You should also probably go with a bowden extruder if you're going for something super accurate. By separating the extruder from the hotend, you can make sure there's less weight for the motors to move around.

As for speed, at my old job I used to work on assorted CNC equipment, primarily for the sign industry. Some of the knife cutting machines for vinyl and what not can do 3,500 inches per minute, (58 1/3 inch/s, 1481 2/3 mm/s for you metric folks) and THAT is scary. Hell, it could cut material at 2,500 without too much difficulty. 50,000 RPM spindle, too. Absolutely terrifying.

What's the deal; with the spidery printers. Look very cool in motion but seem to be fraught with high-level accuracy issues as the angular resolutions (calculations) will be being rounded a lot of the time. Can't help but think I may just neeeeeed one though. LOL

Thanks for that. Just a comment on the Scales etc. They are there for real-time feedback as the driving force is a servo motor, high-torque DC motors. Unlike steppers, they have no idea if where they are so the scale feeds back to the electronics. Scales are not the only option for servos, by far the most common feedback devices are quadrature rotary encoders.

The BIG advantage of servos over steppers is they can never lose steps due to torque overload. They will always arrive at the exact spot intended, or keep trying to. Well, barring a crushed hand causing a panic button to be smacked. Another advantage to servos is they are not speed/torque dependent. Steppers are at max-torque when at standstill or holding. They lose torque (and often steps) the faster they are made to turn and will eventually stall. Servos are DC motors and the more volts/amps the faster and harder they go.

I have five-cnc mills and a couple of cnc lathes, Some use steppers and some use servos and the servos are deadly machines. I have one servo machine (weighs 5200-lbs)  that can do 20"/second and you do NOT see that coming. It has a 5hp spindle motor and can haul a 1200-lbs chunk of steel at that speed. Can we all say "many test runs before putting cutters and material  in?"

What I have in mind so far will be very rigid as I have seen some laughable flappy examples and not just the cheap ones.

OK, so I think I will press ahead with the current layout and may look at servos instead of steppers. This begs a question, just how fast can the plastic be laid down?

I take your points about the quality of design of the hot end being a controlling factor and I would be guided by opinions from here as to which are the better hot ends.

Thanks so far

The higher end Delta Bots can lay as fast as 150mm/s. The limitation in the case of a machine as you described would be the hotend being able to maintain the heat and feed the filament at the speed you wanted to run at. Filament acts as a natural heatsink and sucks heat away from the hotend. Naturally the faster you feed the faster it removes. I really don't see a E3D which is about the best on the market for hotends being able to do much more than 150mm/s. You may have to design your own hotend as well.

The BIG advantage of servos over steppers is they can never lose steps due to torque overload. They will always arrive at the exact spot intended, or keep trying to. Well, barring a crushed hand causing a panic button to be smacked. Another advantage to servos is they are not speed/torque dependent. Steppers are at max-torque when at standstill or holding. They lose torque (and often steps) the faster they are made to turn and will eventually stall. Servos are DC motors and the more volts/amps the faster and harder they go.

I have five-cnc mills and a couple of cnc lathes, Some use steppers and some use servos and the servos are deadly machines. I have one servo machine (weighs 5200-lbs)  that can do 20"/second and you do NOT see that coming. It has a 5hp spindle motor and can haul a 1200-lbs chunk of steel at that speed. Can we all say "many test runs before putting cutters and material  in?"

What I have in mind so far will be very rigid as I have seen some laughable flappy examples and not just the cheap ones.

OK, so I think I will press ahead with the current layout and may look at servos instead of steppers. This begs a question, just how fast can the plastic be laid down?

I take your points about the quality of design of the hot end being a controlling factor and I would be guided by opinions from here as to which are the better hot ends.

Thanks so far

If anybody feels like taking up the torch, I don't know of anybody who has added scales to a 3d printer but they're well known in the DIY CNC Mill community. you can buy kits for mills that add DROs (digital readouts), and those can output to various software (some open source). The trouble would be seeing if it could be integrated into current firmware so the printer could take care of itself. There are even hacks out there for buying cheap chinese digital calipers and turning them into cheap DRO scales

Edit: I knew I knew a project. http://www.yuriystoys.com/2012/01/readi … -with.html Here's how to get the scales to communicate with an arduino. You get a pulse per distance measurement, kinda like a reverse stepper motor. This guy did a great project making his own digital readout for manual mills using an arduino and android.

They work by the same method as digital calipers, so the end result is that the machine can adjust its position to compensate for any movement irregularities. Look at DRO kits to see a bit of what I'm talking about.

So basically realtime hysteresis correction. Sounds pretty neat, and likely more reliable than hardcoded hysteresis correction techniques.

Also a smaller nozzle will create more resolution along XY. Your Z resolution is limited by the mechanics used to drive it--eventually you run out of steps per mm.

The big thing for most 3d printers when it comes to having the best looking prints has more to do with the extruder and hotend set up properly. Motion can be spot on but if your filament isn't coming out at right temperature or at the right layer height you are going to have problems.

If you are trying to go super precise, look at using 0.9° steppers instead of 1.8°. That way you get the same resolution without as much microstepping. Microsteps are inherently less accurate than full steps and get less accurate the more you divide, so you should get better results with 0.9° at 1/16 vs 1.8° at 1/32.

Many of the super fancy machines out there (and i'm talking more about mills than printers here) have scales attached to them. These are basically sensors that report the actual positioning back to the machine. You can either get high end glass scales or you can get cheaper magnetic scales. They work by the same method as digital calipers, so the end result is that the machine can adjust its position to compensate for any movement irregularities. Look at DRO kits to see a bit of what I'm talking about.

What defines the 20+ models?

I am a retired Mechanical Engineer and have a pretty complete machine shop at home and want to design and build a printer. I'd like it be be capable of the best resolution possible without breaking the bank.

I am guessing that the good ones use ball-screws for movement as opposed to belts and a solid mechanical design, the steppers can run in micro-step mode at no extra electronics cost, but other than those, I'd like to know what makes for the hi-res as those three don't account for a lot of extra costs.

Thanks, oh and of course, my design will be all open source hardware using open source software and electronics.

Dave