Got the prototype XY slider sorted. Now I need to sort the rest of the gantry.
Also made a prototype standoff for the walls of the enclosure. Using paper/foam project board for the walls - simplest thing for the first iteration. Later I can drill 1/2" (or 12mm) holes in acrylic sheets. (Large holes and plastic mount should preclude stress-cracking.)
To sort the gantry, I need to sort the mount for the hot-end/extruder. To sort the mount, I need to buy a hot-end/extruder. There is a rather nice review of the E3D Aero from @Thomas_Sanladerer (and Prusa is using E3D), so that seems a good start. (Also I might want to try the “Volcano” option later.)
Went to the @MatterHackers site (as they are just down the street), and the first question: 24 volt or 12 volt?
Right. Well, the community seems to favor 24 volt. That could mean lighter wiring for the same power. Not quite sure what that does to the stepper motors. There is a section on the RepRap site about using ATX power supplies (I have a few), but I missed mention of how you get from 12v to 24v. Some sort of step-up circuit, presumably (and hopefully efficient). Or run the hot-end at 24v and the motion-steppers at 12v? (That does not sound like fun.)
I expect to use an AC heated bed (so that largest-load does not go through the power supply).
I expect to use a Raspberry Pi 3 for the controller. As a software guy, when $35 gets a 64-bit quad CPU with a gigabyte of memory on a widely used board - and it runs Linux - if you suggest something else, I am going to look at you oddly.
So … 24 volt or 12 volt? I am going to assume 24v. Did I miss something?
I expect to use AC for the bed heater. There are existing solutions. Converting AC to DC, then into heat - is not efficient. This also downsizes the required power supply by quite a lot.
There are advantages to using PC power supplies, due to the mass market for PCs. Does the need to step-up to 24v wipe out that advantage? (And will someone please update the RepRap wiki.)
So I’m not sure a raspberry pi would be an easy choice for the control board, but I suspect you’d make a lot of people happy if you could code one up that would do the job properly.
Ditch the RPi unless you plan on porting a realtime Linux distro to it and then running MachineKit. 3d printer controllers almost always run bare-metal firmware on an embedded-systems level processor to achieve precise step timing on lots of IO pins. Non-realtime operating systems cannot readily do high-speed synchronous motion control, no matter how much processor power or ram they have. There’s always some OS housekeeping running in the background mucking up pulse timing. (Not saying it’s impossible, just that it’s hard enough that nobody sane wants to do it anymore. People have tried DMA and stripped-down Linux builds and all sorts of other tricks and it simply isn’t a very good solution.)
On the 24v question, the steppers will be happier with more supply voltage. (As long as you stay away from 8825 drivers.) Higher supply voltage means more top speed and the torque curve falls off slower. Then 24v heaters and wiring are less dangerous for the same power rating. People only really use 12v because in the early RepRap days all you could easily get was 12v ATX. Now 24v PSUs are common for driving LEDs and various other things.
@Preston_Bannister Might not be quite what you are looking for but this is the 24V ATX style power supply I use on all my printers. It has a 24+4 connection. The +4 is typically used to power the bed PCB. I have found them to be highly reliable.
There is a kit that adapts it to a Kycon connector or by removing the end you can wire it into screw terminals. Quite adaptable and very quiet.
@Ryan_Carlyle The realtime aspect is in fact a huge concern.
That said, a single one of the 1.2Ghz CPUs of the quad CPU in a Raspberry Pi is probably massively in excess of anything needed by this simple problem.
It is pretty easy to build custom Linux kernels. I have done this before.
There is also the concept of processor affinity.
On balance, this looks to me as a solvable problem. Bit worried about drivers or devices camping on the bus, but otherwise solvable.
My concern of the moment is choosing a 24v or not for my extruder, so … Thanks. Does this mean folk are not so much re-using ATX power supplies?
You can buy a 350 watt meanwell for about $50. Any manufacturer worth their weight is now using 24v. The smaller wiring, cheaper mosfets, and smaller connectors easily make up for the cost. Also the lower risk of burning your house down.
You may think that the raspberry pi is enough, but core clock speed does not correlate to GPIO capabilities. Even an Intel i7 would run into GPIO bottlenecks, it’s simply not designed for it. We use a microcontroller because it’s exactly what they’re designed to do. The beaglebone has a cape designed for 3d printing, and it is uniquely fitting for 3d printing.
The original poster seems to have disappeared from the Raspberry Pi site. Also I find the thread entirely predictable.
The arguments of “Morgaine” (whomever that was) are valid.
The arguments against, are largely wrong.
Getting realtime behavior out of a general purpose operating system is indeed tricky. This is why CNC machines ran DOS for quite a long time (maybe even now).
But when a $35 device has an obscene number of processor cycles, the problem is not the same.
@Preston_Bannister if you want to be the next person to try to climb that mountain, go ahead. There’s a LOT of demand for it, and many people have tried. There have been some working (for some definition of “working”) RPi printers. But I think you should look pretty deeply into MachineKit on a Beaglebone first, it can cover a lot of the stuff you want to do. For example, tying motion planning to filament heating will be a lot easier to implement via HAL “wiring” than rewriting a standard motion planner.
@Preston_Bannister also, +1,000,000 for wanting to experiment and challenge paradigms. We’re pretty blunt here – hope you don’t take it as discouragement. Just want to help you get up to speed on the “state of the art” so you don’t burn up too much energy rediscovering stuff the community has already looked at (but not documented worth a darn).
@Ryan_Carlyle Thanks. I have been told by others that I was wrong before. Sometimes they were right. Does not bother me.
I do not expect everything attempted will work. But I kind of have a history of making guesses that do work out. And of course I always have alternate plans.
