[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/diy/ - Do It Yourself

View post   
File: 40 KB, 550x458, SVC6S8ETR4334R.jpg [View same] [iqdb] [saucenao] [google]
1987456 No.1987456 [Reply] [Original]

a fully automated cnc shop, with a web application where a customer uploads his cad file, receives an quote in seconds, and can decide to buy the part or not

if he buys, the system checks for the payment, once is done, proceeds to build the part, pack it and sets aside for shipping

say i get a system like this up and running, with minimal maintenance, is it even worth the hassle? is there even demand for it or people just like to talk to their supplier most of the time?

>> No.1987458

>>1987456
>with minimal maintenance
lol

>> No.1987479

>>1987456
there's an awful lot left to automate there anon

>> No.1987483

>>1987479
yes, i didn't get in the details, but you get the idea

>> No.1987485

>>1987456
idk about machining parts but my barrier to entry in pcb manufacture was when they made websites with instant quote tools that auto analysed your design or cost was based on size etc instead of having to send it away, wait for someone to look at it, actually speaking to people? fuck that. i don't want to have to ever speak to another human ever again.

>> No.1987491

>>1987456
Absolutely won't work. Unless your automated system can troubleshoot and rewrite code along with inspect and finish completed parts like the people you're replacing, it's just going to break tools and mess up parts over and over.

>> No.1987492

is there yet a CAM program that 100% auto generates toolpaths reliably? I know that exists for 3d printers which are very much simpler than subtractive manufacturing. Last I was working at a place with CAM the auto toolpath stuff was an expensive addon that was likely to crash the very expensive machine.

>> No.1987505

>>1987485

you mean that you had to talk to someone before getting your pcb to build? what is that you work with?

>>1987491
>>1987492

i guess that would be one of the big problems to solve, but again if you are not solving an existing problem, whats the point

>> No.1987511

>>1987492
What I've heard is that the easier CAMs can do some stuff but get in trouble with 4/5 axis, while the more expensive/advanced ones require years of experience to be used.
So probably not.

>> No.1987544

>>1987505
I think the money would be in selling the solution to the problem, rather than running a shop. No shortage of machine shops that would love to cut out their expensive human programmers.

>> No.1987568

>>1987456
>proceeds to build the part, pack it and sets aside for shipping
What kind of stock are going to have on hand? You're going to need most common grades and sizes. You will need some kind of saw to stock. You probably wont be able to cut very large pieces either. The most automated you can get is probably a KASTOsort (https://www.youtube.com/watch?v=vkij1a_Qgj4)) but a worker still needs to load the bar feeder tables.

t.sawfag

>> No.1987593

>>1987568
i was thinking more of a prototype or small orders oriented system, that looks very expensive, definitely out of my league

also
>2016

i wonder what they have now? how does a regular shop even competes with someone using that kind of hardware?

>> No.1987604

>>1987593
>2016

The shitty resolution and 4:3 aspect ratio tell me it is older than that

>> No.1987612

>>1987456
You’re describing a hilariously difficult chain of AI and robotics technologies. If I were a supergenius roboticist, I would use .01% of my cognitive capacity to brainstorm a more lucrative application for my skills than a fucking automated CNC shop.

>> No.1987673

>>1987456
Wont work. Unless you have time, money, more time, and an engineer.

You are going to have all different cnc machines (along with different robots) to make parts depending what the customer wants. Plus having a system that will route the job to the right machine for said part to be made.

You also make sure that the part will not fuck your machines up. You are going to need someone to make sure that everything looks good before running the part.

Maintenance wise, you will have the normal maintenance that is going to be done along with changing inserts every so often.

There is no 100% fully automated system without problems.

I use to run two cnc lathes with a robot. There is issues and you have to babysit the machines.

>> No.1987677

>>1987456
Allow me to demolish your dreams. You are going to spend well north of $1,000,000 on equipment alone. You (or a hired machinist) will still need to be on site to make sure the machines don't crash and to QC all parts.
>with minimal maintenance
Tools will break. Machines will crash. Tolerances will be lost. Especially when dealing only with small run custom parts. You are going to spend tens of thousands of dollars per year on machine maintenance alone.

>> No.1987679

>>1987593
>that looks very expensive, definitely out of my league
If you are saying that then you obviously cannot afford to start a CNC shop

>> No.1987689

>>1987456
>fully automated
Doesn't exist. Everyone seems to think robots are doing everything in manufacturing today. I blame china taking away most of the west's manufacturing. Nobody gets exposed to it. Are you going to have an automated steel rack that will feed into an automated band saw to cut off blanks? Are you going to have a computer program come up with fixtures to hold odd parts? Are you going to have a computer program decide absolutely all tool paths and just trust that nothing is in the way? How about when the part is done? how will the computer know how to handle the part, where to put it? Do you trust the computer's speeds and feeds are correct and will never break a tool? You're getting into image recognition / AI territory to intelligently handle broken tools, oddly shaped parts, or things just not working out as expected(jams and stuff).

You've seen those completely automated assembly lines with robotic arms welding together car bodies, right? That wasn't setup overnight. It was carefully programmed by probably a team of people to run through the entire programs and tested multiple times in order to get it to run as flawlessly as possible. Even then, shit happens, a human(or a team of humans) need/s to intervene because after all computers are still dumb. All they do is what we tell them and have no agency to decide for themselves. AI is still a fucking meme and you shouldn't take stock in what the "I FUCKING LOVE SCIENCE!" onions boys say.

We're not there yet in terms of technology. If you can make the next big push through in this technological field the last thing you'll want to do is run a machine shop with it, you'll be rolling in so much cash from all the other people that want to lease the technology off of you.

>> No.1987706

>>1987456
If you had one stock metal piece, yes. For example 4" diameter hot rolled 1040 steel. That's it.

You could optimize your software to your machine and the bits you buy and the hardness of your water. Then you could add new stock metal after you prove out the first one.

In this case your profitability is in years of knowledge tweaking your own machine and code to know how to make a part in a few choice alloys.

CONGRATULATIONS, you're a machinist now!

>> No.1987766

Maybe if you like did direct metal laser sintering onto a metal platter that then gets moved into a CNC that cleans up the surfaces but it'd be kinda dumb

>> No.1987769

>>1987766
Those get really expensive fast

>> No.1987771

>>1987769
Yeah like it'd be really dumb and just not feasible (like this thread)

>> No.1987773

>>1987456
OP there are a lot of reasons why this won't work, and most of them aren't even your fault. The kinds of people that want to make small batch prototypes online are also the kind of people who have no idea how to feasibly design a part for manufacturing. I'm sorry bro it is just a waste of time.

However, I don't want to be a complete party pooper. If you really want to make an automated business, I have always wanted to make a machine and website where you can order a custom number of screws/fasteners/nuts in small quantities. You select the screws you want and then order them, and it would cost less than buying a box of 100 or whatever the smallest size is on McMasterCarr. Just a thought for you

>> No.1987774

Even if you fully automate the shop where you don't need s single employee and it requires less then an hour of your time a week. You will still not even be able to buy the metal for less than the Chinese can ship the finished product to your front door.

There is a reason why every screw shop on the continent went out of business or is barely hanging on. The Chinese have such a huge advantage from everything to cheap labor and government subsidies that you will go broke trying to compete.

I know many shops that basically closed up over the years. Just couldn't keep the lights on or doors open anymore. These were businesses that have been around for 50 plus years. The Chinks were willing to work at a loss for years to take their business away.

Thank your government, they allowed Joe6pack to get raped so these large corps could squeeze out a few more pennies in profits by giving China most favored nation status and no tariffs.
Good Luck..

>> No.1987775

>>1987774
Where are these cheap chinese screw shops? I don't mean that in an argumentative way, I want to use it

>> No.1987815

>>1987775
This, fuck fastenal

>> No.1987827

>>1987456
Unless your using 1 material that is soft
Have 1 size bar stock material per machine
Have a bartenders on everything
And have it relatively lose tolerance parts it won't work

>> No.1987830

>>1987456
The only way something like this works is with fuckhuge order quantities, like up in the hundreds of thousands range. Like engine blocks, transmission components etc. For 100 parts or so, the hassle of getting robotic equipment dialed is simply not worth it as it can be extremely finicky and doing it manually is much faster and less frustrating, and takes the same amount of manpower as a robotic loader.

>> No.1987843

>>1987492
Maybe you can outsource that to India?

>> No.1987849

>>1987689
This our vice president of our company is all gunhoe about robots and the retared million+ dollar robot for cnc griding castings still needs someone to manually load and unload the fucking belt to the robot which is always having camera problems and can't pick up anything remotely complex

>> No.1987866

>>1987456
Those robot arms break every time you look away from them, every time you even blink. I hope you like vacuuming metal shavings.

>> No.1987867
File: 33 KB, 334x420, iShiggy bird gryphon i seriously hope you guys dont do this shiggydiggy.jpg [View same] [iqdb] [saucenao] [google]
1987867

>>1987849
>This our vice president of our company is all gunhoe

Do you guys happen to make motsorella, by any chance?

>> No.1987962

>>1987612
>>1987673
>>1987677
>>1987679
>>1987689
>>1987706
>>1987766
>>1987773
>>1987774
>>1987827
>>1987830
>>1987866

thank you all for the input, i'm a control engineer so i do know a lot about of the problems i'd face, i also have a home made cnc, and i want to see how far i can go with this in a small scale at least

i know a system like this would be worth a lot more than what i'd make in a cnc shop, its just that i like that much to see the chips flying

>> No.1987973

I work in a fabrication house and we have all kinds of new expensive shit.
Laser cutters, a self feeding metal punch as big as a house, about 5 CNC machines, welding robots, etc.
The boss will throw money at anything to make production faster or cheaper.

Despite this, fully automating things isn’t quite in reach, each machine needs at least one person on it most of the time.
It’s a nice dream but we ain’t there yet. This shit goes wrong all the time.

>> No.1987980

>>1987867
Gung ho it auto correct
And no

>> No.1987984

>>1987973
Not meaning to shit on your idea tho, OP.
If people like you don’t try, we don’t progress.

>> No.1987988

>>1987962
>, i'm a control engineer so i do know a lot about of the problems i'd face, i also have a home made cnc,

show pics of home made cnc please

>> No.1987989

>>1987456
>order processing
>"automation"
>boxes ready to ship
sounds fucking great OP your are already a gorilionaire!

>> No.1987997

>>1987456

>say i get a system like this up and running, with minimal maintenance, is it even worth the hassle?

A fair question - do you know how much a setup like that would cost?

Or rather, do you know how many tiny parts you'd have to sell to the civilian market to even get your money back?

>> No.1988015

>>1987492
In 3d printing you turn your 3d problem into several 2d problems (i.e. layers).
You go step-by-step along a fixed third axis while every layer is 2d.

That's why it's so easy to automate it, it's comparable to machining a simple relief, or routing a cutout.

Anything more complex than that, especially if you have multi-axis, and you need to switch to different tools for different types of cuts, full automation becomes impossible.

>> No.1988030

>>1987989
english is not my first language, i know i sounded like a faggot, but i'm a engineer working at the R&D of a big automatic recloser manufacturer, i'm not stupid

>>1987997
to be honest i have not done the full business plan yet, i'm more into the technical area

but i have around $450k in my portifolio, that i'd be ready to liquidated and invest to make progress

>>1987988
i'll take some once i'm home

>> No.1988040

>>1987771
theres a machine that does just that, no moving either
look it up

>> No.1988075

>>1987456

so you're saying you want to nix the mouthbreather operators and replace them with a robot arm to load parts and hit the green button?

based

>> No.1988303

>>1987456
>is there even demand for it or people just like to talk to their supplier most of the time?
I'm doubting you're an engineer who actually makes things. Or if so you just graduated.

If a supplier/distributor exists for a part it's definitely going to be cheaper than trying to source it yourself if you want bulk sales. This is because the supplier sets up the entire "vertical integration" (pic related) supply chain from start to finish including design verification, raw stock acquisition, manufacturing, quality control, and shipping. Or they pay someone else to do any or all of these steps. They of course try to optimize this as much as possible to lower costs. You're expecting a customer to copy and create the design themselves and then get your robots to do the rest (you'll have to source the raw material yourself unless you think robots can shop around for raw materials). That's never going to be cheaper than what the supplier pays to make the part + their profit after all their efficiency optimizations, unless you are a massive company that is going to actually do all those supply chain steps and keep the profits yourself.

So this leaves a customer demand for single or low production parts. Therefore your automation idea would compete with CNC shops that get paid to make 1 or a small batch of parts from CAD files. As others have said your automation software simply doesn't exist ye (at least I've never seen it, maybe a company has made it but it would be $$$). Toolpaths are really freaking complicated to make and it takes months to years to get proficient enough to do it correctly and at a production pace. So the automated tool path software would make parts would almost definitely be shitty compared to a machinist with many moths to years of experience. You'd have to lower your price considerably to compete with them, ie worse tolerances for lower prices.

So basically I'm sorry to say but this plan wouldn't work.

>> No.1988310
File: 45 KB, 718x864, Vertical-Integration-Info-Graphic[1].png [View same] [iqdb] [saucenao] [google]
1988310

>>1988303
forgot pic

Also forgot to say 3D metal printing is something you might want to look into. It's certainly still not fully automated but it doesn't require you to use "tool paths" so if higher automation is your goal it's the best route so far.

And of course if something is possible others are already doing it. Lots of people have already made 3D online metal printing services. They're not cheap. I read a university had one of these 3D printers and made a mold for a manufacturer and it cost $40,000, the mold only weighted about 15 pounds and was smaller than a lunchbox lol.

>> No.1988320

>>1987456
>customer uploads his cad file
>receives an quote in seconds
>proceeds to build the part
>with minimal maintenance

>> No.1988427

>>1987492
3d printing CAM appears simple because 99.9% of people only dip their feet in it and are good with 0.5mm part precision
forget that in a machine shop

>> No.1989417 [DELETED] 

This is already a thing. There are companies that automate quotes, and have many cncs with large standardized tool changers that can machine 3d parts with little human intervention. Turn around in just a couple days.

>> No.1989676

>>1987456
>customer uploads his cad file, receives an quote in seconds
fuck the machine shop
make billions building these quote machines

>> No.1989754

>>1987456
Nobody signs up for you to just make them a part without qualifying and inspecting the part PPAP. Its a long process based.

Quoting random parts is hard if your already in production. How much time is available on the CNC. Is this product worth making vs what is currently being made. What is the volume of sales. If your making 50k parts a year thats good profit depending on cycle times.

>> No.1989938

>>1987456
The setup you're talking about would cost $2M when its all said and done and you would still need to supervise it.
If you want to get into fabricating, buy a good used mill and go bid for contracts. It really is that easy.

>> No.1990007

>>1987456
>minimal maintenance
bro have you ever machines

>> No.1990009

>>1987456
>a fully automated
one software bug or one bad update and you have to revert to a manual system. it wont work.

>> No.1990022

Run a successful machine shop first and you will have your answer. If you have not surmounted the lesser complexity of that task you won't even know what questions to ask.

>> No.1990080

>>1987456
I can tell you have never worked in manufacturing or software development. Just for the quoting and ordering system, companies would pay you billions if you could get it right.

>> No.1991075

>>1987456
Companies do this now. Protolabs for example.

>> No.1991198
File: 122 KB, 333x500, 1600967003683.png [View same] [iqdb] [saucenao] [google]
1991198

>>1987456
you'll still need manpower for
>machine supervision and gcode tailoring
>quality control and part inspection
>tool changing
>stock placement and preparation
>part postprocessing (deburring and finishing)
>packing and sending the part
you'd need a massive upfront cost to automate even half of this with assembly lines and control arms. doing this on a small scale would make ROI unattractive for a home gamer, would probably be more feasible to do the same system but with 3D printing (machines like the CR-30 have belt platforms now)
t. mechanical engineer with specialty in manufacturing

>> No.1991203

>>1991198
plus, it sounds like most of your work and effort would be going into the frontend and cohesion of the operation - setting up the website, making software that can automagically take a 3D file submitted and generate reliable tool paths no matter the geometry of the object, calculate the running costs based off the volume of material used and wasted, amount of machine power used, leeway for any problems like tools breaking and ruining the part, and all this within minutes - sounds like you need big bux, not a one man band operation sadly