you seemed to be in your last post...Quote:
Originally Posted by BanduraMaker
and i was originally. eep!
Printable View
you seemed to be in your last post...Quote:
Originally Posted by BanduraMaker
and i was originally. eep!
The point I was trying to make is that the strength of the platform is in the software, not necessarily the hardware. Hardware shortfalls, even a lack of adequate I/O, are more easily made up for than shortcomings in Software.Quote:
Originally Posted by ihavenofish
Taking something like EMC2 as a start and moving it to a small platform sounds like a really good way leverage a lot of prior experience in a new HW/SW combo unit.
oh, i see what you mean. having a basic app that did very little would indeed make it kinda pointless for most people.Quote:
Originally Posted by BanduraMaker
what i plan on doing soon is dismantling an acer netbook and turning into an aluminium control panel. its a full windows/linux pc with lcd, keyboard, and all the trimmings for $199. i came up with that after i realised the rpi would take a very long time before it was ready for my use. the drawback to the netbook is that its got usb, ethernet, and not much else for io.
As they say "misery loves company", so I quite understand that someone who has shelled out (say) $2000 for an all-singing all-dancing solution does not then want to hear that he could have had a better system for $200. Still... it's amusing watching that person desperately clutch at every little straw in the hope that it might one day prove how the $200 thing is really a POS. And struggle to appear open minded as he does it! :-)Quote:
Originally Posted by BanduraMaker
Which is to say, why couch your message in terms of so many negatives? Where did all those negative assumptions come from? As an embedded programmer I can assure you that a 700MHz ARM with 128MB SDRAM and an 8GB SDHC sdcard (acting as the hard disk) is fully capable of implementing a program of this type to the highest possible standards: in fact it's vast overkill. Plus it will do so in a far smaller "footprint" than a PC. The only valid objection was a possible shortage of PIOs - which if true would be easily addressed using a daughterboard containing an SPI-controlled PIO shift register, or similar.
Well, if you have the tool under control then you have it under control - you can't outperform perfection. I would expect step size to be limited by the motors, not the ARM. The display thing is an example of straw clutching. The only reason a display was even being discussed was because I, a non-CNC person, didn't understand that a display would be needed (and am still not convinced) - so I didn't want it taking up space in my workshop. However the issue was never that the RPi had any trouble whatever in handling a display - it has an HDMI output and can therefore drive a modern FullHD LCD panel [ame=http://www.youtube.com/watch?v=TgR74Kp6Ws4]Demo[/ame]. As it happens I recently implemented an HDMI out on an embedded project and the code was trivial: I2C to set up the HDMI controller timing regs (i.e. select a resolution), then set up a DMA channel to keep it fed with video data. I expect it would be the same here (obviously we aren't writing an MPEG decoder, so the clip is just here to scotch a potential urban myth about RPi having an inferior display capability).Quote:
Originally Posted by BanduraMaker
youre still completely out to lunch on this topic unfortunately.
the $200 solution is a regular PC, and $200 is pretty excessive really, you can often get old oens for free these days.
the rpi is a $25 solution. and yes, of course you need a display, how else woul you be able to interact with the device, jogging, knowing what position youre at, zeroing the machine, setitng up tool tables, getting graphical feedback, errors, editing gcode, tuning the machine etc.
even the cnc machines from the 70's had displays :)
you still seem to be stuck on the rpi being some middle man between the pc and the drives. the rpi IS the pc, otehrwise its got no reason to exist in the chain at all since it only does LESS than the $200 PC.
I see absolutely no point in arguing the merits of a machine like this. It's a good idea, and not just because it was mine. If it does not appeal to you, use a normal PC. Nobody is going to stop you. What no one seems to mention is that everyone has different needs and wants. We have a local club with half a dozen maker bots, and all the owners own RPi's, and they would all JUMP at an SD card with the proper software.
If you don't think its a good idea then don't contribute. But then don't try to convince everyone else it wont work. Every single truly good idea I have ever had came with several people trying to convince me it would not work, but still it did.
Given the data this project is certainly possible. If you don't think it is, thats fine just don't try to shoot everyone else down.
As far as creating your own linux distro and stripping down package to a minimum...this is not that hard. I have several friends with their own compiles. DSL linux was somewhere around 40mb last time I checked. These distros are already out there even, and may just need a bit more tweaking to be ready for EMC2.
Alex
Mpack, I think I see the problem now. I was pretty certain that you couldn't be as dense as you're coming across, it's simply that you're ignorant of the issues at hand. So instead of responding with snide remarks I'll give you a little education.Quote:
Originally Posted by mpack
Controllers like Mach 3 and EMC2 have thousands of man hours into them because there's no such thing as perfect control of the tool, at least if you live in the real world. CNC controllers make compromises in the tool path, or not, for various reasons which wouldn't be obvious if you were completely ignorant of how CNC machines work. e.g. We who work in wood prefer to use "Constant Velocity" mode to keep the bit from burning the wood. This mode looks ahead into the code to look for changes in directions and will make adjustments to the trajectory to try to maintain velocity.
Implementing something that will perform coordinated motion from point A,B to point C,D to point E,F is trivial but, implementing something with CV mode and acceleration curves (and there's different types with different performance) and backlash compensation and screw mapping and, and, and, is another story.
This is very true.Quote:
Originally Posted by crane550
Right now, you can buy an industrial controller starting at about $1,250 + Computer with WINCNC
Flashcut is apparently up there in terms of quality and performance for about the same price for software and signal generator. AFAIK, it doesn't have the customization options that Mach 3 Does though.
Other industrial applications go up in price from there.
Next we have Mach3 which is close to industrial controller level but not quite there yet and has bugs and performance issues (smoothness of cut, CV acceleration bug possible, trapezoidal acceleration vs. S-curves). It's a great value and has a ton of support for $150.
We also have EMC2 (which I have no experience with) but it apparently is neck in neck with Mach 3 as well at zero cost.
Finally, there are Arduino based controllers which I know absolutely nothing about. Judging from the websites I've looked at though, they only have pretty basic functionality but it's enough to control something like a makerbot without much difficulty. There's a kit I've found for $370 that includes stepper drivers on the makerbot site. It comes in at <$100 without the driver.
The question I think becomes where in the continuum do you want this project to lie?
Personally, having an embedded type controller with all the features of Mach (minus the bugs) with a high speed clock for smooth control and a display (touch screen?) for <$400 would be super cool. Especially if it could be customizable with macros etc.
A <$100 very basic controller isn't very interesting to me but hey, there's probably many who would be into it!
sorry, hes just brought out my need to correct people who are wrong, haha :PQuote:
Originally Posted by crane550
i fully agree with you that the rpi is a good idea. :)
i just wish i HAD one now!
A lot of arguing going on here...
I think it would be great to see the RPI run LinuxCNC, although even if the LinuxCNC team decided to get it running on there, the support would probably be not much. How much of their limited resources are they going to put into a platform that is used by 0.01% of the userbase? The only hope would be that one of the programmers there goes crazy over RPI and decides it's his life mission to get it working on it. RTAI does have some work done on it if you search for it, here's from the website:
ARM (StrongARM; ARM7: clps711x-family, Cirrus Logic EP7xxx, CS89712, PXA25x)
I'm not sure if RPI falls into that, I don't think it does, but i'm not expert on arm. Also I think some versions of LinuxCNC used to use rtlinux, or i'm not sure what it's called, but one of the other realtime apis, but i'm pretty sure they use RTAI exclusivily now.
If cost is your primary purpose for attempting to use RPI, you may look at intel atom motherboards. I have used over 5 of them, and they all work fine, and they come with parallel ports (well some of them). I recently got 2 motherboards for $100 off ebay, new, but a reliable source like newegg has them for $75, you only need to add $10 worth of memory. They are not very power hungry either, although they will use more then RPI i'd imagine. So in my case, $50 motherboard, with $10 ram, that's $60 vs $35 for the RPI, and LinuxCNC works out of the box. I use a $15 USB 3.0 flash drive (not that the board even has usb 3.0, but the 3.0 flash drives are usually a bit faster), plus $15 power supply, and I made my own case out of 1/2" pvc sheet, $10. So i'm at $100 for a PC that runs some of my machines. I actually only made my own case for one, i had a bunch of old cases I threw the others into.
Not that my intel atom has anything to do with RPI, i just thought i'd mention there are low cost alternatives that are out today. Now if your main reason for wanting to use it is size, then the RPI destroys the mini-itx intel atom boards.
I do plan to have a RPI connect to a arduino/grbl setup.
I'm not trying to pick a fight, but...Quote:
Originally Posted by KuchateK
Why is everyone bashing the poor old parallel port? In what way is it unreliable? It may be outdated technology, but a PC with a parallel port has more GPIOs than the RPi. There are thousands of happy Mach and EMC users using a parallel port. The only issues I read about is incompatibility of some parallel port boards. You can buy a brand new Atom motherboard for under $80 which has a parallel port.
Yes, the $ add up when you need to add memory, disk, power and case, but if $ is an issue, find an old PC.
Has anyone considered the Beagle Bone?
Similar power, 700MHz ARM. More expensive, $89, but lots of IOs.
BeagleBoard.org - bone
i dont see any support for a monitor. limits its uses a fair bit.Quote:
Originally Posted by ftkalcevic
BanduraMaker Stated: ""Conceptually, a stand alone replacement for Mach 3 and/or EMC2 at a cheaper price is truly an appealing thing. However, if it's a step backwards in terms of features and possibly performance, much of its appeal is lost, at least for users such as myself.
I'm a Mach 3 user and while it leaves many things to desire, it does have a very rich feature set and is quite flexible.
Consider that a SmoothStepper is about $150, A 1GHz laptop is about $100 and Mach 3 is about $150. If you can out perform Mach 3 in terms of outright performance and feature set (including display) for <$400, you've got something that would be of interest to me.
If on the other hand, you make something for $100 that has a crippled and poorly implemented G-code interpreter, lack of flexibility and a lack of features (e.g. spindle control, macros etc.) then you may have something of interest to some but certainly not to advanced users.""
So far Andy's [BanduraMaker] comment is the only one that is really appealing as I gotta pay the rent and eat.
Plus considering the fixed investment in equipment, computers, software, plus the time investment in learning the various software packages, and getting them all to play nice, and developing a workflow to turn out product, these investments too make one, especially one with an economic/financial orientation, very reluctant to hop on a whole new direction for the CNC-world thingies.
I like concepts, especially if it shows up with lots of wires and is in a black box with blinking lights.
However, if one stands back and takes a look from a wider perspective, the enormously differing opinions [some with less than civil tone and content] over hardware, interfaces, software [especially all the CNC-related specialty variety], middleware, operating systems, and on and on, really leads one to the conclusion that the RPI idea is not yet a prime time player, and that it will take a long long while before its even in the ball park.
Not that it, or something like it, won't one day be the way to control a CNC machine, its just with no central tendency on implementation for CNC purposes, its not going to be any time soon.
I/O is easier to get than monitor connection. CNC without monitor has no sense today. Raspberry PI is $40 shipped.Quote:
Originally Posted by ftkalcevic
I don't think I did.Quote:
Originally Posted by BanduraMaker
I agree.Quote:
Originally Posted by BanduraMaker
I think that RPI solution would be equal to Mach/EMC for a lot less.Quote:
Originally Posted by BanduraMaker
Above price is not so bad if you're talking about few thousand dollar machine. But for me this is half the cost of my desktop router. Not very appealing. I can get much more needed digital oscilloscope for that.
We already have that. GRBL is such solution.Quote:
Originally Posted by BanduraMaker
I think this new platform will allow creation of something that can match existing PC based solutions.
I think even more important than this is popularity. There is a lot of great powerful development boards. But GRBL was made for Arduino, possibly one of the worst, but the most popular.Quote:
Originally Posted by BanduraMaker
Speaking about I/O... A lot things CNC controller does is not time sensitive. It can be served with one of those serial interfaces through external boards. We are using BOBs for LPT, so it won't be that different.
I think someone will go all the way to FPGA. They are pretty cheap and common these days. We can have small and cheap external board with ton of I/O and perfect timings.
Both Mach and EMC2 code is a fork of EMC project started by NIST. Yes, there are ton of work in those projects. One of them is open source. So do you really think we want to start from scratch? Because I think we will take as much as possible from EMC2 and just write whatever is necessary to use that on new hardware.Quote:
Originally Posted by BanduraMaker
As I said it is already implemented and working. Job done.Quote:
Originally Posted by BanduraMaker
Hundred here, hundred there...Quote:
Originally Posted by BanduraMaker
My CNC budget is $1000. Mechanics and electronics. I'm not going to buy Mach. I'm not going to buy SmoothStepper or a computer for it. It has to be cheap.
Cheap is what makes it popular and then it will get better. Like RepRap. That project is great example. RepRap didn't became popular because it was good but because you could build it for next to nothing from wood, hardware store and printer parts. The very first reprap prints were horrible. When I saw them I laughed. Right now that community achieved results comparable or beyond commercial machines worth 20-30 times more. They have comparable user interface, software, object database, documentation... We need the same thing for CNC.
As I said. They are both continuation of single project. They didn't start from scratch. They took something good and built on top of that. We are not going to abandon all that work but continue on top of that as well.Quote:
Originally Posted by BanduraMaker
Starting from scratch is simply stupid. That is not why Richard Stallman started Free Software movement.
I think you are underestimating the interest in that platform.Quote:
Originally Posted by klick0
They sold 10000 of those boards in minutes. They said they have registrations for hundreds of thousands more. I've registered and in reality I'm planning to order a lot of those to replace several atom based servers and for projects where Arduino is a little to small and atom is an overkill.
Remember that this product is targeted first at programmers and schools and then at hackers and makers. Basically platform to create on top of it. Even before release it was running stepper/leds/adc/dac as a demo! CNC controllers and other mechanical devices will grow on this a lot faster than you and me can imagine.
Arduino is just 300 000 boards strong. It was started in 2005, although first few years it was rather quiet compared to RPI. Right now Arduino has its own CNC controller and thousands of RepRaps running on it.
Raspberry PI seems to be a lot more popular.
RTAI is multi platform API. It runs on x86, PowerPC, MIPS and ARM7. It will work on ARM11 chip in Raspberry PI very quickly.Quote:
Originally Posted by klick0
Quote:
Originally Posted by klick0
As of today you are both correct. Yesterday I got perfect PC for CNC for free.Quote:
Originally Posted by ftkalcevic
But... I'm working in IT. I see those old computers thrown away every day. This was last LPT PC where I work. I know few barely working computers with LPT at other locations and then what?
I know exactly that LPT in PC market is long gone. That "brand new" term you used is very relative. Those new motherboards you can buy today are last examples of those old interfaces available on new products.
Soon it LPT will be only on second hand market and specialized industrial products and then it will be horrible. Go to ebay. They want an arm and a leg for old parts. And then you have to pay for the shipping more than RPI is worth. Don't get me started on industrial PCs and controllers.
What then? What are we going to use in 5-10 years to run our CNC machines? SmoothStepper for $150 plus mach license for everyone? It's not going to work. CNC technology needs to drop in price. Every school class/home workshop should have small router. If controller + software + PC costs $400 it will never happen.
It's not as bad as all that.Quote:
Originally Posted by klick0
First off, the CPU is an ARM (at least the CPU of concern to us is an ARM), which is already a very important platform for Linux. True, a set of drivers needs to be written for the RPi peripherals, but that's not really a big job - we don't even need a complete set, we only start with the stuff actually used by EMC.
Of course EMC doesn't run on any old Linux, it only supports a particular few variants with (their own variant?) real time kernel. We would need to port that kernel to ARM, if it hasn't been done already.
And I remind everyone that Linux is not a necessity. An OS is a convenience if you need a fancy shell, fancy filesystem, ability to choose any of several programs to run, run several apps at once etc... if you need none of those things then you can dispense with the OS.
I had a look at the EMC2 sources. It does indeed have what it calls a HAL (hardware abstraction layer - which hopefully also abstracts out the OS) - but whether this is intended to do what I hope it does is too soon to tell.
There is mention of an HDMI daughterboard being available.Quote:
Originally Posted by ihavenofish
at that point though your at a higher price than many regular pc's, with not many advantages.Quote:
Originally Posted by mpack
KuchateK: "If controller + software + PC costs $400 it will never happen."
I am building my second 3-axis CNC router, which is designed to be a 6-axis machine eventually. I have approximately $5000 in the frame, rails, carriages, stepper motors [including spares], wiring, nuts and bolts, labor and materials for building a variety of parts, e-track, miscellaneous electrical and electronic parts.
I have a gecko G540 and 50-volt power supply. I use MACH3, its quirky, but I have got it stable. I use Alibre and Sketchup Pro. I use Cam Bam.
The Dell computer [included monitor, KB and mouse] I got free, and I bought and installed two parallel ports, plus a very good video card. The operating system is Windows 2000.
I design and build furniture.
Given the above facts and circumstances, $400 is pocket change.
Time does have value.
"Cheap" is very expensive.
Though the RPI appears to be very cool, I care not until its been proven, it and all related technologies are accessible, easy, reliable and dependable. And all the aforementioned plays nice with ALL CNC-related software; i.e., CAD, CAM, controllers, and the many add-ons thereto.
Lots of very cool pieces of technology have come and gone.
Sure looks like a lot of folks [and you all are really smart and clever] see the leaves and trees, though are missing the forest.
Yes, YOU have spend a lot of money on your machine....but not every has. This is not meant to control a $50,000 machine. It's made to control high school projects, Repraps, Makerbots, PCB mills, and in general smaller products where price is a factor.Quote:
Originally Posted by zool
If someone was to produce a software repository or a linux distro that would run on the RPi I have no doubt in my mind it would sell like hot-cakes.
$25 for the Rpi
Lets me generous and assign $50 to a breakout board....wer are not looking at a $75 solution. Add a mouse and monitor which most people already have and your good to go.
Now all of the sudden this is looking like a good solution. Most makerbot guys, bench top mills, PCB etchers and such don't usually have a dedicated PC anyways. This is something that makes these devices a lot more portable.
It is rather arrogant to point out the impracticality of a device like this in the industrial market. It's not the point. This is exactly the same as some filthy rich man ridiculing some farmer and his beloved old pickup because it does not have the refinements of a Bentley. Peoples needs are different. Everyone needs to grasp this. For the longest time I had a guy ridiculing my current machine because it didn't have the tolerances of the industrial chip manufactoring equiptment at his company. Sheesh! Really?
I'm getting tired of us arguing the merits of a device like this. I would like to see some of the supporters of this project start moving towards implementation. I got some linux nerds in my family I will talk to over this next weekend and get some educated takes on this on what needs to be done. It's been well demonstrated that this project is possible. Lets find out how to make it work. Nay sayers please start your own thread.
@zool. You write as if I've been trying to persuade you to abandon your beloved CNC gizmos - I have not. If you are happy with your existing setup then fine - that is absolutely no concern of mine!
My question (and I think the question behind this entire thread) was about whether the RPi would make a good CNC computer... and I think it's pretty clear that it could. It is equally true that it isn't a CNC computer right now - and won't be for a while (e.g. first step is for the RPi to become something other than vaporware). Ultimately however I believe that it provides more direct control over I/O and timing (as would be the case with any similar board of course), and it takes up much less space, so it has the potential to become a superior CNC machine. Which does not mean that I'm advocating that you give up solutions you are already happy with.
Cost is not the primary consideration for me, except insofar as it allows for a cheap experiment - and obviously a cheaper but superior final result would always be an attractive proposition.