Was thinking through various ideas for designing a very fast printer tonight.
In your experience, what’s the limiting factor when it comes to increasing speed (extrusion rate, nozzle travel speed, thermal cycling issues, etc.)?
After inertia of the moving parts, probably cooling of the print.
8 bit arduino will constrain your maximum # of commands sent per second which constraints’ your top speed. Something like the Smoothie board would solve that issue as it’s 32bit. After that, mechanically, you’re looking to minimize rolling weight as much as practical, while increasing the rigidity of the bot. Practically, using machined Aluminium and core x/y would solve both (IMO), you can always use more powerful stepper motors to ensure you have the holding torgue to compensate for high accelerations. Hot end would probably have to be water cooled. Electronics and steppers air cooled. Syncromesh for belts.
I gotta say, tim and michael pretty much hit the nail on the head. weight/inertia as well as top speed of your microcontroller are the current limiting factors. This is the whole reason why the ultimaker is so fast, the XY overhead gantry is possibly the best method for speed.
@Tim_Rastall Have you heard of anyone using the synchromesh successfully with corexy/hbot? I know @Whosa_whatsis was talking about testing this. I’m still researching the core xy for my upgrade and finally gave up on spectra after seeing polygonalhell’s blog. Anyways, sorry to hijack thread, i was just curious to the synchromesh as it’s nice and small, but people have had reservations regarding backlash, which is less of an issue in core xy, but obviously still a major concern.
Thanks for the feedback guys. I figured lighter=faster (like most things 
but I wasn’t sure if there was a point where axis speed would be limited by how fast the extruder could move plastic, or if you’d hit a limit where the plastic going from liquid to solid would be the pinch point.
From the sounds of it, for a small printer with an x/y gantry the controller cpu would be the “hard wall” then; would you agree?
With the setup we use for the Bukobot and Bukito (both of which have moving platforms, and the Bukobot carries it’s extruder motor around too), the limiting factor is the stepper driver control pulsing maxing-out at 355mm/s. This could, of course, be doubled by going from 1/32 microstepping to 1/16, but we like the super-smooth surfaces that the increased resolution combined with the smoothness of synchromesh (I don’t know why everyone thinks it has backlash issues, but they’re wrong) gives us.
Inertia quickly becomes a bigger issue than top speed, which is why you will never get a decent printer by strapping an extruder to a milling machine that is capable of moving at a meter per second, but only with zero jerk and 10mm/s^2 acceleration.
Also, most hot ends these days have short hot zones, which increase filament control but limit extrusion speed. If you want to print super-fast, you’ll need a longer hot zone – especially if you’re using a larger nozzle – so that the plastic stays in the hot zone long enough to reach extrusion temperature. You can only do so much by pushing the nozzle temperature higher.
I was baffled by @Tim_Rastall 's suggestions for a super-lightweight X/Y system followed by the suggestion to add water cooling to the hot end. Unless you are going to spray water at it from a distance, that would require carrying tubes full of water around with the carriage, defeating the purpose of all the other weight reductions.
@Whosa_whatsis Ah, yes. I didn’t explain that last bit very well did I. Allow me to at least attempt to unbaffle you:
Firstly, I assumed that if you are looking to hit those really high speeds, you are going to be pushing plastic through the extruder at extreme rates, as you know there is a consequential need to increase the hot end temperatures. At some stage your hot end would benefit from water cooling rather than air cooling, particularly if you want to decrease oozing issues that I assume would occur when the filament suddenly stops flying though the extruder and finds itself in a super-hot melt zone.
Secondly, if you were to run multiple nozzles and there’s a compelling argument to have at least 2 to maximise the quality/speed ratio (1 fine perimeter nozzle and a fat infill nozzle). Water cooling becomes even more useful as you can do away with 2 x heatsinks, 2 x fans etc, assuming the dual nozzles are cooled together.
Thirdly - water and tubing weight: Complex thing to work out as the inflow-outflow tubes would be fixed on one end and moving with the carriage on the other, but assuming a tube size with 5mm ID and say, 600mm in length, give a volume of 11,780 cubic mm, or 11.78 grammes pure water. Even twice that for the in and out tubes plus weight of the the tubes themselves doesn’t seem like a big addition.
Due deference to your knowledge of this stuff - I know you do it for a living but that was my reasoning.
@Eric_Moy No, although as you say @Whosa_whatsis has mentioned aspirations to that effect. Syncromesh should be less prone to backlash than belts - it’s got bugger all stretch and the way it meshes with the pulleys would mean it would be very resistant to slipping. At some stage I’m going to buy some and have a play. Which blog post has discouraged you from using spectra? I can’t find anything on polygonehell’s pages.
@Tim_Rastall took me a while to find, but crispy posted on http://reprap.org forum quoting an sdp-si apps engineerineer:
http://forums.reprap.org/read.php?1,181517,182360#msg-182360
although I believe he was quoting second hand. And I do take what applications engineers of distributors say with a grain of salt, as they aren’t generally well versed with all the products they sell (no offense to any AE’s out there)
@Whosa_whatsis I think the concern with backlash comes from the geometry of the cable. I’ve never used it, so I’ll just speak from my own assumption (no experience with it at all though). For a standard timing belt, back lash is a result of tensile deflection (stretching of the belt) and deflection of the teeth. This is mitigated by the steel strands in the belt, and the improved by the large contact patch of the GT2 (and maximizing # of teeth in contact).
With the syncromesh, you need to worry about stretching of the cable AND twisting of the cable, as twisting of the cable is like unscrewing a screw. Given the tiny diameter of the center filament of the synchromesh, there doesn’t appear to be a lot of torsional rigidity. Albeit if the cable untwists, the spiral threads of the cable will press against the grooves in the drive pulley, resisting the twist. I guess this would be a non-issue…
So honestly, it comes to a comfort factor thing for me. I know you have way more experience using the both GT2 and synchromesh than I do, but given the belt tension requirements of an H-bot/core xy, I’d rather wait for somebody else to test it out first, as I have neither GT2 nor synchromesh around to test, so I’d have to drop cash to test either. I really want to hear that the synchromesh works for core xy as it’s MUCH smaller and the pulleys and cables can be hidden easier. (hint hint) 
But then again, GT2 has become really cheap on aliexpress, so my wallet may get the better of me.
In all honesty, I can’t find any 3d printer projects out there that stuck with core xy, other than polygonalhell, but he used spectra and is having to use many more parts to accomodate this. Most designs have reverted to h-bot, which I’m not willing to do given the racking issues (which are apparent in the youtube videos) as I’m looking for a kinematically robust design that will be stable enough so I can spend more time printing, than tweaking mechnical parts.
@Eric_Moy @John_Driggers has a working core xy, he’s just keeping it a secret.
@John_Driggers if the core xy printer is the one that made the prints you’ve been posting, please please please post some pictures of your printer. You’ve got some sick prints in your posts, and I’d love to pilfer some designs from you
@Eric_Moy Good luck with that. IMO @John_Driggers is a big tease when it comes to this fabled corexy. In saying that, maybe he’ll be more responsive to a guy that has GIANT GREEN FISTS!