informal poll... let's stick to dumb controllers

Are you concerned about your reprap getting a little too smart? How much processing power do we really need for our controllers? Maybe this technology is a bit too exotic to hook up to your WiFi. What do you think?

something's would be nice to be able to do via Wi-Fi, like printing from apps. I need to connect my Laptop to the printer each time I print.
in saying that it was my choice not to go for an LCD controller or SD card adapter to be able to print without a computer.
my main driver for parts for my first printer was cost so I wouldn't have taken up Wi-Fi as an option, however if/when I build a second one I would like to be able to monitor it remotely through a smart phone or tablet. can setup a push notification to come through if the printer stops for any reason, like maybe the thermistor fell out that sort of thing.

My new printer project will not need a computer to run.
Mind you , my current printer doesnt need a computer.
So i should say, My new printer doesnt need a computer to Slice or Print, or download files from the web..

It's something that i am working on right now. Along with a few other projects.

My tendency goes towards very simple controllers. Do one thing and to that right.

If you want to have a standalone printer, add a small PC like a RaspberryPi or a BeagleBone. These can slice, too, serve a touch screen display and in emergency cases even connect keyboard and large screen. Having these two computing untis separate also means you don't have to worry about realtime issues on the CPU running the web server.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

personally i'm partial to the 8bit arduino based stuff we are currently using, i'm a bit against the movement towards ethernet and wifi capability as the temptation for people to do a lot of remote monitoring from work becomes too much for some people

---

-=( blog )=- -=( thingiverse )=- -=( 3Dindustries )=- -=( Aluhotend - mostly metal hotend)=--=( Facebook )=-

Plus you need reliable safety measures in place should something like a fire occur when you're not there.

---

greghoge.com

HUGE 3D PRINTER PARTS SALE!!!

The farther you go with complex electronics, firmware software and mechanics, the smaller would be the amount of people that can enjoy it. If as of now with all its "simplicity" a lot of people struggle to understand it and implement it, with complex components you will eliminate the desire and willingness of new generations to just try it.

Now, with all that been said whoever want to do it it's free to do it, just don't make it the standard.

Simple=masses, complex=minorities

Edited 2 time(s). Last edit at 10/21/2014 09:35PM by ggherbaz.

Well thinking about it.
Standard should be a dumb controller to attract alot more users.
And after that, after learning the system. Should be able to upgrade. At the users discretion.
I agree with everyones response, so just roll it all together and you will have your answer.

If by 'dumb' you mean 8-bit arduino based, then I disagree.
I much prefer my Smoothieboard to the Ramps I had before.
Most of my reasons don't really have to do with processing power, but the overall usability of the board. That said I've tried stream printing over ethernet on it, and that feature is not really ready for 'production' use just yet. The software isn't as mature as e.g. Marlin yet but it's getting there.

I would venture to guess it's easier to get started with for a complete noob as well, but I wouldn't know, I had over a years experience when I switched.

/Andreas

No need to stick with 8 bits when designing for just core functions. There are ARM CPUs out there which are of pretty similar complexity as ATmegas, just 32 bit and a bit faster.

32 bit speeds up a lot, even at the same clock frequency, because moving axes is mostly done with 32 bit integers. A 32 bit addition on an ATmega is 8 clock cycles, IIRC, on a 32 bit CPU it's just one clock cycle.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I use dumb controllers, but it's a bit of a headache (and unnecessary) for me to have to put my STLs or gcode onto a sd card and take it down to the basement where my printer lives. I'll probably setup a print server with a little more 'brains' to expedite that whole process.

As for 8 vs 32 bit.. I find it at least a little dumb that some performance features are stifled by the performance of a same-cost 8-bit microcontroller when cortex-m4's are so ubiquitous and inexpensive and easy to program. In some imaginary life when I have more time than I do now, I'll setup some stepper electronics and either port myself or run someone's ported code on my $10 168MHz nucleo.

Smart Fast controllers!!!
If they were still using dumb controllers in ink printers
we would still have dot matrix.

If you didn't have Graphics Processing Units (GPUs) your World at War game would look pretty lame!
Hey a computer the size of a credit card (Raspberry Pi) can do amazing things with out PHD EE!

Smart controllers open up postscript type applications
and doing things at the machine resolution instead of PC resolution.

CNC guys use much bigger G-code set ARCs, Radius circle
Not just 128 line segments on a circle.

Control algorythms to control acceleration, jerk and other motion control

OOGH! ---- got to get out of cave!

I don't know. The controller market seems a little bloated to me. The dot matrix analogy is really apples and oranges. A pixel is a pixel. A micron is a micron.

Let me say this... There is no reason a fully functional 3d printer controller board, with steper drivers, can't be sold for 15-20 dollars. With free shipping. If I had the money and electrical engineering capacity to make that happen, we'd be living in a much different world.

Edited 1 time(s). Last edit at 10/24/2014 11:23PM by Robert_Paulson.

Quote
Robert_Paulson
I don't know. The controller market seems a little bloated to me. The dot matrix analogy is really apples and oranges. A pixel is a pixel. A micron is a micron.

Let me say this... There is no reason a fully functional 3d printer controller board, with steper drivers, can't be sold for 15-20 dollars. With free shipping. If I had the money and electrical engineering capacity to make that happen, we'd be living in a much different world.

Your pricing isn't realistic. The AVR and drivers alone in significant quantity are not quite that. The board will be a couple of bucks, then connectors, PCBA, testing, packaging, getting them from Shenzhen to the US. Once they are here packaging and shipping. A small box and USPS first class parcel is looking at $2-3 plus what ever you pack it in. Then there is the labor to handle it, account for the money and any transaction fees and business overhead. This assumes you already have a contractor that you know and use otherwise there is the overhead of making that arrangement. It's not something you want to do over the Web on Alibaba. To get a US$15 retail price, shipped and make enough to sustain a real business you'd have to get them made, tested, landed in the US then distributed for about $3 each. Ain't gonna happen. About a year ago I looked at doing 1000 Brainwaves and my base cost was about US$30 ea, not including any shipping, test fixtures or test programs or methods. Were I to go over and pound the pavement for a couple/few weeks instead of going through an English speaking middleman I'm sure I could knock the price down but that adds another cost as well. As a comparison my BOM when I built five prototypes was about US$50 ea in just parts from Digikey/Mouser and boards from Seeed.

Who said anything about Shenzhen? Why should they dictate the price of electronics. Give me some h-bridge drivers, some atmega644's and some PCB. I'll find a way to make it work at $15.

Quote
vegasloki

Quote
Robert_Paulson
I don't know. The controller market seems a little bloated to me. The dot matrix analogy is really apples and oranges. A pixel is a pixel. A micron is a micron.

Let me say this... There is no reason a fully functional 3d printer controller board, with steper drivers, can't be sold for 15-20 dollars. With free shipping. If I had the money and electrical engineering capacity to make that happen, we'd be living in a much different world.

Your pricing isn't realistic. The AVR and drivers alone in significant quantity are not quite that. The board will be a couple of bucks, then connectors, PCBA, testing, packaging, getting them from Shenzhen to the US. Once they are here packaging and shipping. A small box and USPS first class parcel is looking at $2-3 plus what ever you pack it in. Then there is the labor to handle it, account for the money and any transaction fees and business overhead. This assumes you already have a contractor that you know and use otherwise there is the overhead of making that arrangement. It's not something you want to do over the Web on Alibaba. To get a US$15 retail price, shipped and make enough to sustain a real business you'd have to get them made, tested, landed in the US then distributed for about $3 each. Ain't gonna happen. About a year ago I looked at doing 1000 Brainwaves and my base cost was about US$30 ea, not including any shipping, test fixtures or test programs or methods. Were I to go over and pound the pavement for a couple/few weeks instead of going through an English speaking middleman I'm sure I could knock the price down but that adds another cost as well. As a comparison my BOM when I built five prototypes was about US$50 ea in just parts from Digikey/Mouser and boards from Seeed.

Robert_paulson = ohioplastics , might give you an idea where the idea came from

---

-=( blog )=- -=( thingiverse )=- -=( 3Dindustries )=- -=( Aluhotend - mostly metal hotend)=--=( Facebook )=-

Quote
Robert_Paulson
Give me some h-bridge drivers, some atmega644's and some PCB. I'll find a way to make it work at $15.

Send me your design and I'll mill you the PCB. I'm sick of these silly excuses, such an effort can be done for less than $100. If you didn't have these $100, you didn't have a printer.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Quote
iquizzle
As for 8 vs 32 bit.. I find it at least a little dumb that some performance features are stifled by the performance of a same-cost 8-bit microcontroller when cortex-m4's are so ubiquitous and inexpensive and easy to program.

RepRap community has well proven that writing code isn't a business model. Even if you do it in your free time, you have a high chance to either get flamed or ignored. All you can achieve is cheapo-chinese to pick up your design and selling it without sharing a penny. Then you, the developer, have all these silly buyers in your neck, making matters even worse.

RepRap asks for it and now they get it. Doing the design is the simple part; it's the software that matters.

Sorry for being so unvanished. That's my experience and there's no sign hinting to a change.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Quote
Taumflug
Send me your design and I'll mill you the PCB. I'm sick of these silly excuses, such an effort can be done for less than $100. If you didn't have these $100, you didn't have a printer.

God damn it man, I'm a machinist, not an engineer.

On this discussion I lean strongly towards the 8 bit camp. When you are working with 8 bit microcontrollers it is both useful and easy to know something of the structure of the microcontroller - ports, direction registers, instructions etc.. This gives anybody working with these a greater connection with the hardware. With the 32 bit processors the software is so abstracted that this sense of connection is removed. Object oriented languages on complex, feature rich microcontrollers are too much like learning to say "abracadabra" in just the right way - it may be easy but it is not much fun.

I am not against the ARM microcontrollers, having programmed Acorn ARM microprocessors back in the late 1980s, but I am now a hobbiest and everything I do on 3D printers is just for fun - including the feeling that I understand everything that is going on in the printer.

Quote
leadinglights
With the 32 bit processors the software is so abstracted that this sense of connection is removed. Object oriented languages on complex, feature rich microcontrollers are too much like learning to say "abracadabra" in just the right way - it may be easy but it is not much fun.

Looking at Teacup firmware, variants for ARM and AVR don't differ much. It's pretty much the same C code. Exception is this processor setup file, which has to exist for ARM. Other than that it's just pin names and such stuff, but the very same algorithms and code.

Yes, I do see how some firmware developers try to abstract things. That's entirely fine of you have processing time and program space to waste. But not good if your task is to create clock-precise code and hunt for these last 5% of performance. Interestingly, some of these higly abstracted firmwares use considerable amounts of assembler code, totally the opposite of abstraction.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY