out of polystyrene, and then having them replace it with another material somehow.

This is possible, but it will be difficult for the company to give you an estimate.
You will need to provide the company with information about what kind of case you want in advance.

If you want to find a company who can take the one-of-a-kind original model and complete the case, you can expect to pay a considerable amount.
I wouldn't want to take on such a job because it's too risky...

It took 3 minutes to create a 3D model of the M.2 2280 NVMe.

Measuring an object with calipers and creating a 3D model is something even a CAD beginner can do.
This is why I recommend modeling in CAD.

I think CAD is a lot easier than you think.

It's easy if you just make the parts.

Assembling and designing them is a bit of a challenge, but...
Once the parts are converted into 3D CAD data, you should be able to send that data and make a request.
You can explain the finished product with hand-drawn illustrations or photos of the actual item.

I think this approach will be easy for many companies to accept.

 

[ziomario]

---> If you want to find a company who can take the one-of-a-kind original model and complete the case

I want that the company makes only a generic box that has the correct height, width and depth. I will finish / refine the box my myself,creating the holes using a rasp. For the rest of the modifications that I can't do manually,they can be determined by examining the basic 3D model of the box in STL that I've got from the RockPro developers.

Update : It seems that the 3D companies want only the 3D model already completed. At this point I'm forced to find a box that has been already created and that has the right dimensions by myself on the internet.

 

[qxp]

Look, learning how to do good output from kinect will take longer than learning CAD. Try FreeCAD, simple enough to use and you can 3d-print a prototype box yourself.

 

[ziomario]

Look, learning how to do good output from kinect will take longer than learning CAD. Try FreeCAD, simple enough to use and you can 3d-print a prototype box yourself.

Using Kinect and Skanect was the old idea. I've explained the new idea on post #6 and #8.

 

[DirtyDingusMcgee]

So i've been watching this project roll along. What is the end goal of this? Is all of it just so you can have one, single FreeBSD tablet? Is this something you are expecting to be commercially viable? Do you expect to sell a lot of them? 10? 100? 1000? 10,000? Going from prototype to a production unit is going to cost a lot more than I think you think it will.

 

[S-TKUC]

So you choose the dimensions you need from a selection of finished cases, purchase them and then process them yourself?

In those cases, I often look for them at TAKACHI ELECTRONICS ENCLOSURE CO., LTD.
PLASTIC ENCLOSURES・HANDHELD ENCLOSURES
The case your LCD screen will fit in is limited to the PF series PF24-4-24W, etc., and you probably won't like the size...
or
ACRYLIC CASES・SKELETON CASES・TRANSPARENT ENCLOSURES

This product might be good. CUSTOM SIZED ACLYRIC BOX with CORNER GUARD - SKGF series
SKGF□-□-□K15□・W201-250mm, D101-150mm

The material is acrylic, so it does not meet your requirements...
But among the company's cases, it has the smallest difference between the outer and inner dimensions, and additional processing services are also easy.


For me, it's easier to 3D print a case myself than to find an existing one.
I think it's easier to learn 3D CAD.


Honestly, I think it's necessary to start by learning the basic steps of manufacturing...

You should find a way to realize it from "what has been done," "what exists," and "established methods."
When you find a problem, look for a solution.
If you can't think of a way to realize it, the quickest way is to improve your skills.

It's a waste of time to diverge your thoughts on imaginary things, hypothetical things that might be possible, or unknown methods...

 

[DirtyDingusMcgee]

Honestly, I think it's necessary to start by learning the basic steps of manufacturing...

You should find a way to realize it from "what has been done," "what exists," and "established methods."
When you find a problem, look for a solution.
If you can't think of a way to realize it, the quickest way is to improve your skills.

It's a waste of time to diverge your thoughts on imaginary things, hypothetical things that might be possible, or unknown methods

Well said. You aren't designing an A-12/SR-71...

You are designing a cheap tablet in a market that is flooded with cheap tablets. Just the guts alone cost more than many tablets of similar power.

FreeBSD is not going to be much of a selling point. Does it run WhatsApp, youtube, music, and porn for less than $100-200? That is the concern of the cheap tablet market.

 

[S-TKUC]

I have predicted from the beginning, including in other threads, that you would need to improve your skills.
That is why I have been trying to advise you to clarify the direction of your skill improvement and what you need.
You have to realize for yourself that you need to improve your skills, and pursue it with ambition and high motivation.
This is very important.

But you have not shown any intention to improve your skills.
You have not even once calculated the formulas and methods of calculation that I have already clearly shown you many times.
No matter how much I showed you how to calculate battery life, you just kept asking questions until I gave you an example of how it works.


After that, it's up to you to do as you like.
You shouldn't involve other people in your activities.

*I'm only talking about your attitude so far.

If you change your mind and decide to improve your skills with ambition, I'll be happy to answer questions again and help with any guidance you need.

 

[Dr. Dro]

Interesting thread. I have a question, semi-related I guess. I have an original "Kinect for Windows" kit here, which I suppose is the Xbox 360's version of the Kinect. I never used it, my brother used it once I think? It's been on the back of my drawer for a few years now. Can it also be used for this purpose? Might look into it sometime.

 

[qxp]

Using Kinect and Skanect was the old idea. I've explained the new idea on post #6 and #8.

Not good ideas either. How much money can you spend making and mailing prototypes back and forth ? More than likely the first ones will not have parts fit, or the holes won't align, or so on. You'll spend more time using the rasp than it takes to learn CAD.

 

[lilhasselhoffer]

Interesting thread. I have a question, semi-related I guess. I have an original "Kinect for Windows" kit here, which I suppose is the Xbox 360's version of the Kinect. I never used it, my brother used it once I think? It's been on the back of my drawer for a few years now. Can it also be used for this purpose? Might look into it sometime.

Yes, and no.

The Kinect was a tool designed with a certain level of accuracy in-mind. It uses different scanning, namely a combination of visual calculation and audio cues, to extrapolate a 3d environment. The hope, nearly two decades ago, was to take this piece of hardware and make it amazing. Unfortunately, a lot of the three dimensional mapping was highly tied to the xbox...and it took a while for people to get to the point where they could take a relatively cheap Kinect, slap it to a Raspberry PI, and have an enclosed 3d sensor that could give them something.

In the process, I was not aware of them ever being able to be exceptionally precise. Without known points that never move, the 3d calculations were largely about motion in a three dimensional space, which is why things kinda went pear shaped once you tried to use the thing as a measuring device. The best examples I've been able to find are novel approximations of 3d solids, like human beings. It never, to my memory, was usable as a competent 3d scanner down to mm range precision...as the OP had somehow assumed?



To the OP...you have a vision, 1/4 of a thought process, 1/10th of a plan, and 0% of the knowledge required to deliver. Based upon your multiple threads, let me start this by saying you really, really, really need someone to help with your thought process. That may sound insulting, but you cannot seem to articulate what you even need help with.
What I think your endgame is, is this:
1) I have the 3D filed for a single board computer.
2) I have purchased a battery.
3) I want to shove the single board computer into and enclosure, along with the battery, and make my own custom computer enclosure.
4) I need somebody to print a custom enclosure that will both fit my selected board, and my battery, along with giving me access to the I/O from my board of choice.
5) I have zero 3d modelling skills. I also have no hardware, and no desire to learn anything to make this happen.
6) I have no set budget for the project. I have no idea how to build a charging circuit into the computer. I also don't seem to know how to communicate with engineers...so I need interpretation or a really good plan.

I would suggest that you not proceed with your current project until you can either assess your skills and understand what you actually need, or you can understand something of worth regarding the rest of this project. Let me help you with some parting education:
1) Polystyrene is almost exclusively a foam. Styrene is a solid. People don't print in either because they are brittle. Models cut out from them, usually via hot knife, are not dimensionally stable or durable.
2) Most people who will print you anything need either an .stp or .gcode file. One is a model, and the other is code used to guide a machine in fabricating your stuff. These are not the only option, but you'll note that "shipping them a thing" is not on the list.
3) You genuinely seem to not know what you are doing...and definitely shouldn't design a charging circuit. For safety's sake, please consider this before building anything.
4) You seem to want to use this as a replacement power pack+case for a tablet that isn't working...given that you used an SBC in the last thread. Why in Hades do you have the board, a monitor, and a battery pack, but no case? If you've got all of this together, and intend to do what it seems like you are asking, please consider that when you burn your house down we are not responsible. Lithium ion batteries need a special charging circuit, not just to have power pumped in like a lead-acid battery. SAFETY FIRST.
5) Maybe go to Hackaday.com. They're more into crap like this...and they might be able to hook you up with a local makerspace that could help get the ideas out of your head, and put them into a workable format.

 

[qxp]

You seem to want to use this as a replacement power pack+case for a tablet that isn't working...given that you used an SBC in the last thread. Why in Hades do you have the board, a monitor, and a battery pack, but no case? If you've got all of this together, and intend to do what it seems like you are asking, please consider that when you burn your house down we are not responsible. Lithium ion batteries need a special charging circuit, not just to have power pumped in like a lead-acid battery. SAFETY FIRST.

You made a lot of valid points, but I think having a bunch of parts working together before you find the box for them is very reasonable. For once, you might discover you want more parts than you thought when you started

There are exceptions of course. For example, one big engineering project that designed a box before the parts was the first stealth airplane. But then the box was the main feature

 

[DirtyDingusMcgee]

Don't get me wrong. It's a neat little project, and it shows some good thinking.

I don't seek to discourage, but it's a long way to the top if you wanna rock n roll.

 

[S-TKUC]

Don't get me wrong. It's a neat little project, and it shows some good thinking.

I don't seek to discourage, but it's a long way to the top if you wanna rock n roll.

That's right!

I was intrigued by the OP's attempt to build a tablet PC with the hardware configuration of his choice, so I posted on the thread.


If I had the same goal as the OP and went all in to build my own tablet PC, here's what I'd do...

- Make a custom battery pack.
Purchase a battery spot welder and some battery metal sheets, and figure out the maximum battery count and placement that would fit into the shape I want and output the voltage I want.

- Select or make a BMS circuit.
As I mentioned before, I have experience building and testing BMS circuits for much larger batteries, so this shouldn't be a problem.

- Establish communication between the BMS and RockPro64.
Monitor information such as remaining battery level, voltage and current, and allow the system to recognize the battery.

- Select a display.
RockPro64 has video output terminals suitable for tablet use, such as LVDS, eDP, and MIPI-DSI.
EDIT: LVDS wasn't there, but eDP and MIPI-DSI are good enough options.
Carefully select a display that is thinner and easier to incorporate.

- Changed power input method.
The RockPro64 can now be powered via a space-saving JST XH 4Pin connector (CON15) rather than the typical DC barrel jack.


There are also various other factors such as airflow design.
If I were to take the easy route, I would choose a different SBC instead of the RockPro64, but that's a matter of personal preference.

 

[ziomario]

So i've been watching this project roll along. What is the end goal of this? Is all of it just so you can have one, single FreeBSD tablet? Is this something you are expecting to be commercially viable? Do you expect to sell a lot of them? 10? 100? 1000? 10,000? Going from prototype to a production unit is going to cost a lot more than I think you think it will.

The end goal is to build only one FreeBSD tablet,the one for me.

You are designing a cheap tablet in a market that is flooded with cheap tablets...

On the market there aren't tablet based on FreeBSD.

FreeBSD is not going to be much of a selling point.

A FreeBSD tablet is for sure addressed to those who knows the OS...But I think that offering the adequate information, some other people could also accept the challenge of purchasing a new product...anyway,ok,it's not for every need. Infact it shouldn't be. I create it for me,for pleasure and for learning.

Does it run WhatsApp, youtube, music, and porn for less than $100-200? That is the concern of the cheap tablet market.

Who said that it should be a cheap tablet ? This kind of tablet is not for everyone,so it should not be used only for the basic tasks that you mentioned.
It's a professional tablet addressed for the experienced or semi-experienced FreeBSD or maybe,Linux users. For sure not for the Windows point and click users. So,if the use is intended for professional users,the price will be calibrated for a professional use.

Not good ideas either. How much money can you spend making and mailing prototypes back and forth ? More than likely the first ones will not have parts fit, or the holes won't align, or so on.

I don't understand this point. How can it be possible that the holes I make with the rasp and which are therefore correct at the root, they will become wrong just when the CAD modeller takes the same object to make a digital model of it ?

You'll spend more time using the rasp than it takes to learn CAD.

It seems to me that your evaluation is calibrated on your skills,not on mine. Anyway,I don't exclude to try.

1) Polystyrene is almost exclusively a foam. Styrene is a solid. People don't print in either because they are brittle.

I found a lot of websites of companies that tells to be able to print every kind of box using the polystyrene. I've talked with them. The words used on their websites are pretty clear. I asked to a polystyrene manufacturer to create a box for me. He replied that he can't because he does not have the right mold. Then,I asked : "but on your website you say to be able to create any kind of box of any size and decoration. He replied : "yes,but there are
exceptions". So,your website is lying. Ot is it only marketing ? Marketing = telling lies ?

Models cut out from them, usually via hot knife, are not dimensionally stable or durable.

The case of polystirene should not last forever. Its purpose is to convey the most accurate information possible to the CAD modeller. He will produce the STL file for me having the polystirene enclosure under his eyes and then he will send me the STL file that I will send to the 3D printer company.
I think this way is faster than learning CAD. I suspect that you don't have a clear idea about how much effort and time it is needed to learn CAD from zero. I'm sure that you don't have idea simply because you are already able to use CAD. So,you assume,implicitly,that if you have been able to learn it / use it,everyone can do the same.

5) I have zero 3d modelling skills. I also have no hardware, and no desire to learn anything to make this happen.

Very false and offending. You simply don't tolerate those people who try to get to the result by finding some /anyway,licit/ workaround to work less. Why? Because you are a professionist,you work well,maybe even too much. You are too fussy. So fussy to don't understand that my project is a little project. For sure I want to be satisfied by the result,but I have a limited budget and no commercial ambition. And anyway,you can't blame if I have no experience,I'm here to learn,for sure I'm not here to be discouraged by someone,even if he is experienced. I see a lot of experienced users here. So,every suggestion I see is useful but not essential until I will understand what will be the next step to do for my project.

If I were to take the easy route, I would choose a different SBC instead of the RockPro64, but that's a matter of personal preference.

Which kind of SBC do you want to choose ? Do you know that it should offers a good compatibility with FreeBSD ? And FreeBSD has a limited SBC ARM boards where it works decently.

 

[S-TKUC]

Which kind of SBC do you want to choose ? Do you know that it should offers a good compatibility with FreeBSD ? And FreeBSD has a limited SBC ARM boards where it works decently.

To begin with, I'm an amateur when it comes to OSes, so I wouldn't select an SBC assuming it would be FreeBSD.
For me, the OS is a means to an end, not an end in itself.

For example, if I want to use a camera for real-time image processing, I would choose an SBC such as Jetson or KakiPi.
EDIT: If I were to target a low profile system with the Jetson, I would choose the Hadron-DM for the carrier board.

If I want to use it as a x86 Windows PC, I would choose a LattePanda.

If I just want to watch videos...FireTV Stick...no, I know, I should buy a FireHD.

 

[ziomario]

Not good ideas either. How much money can you spend making and mailing prototypes back and forth ? More than likely the first ones will not have parts fit, or the holes won't align, or so on.

I don't understand this point. How can it be possible that the holes I make with the rasp and which are therefore correct at the root, they will become wrong just when the CAD modeller takes the same object to make a digital model of it ?

 

[S-TKUC]

When processing brittle, easily deformed materials such as polystyrene, there are various possibilities for deformation and damage.

During packaging, transportation, unpacking, while the 3D designer is measuring, etc.
Temperature, vibration, humidity, improper packaging, etc.

Unless the 3D designer and the client can completely agree on the uncertain factors, the job cannot be completed.


For example, let's say I receive a polystyrene case.
I've been following the OP's thread, so I already know the rough dimensions of the parts.

If the display opening of the case I receive measures 1mm smaller than the display's visible size, I don't know which to believe.
If the case is slightly curved, I don't know if that's by intentional design or the result of deformation.

This type of inquiry will be made for almost everything.

It's a waste of time, and the time spent making inquiries is of course work time, so a fee will be charged.

From the OP's perspective, it will feel like they're trying to extract fees by finding fault.
From the party accepting the request's perspective, they just want to prevent the worst-case scenario of the case being unusable after delivery.


If you want to do it without ever touching CAD and without incurring any costs...
Instead of sending the polystyrene case to the 3D designer, you would be better off sending all the parts to be assembled.

The 3D designer can then create an accurate 3D model of the part based on a new, off-the-shelf product and design the case reliably.
It would be cheaper to buy another set of all the parts than it would be to pay the risk management costs for a case that may be deformed.

*Of course, it would be better for the OP to cover the costs for the 3D designer to purchase all the parts, rather than the OP sending all the parts to the 3D designer.

When outsourcing this kind of work, you shouldn't attempt to do it in a field in which you are not an expert.
In other words, it is fundamentally wrong to create a mockup of a case and ask for a digital reproduction of it when you are not an expert in case design.

Here's what you should be asking for:

"Please make a case to fit this component in."
"This part is a display, so please make an opening for the display area."
"Please fit this part into the case with this component connected to this connector on this part."​

"Please make this connector of this part accessible by providing an opening on the outside of the case."​

​

The following requests will double some of the work required for the initial review, but they will be less expensive than having to make multiple retakes without any prior notice.

"Please consider two examples of part placement that meet the above conditions, one that prioritizes thinness and one that prioritizes screen occupancy (thin bezels), and confirm them before proceeding with detailed design."​

​

If you make the following requests, the design cost will skyrocket, but let's take an example...

"This part of this component generates a maximum of x W of heat, but please design the airflow if necessary to keep it below y °C. The voltage that can be used for the forced air cooling fan is z V and can be supplied from this connector."​

If you have the thermal design capability to make changes to these requests, you may be able to get away with just the basic costs.

"This area of this component will be the intake. Prioritize the opening on the circular outer perimeter and provide a fan grill with an opening rate of 85% or more."
"Air at a maximum temperature of n°C will come out of this side of this heat sink. Use this part of the case to exhaust it to the outside. The opening area of the exhaust port must be at least x mm^2, the length of the flow path must be no more than y mm, and the cross-sectional area of the flow path must be at least z mm^2."​

​

This type of request method allows you to make a request without intervening in any specialized fields regarding the case design.



If you are really, really, really determined to send a polystyrene casing to a 3D designer, no matter what...

The following agreements will be necessary.
These are just some examples.
- All curves or irregularities of less than ±0.5mm per 30mm length will be considered to be straight lines.
- The thickness of the casing must be 0.8mm or more at its thinnest point.
- The gap between the casing and the lid should be 0.4 to 0.8mm.

​

There are many other agreements, such as the above.
The OP himself must fully understand what these agreements mean and what they bring about.

I think that if you can fully understand these agreements, you will be able to create 3D CAD design yourself without needing someone to do it for you.

 

[qxp]

I don't understand this point. How can it be possible that the holes I make with the rasp and which are therefore correct at the root, they will become wrong just when the CAD modeller takes the same object to make a digital model of it ?

Simple. Take a model, take a pair of calipers, measure distance between holes, get it wrong by 0.5mm due to odd position of holes or calipers, now it does not fit. Even if you design a 3d-printed model with exactly correct hole spacings, it might shrink after printing and the holes will be wrong. It is an iterative process - make it once, check whether it fits, make adjustment.

In the industry for one-off instruments you would design the box and have it cut out of metal, which does not shrink appreciably as far as hole spacings are concerned. You still have to worry about clearance - if you have a connector to pass through and make the dimensions exact, it will not fit. Need to add some tolerances.

And guess what - we are talking about tolerances, so you learned some CAD already.