Yes, you do need a special PC. This means a PC you use for Linux CNC and not the one already on your desk. Again think of the normal ink-jet printer. Do you connect it to a special "printer PC" that is only used to drive inkjet printers? No. In most cases, the inkjet (or laser) printer is just on WiFi and can print from any iPhone of PC that is also on the same WiFi.
When designing something new one must be VERY careful not to be one of Henry Ford's customers. In an interview, someone asked Henry Ford (Who was famous for saying "You can have the Model T in any color you like, so long as you like black.") Mr. Ford said that if he had listened to his customers that would have told him all they wanted was a faster horse. This is a common thing. Users typically get used to what they have and ask for only tiny, incremental changes. But the really good products are disruptive. Look at what Apple did to the cell phone market, now everyone uses an iPhone or iPhone copy. Same with the Mac. now everyone uses a mouse and a windowed graphical interface. Neither of these ideas was something a customer asked for. I think a CNC machine that works like a $180 3D printer would be disruptive. It would dramatically lower the skill level required to get into CNC. Basically you have to get rid of all the parts that cause problems. Look at forum posts to see what those are. That said, disruptive change is really hard to do as it takes a change of mind-set. You have to kick the problem around for a long time before you see a completely different solution. But the big thing to NOT do is to be one of Ford's customers who is asking only for a faster horse. It is all to easy to fall into that trap. What am I working on now? Not this. I'm frustraited in the current fitness tracker ecosystem. Why can't I comine data from my Fitbit heart rate monitor withthe cadance sensor on my bike and my WiFi enebled bathroom scale and get it all plotted on the same system? Te fitness tracker market is too fragmented. There needs to be an easy solution. THis seems off-topic for a CNC forum, but not. The way to re-think fitness tracking is to think what people REALLY need and forget abot incremetal improments to existing products. Step way back ad look. I think peope like me and a few million others just want to zip-tie a tiny sensor device to the crank arm of a bike or ther running shoe or their wrist and then have some kind of "AI coach" consult with then about what needs to be done to reach their goal. People would live to be able to mix brands of sensors and never have to fiddle with phone apps and web sits and bluetooth paring and the 100 other details. Back to CNC. Step back and people just want to make parts, not fiddle with 100 details of how the CNC machine works. That is setting the bar really high. I've done this for other projects at work. The first step is to fly out to where your users live and watch them work. See what they do and how they spend their time. The low-hanging fruit are tasks that take a lot of time but not much brainpower. User productivity goes up when you automate those tasks. On Wed, Feb 12, 2020 at 3:47 AM Les Newell <les.new...@fastmail.co.uk> wrote: > > > You care where the loop are because if they are in your PC you need a > > special PC with a special OS. > > Special PC? Pretty much any PC will work. I guess you could call the OS > special but not nearly as special as an RTOS on an external processor. > If you are really that terrified of Linux, run LCNC headless and write a > Windows or web app to talk to it. The LCNC PC then becomes your external > CPU box. > > > But look at how most 3D printers work. > > They are just like 4 axis milling machines, typically using 4 stepper > > motors and no one needs a special PC or OS to make prints. In all cases > > the designers have pushed all the real-time function out to a $2 chip. > > That $2 chip is absolutely packed to the gills and it does one job - > running a basic 3D printer. It's so full that if you want to add any > features you have to take others out to make room. The whole ethos of > LinuxCNC is to be versatile. Want to run an automatic tool changer? No > problem. Does your machine have an electronically controlled 3 speed > gearbox? Again it's easy enough to add functionality to do that. Servos? > stepper? Any mix of the two? Again no problem. I've used quite a few > different hobby grade controllers and for versatility LCNC blows them > all out of the water. > > If you just want to run a basic desktop router on the cheap get an > Arduino and chuck GRBL on it. You now have an external board running a > motion controller with the PC just feeding it g-code. It's as close to > your 'printer' concept as you're going to get. You can even run it > completely standalone off a SD card. Of course there is no way GRBL can > handle a bigger more complex machine. > > Admittedly these days you can get slightly more expensive 32 bit CPUs > that could run HAL. Moving the real-time stuff to such a CPU is doable > with the investment of enough man hours. Who is going to provide those > man hours? A PC is a convenient, cheap source of a very powerful CPU > with a bunch of storage and useful interfaces. You can run LCNC on older > PCs that are often given away for free. > > > Doing the work on the PC under Linux was a cost > > cutting measure but today it dramatically raises the cost because of the > > difficulty of interfacing a PC to a milling machine. > > Are you talking about the cost of something like a Mesa card? Your $2 > processor doesn't have nearly enough I/Os to do the job on a decent CNC > installation. By the time you get a processor with lots of I/Os, add the > cost of a board and the support components needed you are likely to > easily exceed the cost of a Mesa system. Arduinos for example are dirt > cheap because they are manufactured in vast quantities. CNC is a pretty > niche market and any board specifically designed for CNC use isn't going > to have the low cost you get from quantity. > > > Another example of real-time motion control software "done right" is > > Cleanflight or BetaFlight (both are nearly the same) > > Again, they have one job so the code can be carefully optimized for that > task. CNC machines vary a lot so your control needs to be versatile. If > you want to add some new functionality to your flight controller you're > looking at a lot of work. > > There are massively more drone users than hobby CNC users. Therefore > there are more people willing to donate their time into making > Cleanflight etc as slick as they are. If enough people were to work on > LinuxCNC it could become very slick with a pretty GUI to configure it. > The problem is that there aren't enough people willing to donate the > man-hours needed. Even then the GUI wouldn't be able to handle the more > esoteric setups. Are you willing to dedicate a year's work to the > project? You could get a lot done in that time. > > Les > > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users > -- Chris Albertson Redondo Beach, California _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
