MakerGeeks.com Mini Kossel Kit - Junk!
Posted by GeoDave
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! January 25, 2015 03:44PM |
Quote
rogueqd
I'm looking at buying a Kossel as my first printer. I'd already decided not to buy from MakerGeeks afrer seeing the youtube video.
I found this site that looks good (looks dont count for much) and my emails have been responded to quickly and helpfully. Does anyone have any experience with them?
[builda3dprinter.eu]
Any other comments would be helpful too. I have some previous Arduino and multicopter building experience.
I bought a set of printed parts, rails to wheels conversion kit and the extruder spur gear from builda3dprinter.eu early last year and everything I received from them was high quality:
A photo of my printer is on their showcase page (Tristan's Kossel UK): [builda3dprinter.eu]
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! January 25, 2015 04:59PM |
Quote
rogueqd
Thanks Tristan. I like the way you've mounted the filament reel at the top did you design that yourself?
After more research I'm wondering if an XL is the way to go. The creator of the Griffin printer said Kossels dont have enough rigidity and can wobble a bit at larger sizes meaning you have to print slower to maintain accuracy. Ardon from builda3dprinter.eu also mentioned they could be wobbly but said it didn't affect the prints. Have you noticed anything like that?
Yes, I designed the spool holder myself!
I wanted to make use of the otherwise wasted space at the top. The design is a bit weak where it attaches through the lid but I designed it around some bearings I already had, i.e. cheap!I have noticed the effects of "frame wobble", it mostly shows up when the extruder changes direction suddenly, like at a 90 degree corner. The result is a "ripple" on vertical walls after the direction change:
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! January 25, 2015 06:47PM |
It's a nice use of the space. Do you think the mass of the filament spool helps prevent the wobble, or makes it worse?Quote
Tris
Yes, I designed the spool holder myself!
I have noticed the effects of "frame wobble", it mostly shows up when the extruder changes direction suddenly, like at a 90 degree corner. The result is a "ripple" on vertical walls after the direction change:
It seems to me that twisting is where the frame is weakest. Do you think little struts from the top bar of the bottom section, near where the motor mounts meet the frame, would hold the uprights stiffer and stop it wobbling? I guess you could even make large triangles from the middle of each side as long as the rollers could still come down far enough.
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! January 25, 2015 08:47PM |
Ok, I get it. I'm not an engineer, but the resonant frequency will come from how springy the frame is, right?Quote
Tris
It may make it better OR worse, it's all about the resonant frequency of the frame vs how fast the effector is oscillating...
Corner braces would certainly improve rigidity. There's already a good design on Thingiverse for such a brace: [www.thingiverse.com]
Nice to see the brace, so I was on the right track then.
Hey it's great having you answer my questions, thanks, but is there a good place for a beginner to go to learn about all these things. Build guides are a good start, but they don't teach you much about the theory or operation. Forums are also great, but you generally have to read through a ton of irrelevant stuff to find the little nuggets of knowledge.
Is there something like a beginners guide to owning a 3D printer? Preferably kossel, or delta, specific?
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! January 26, 2015 07:22PM |
All you really need to know is that when the effector is accelerated or decellerated in a particular direction, the frame will want to move in the opposite direction (albeit less as the mass of the frame is much greater than that of the effector!)
If the vertical extrusions are not rigid enough, they absorb the forces exerted by the effector by bending, causing momentarily inaccurate effector placement.
Make the frame as rigid as possible to prevent it from absorbing these forces and deforming. Of course, this energy has to go somewhere so I mounted my printer on vibration dampening rubber feet so the frame can move about slightly as a whole and not distort so much. Doing this also relieves the strain on the moving parts somewhat, especially the ball joints.
As for a beginners guide, I havent seen one yet. I had to do the whole "knowledge nugget" mining thing myself too!
If the vertical extrusions are not rigid enough, they absorb the forces exerted by the effector by bending, causing momentarily inaccurate effector placement.
Make the frame as rigid as possible to prevent it from absorbing these forces and deforming. Of course, this energy has to go somewhere so I mounted my printer on vibration dampening rubber feet so the frame can move about slightly as a whole and not distort so much. Doing this also relieves the strain on the moving parts somewhat, especially the ball joints.
As for a beginners guide, I havent seen one yet. I had to do the whole "knowledge nugget" mining thing myself too!
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! May 17, 2015 11:34PM |
I am SO close to taking the whole assembly and tossing $1200 in the garbage. I have invested many months of re-engineering and probably another $500 and I still can't get a decent print, if at all. I bought it last October! The entire spider assembly, every part, has been replaced with better or self-designed parts. It has become the biggest waste of time and money I have ever encountered in my life of hobby and professional engineering.
-The hot-end jams too easily, despite that I replaced it with an E3D V6. I don't know how others make 3mm filament work on a delta, it just seems to require too much force..I have to hang a tremendous weight on the extruder hob clamp to keep it from stripping eventually, although that causes it's own problems. It might run for a while but then it starts starving due to hob grinding. Taking it out to unplug the hot-end is such a major pain that haven't heated up the hotend in months, I'm just afraid to deal with it without some hope I have fixed anything. My suspicion is that a 3mm filament has to go slower han a 1.75, which allows heat to actually travel backwards inside the filament and make it squishy, which then causes jamming. I actually got my best prints by stopping attempting to take it slow; I just jumped to 100mm/min print speed and suddenly it worked, for a while anyway. It was a short-lived triumph. No amount of heat-breaking can improve such a thing, only really fast printing!
-The frame has all kinds of skew on it. I've tried to straighten it out and it just can't be done with my tools. I have a sideways skew on my X/Y plane of 1mm on a 100mm square that I can't figure out, but I suspect my towers are not parallel and it's really hard to measure without a good large internal caliper, not to mention that nothing is orthogonal so conventional squaring techniques are also useless. I have inconsistent dimensions from the center to the towers and different types of dimensions seem to contradict each other. It is VERY difficult to de-transform measurements of errors made on a delta back into cartesian space, so intuition is almost useless. G30 autocal just wanders away from realistic parameters the longer it runs, which probably means there are too many separate distortions.
-I am about to lose interest in the whole 3D printer thing anyway if it is going to be so high-maintenance. I have too many other projects to work on and thought the delta could make parts for those projects, and maybe tweak improvements, not become a black hole filled with pus-spewing flames of hell.
-Josh sent me a free box of injection molded vertices but as commented before they are sloppy and useless. Plus the bottom ones were replaced with two smaller ones stacked up so that one delta could use 9 identical parts, a good approach for injection molded parts but now the lower vertices are weaker because they have lost shearing strength! WHAT IS GOING ON OVER THERE?
If I decide to just buy a working assembly how much can I expect to use it vs doing maintenance? Should I bother with a delta or stick with conventional cartesian? It seems like the biggest attraction for most people is 'It looks cool'.
I am generally interested in robotic/mechanical parts for other engineering projects so quality and accuracy are high priority.
BTW, with the zillions of pre-made printers out there, I would like to find some comparative reviews by experienced people, not just 'I had no problems' kind of feedback. Any good places to look? I would probalby go with a Cartesian given the filament limitations of the deltas because of the bowden tube approach. I want full access to the variety of filaments out there and squishy ones are hard to feed in the deltas.
thanks
-The hot-end jams too easily, despite that I replaced it with an E3D V6. I don't know how others make 3mm filament work on a delta, it just seems to require too much force..I have to hang a tremendous weight on the extruder hob clamp to keep it from stripping eventually, although that causes it's own problems. It might run for a while but then it starts starving due to hob grinding. Taking it out to unplug the hot-end is such a major pain that haven't heated up the hotend in months, I'm just afraid to deal with it without some hope I have fixed anything. My suspicion is that a 3mm filament has to go slower han a 1.75, which allows heat to actually travel backwards inside the filament and make it squishy, which then causes jamming. I actually got my best prints by stopping attempting to take it slow; I just jumped to 100mm/min print speed and suddenly it worked, for a while anyway. It was a short-lived triumph. No amount of heat-breaking can improve such a thing, only really fast printing!
-The frame has all kinds of skew on it. I've tried to straighten it out and it just can't be done with my tools. I have a sideways skew on my X/Y plane of 1mm on a 100mm square that I can't figure out, but I suspect my towers are not parallel and it's really hard to measure without a good large internal caliper, not to mention that nothing is orthogonal so conventional squaring techniques are also useless. I have inconsistent dimensions from the center to the towers and different types of dimensions seem to contradict each other. It is VERY difficult to de-transform measurements of errors made on a delta back into cartesian space, so intuition is almost useless. G30 autocal just wanders away from realistic parameters the longer it runs, which probably means there are too many separate distortions.
-I am about to lose interest in the whole 3D printer thing anyway if it is going to be so high-maintenance. I have too many other projects to work on and thought the delta could make parts for those projects, and maybe tweak improvements, not become a black hole filled with pus-spewing flames of hell.
-Josh sent me a free box of injection molded vertices but as commented before they are sloppy and useless. Plus the bottom ones were replaced with two smaller ones stacked up so that one delta could use 9 identical parts, a good approach for injection molded parts but now the lower vertices are weaker because they have lost shearing strength! WHAT IS GOING ON OVER THERE?
If I decide to just buy a working assembly how much can I expect to use it vs doing maintenance? Should I bother with a delta or stick with conventional cartesian? It seems like the biggest attraction for most people is 'It looks cool'.
I am generally interested in robotic/mechanical parts for other engineering projects so quality and accuracy are high priority.
BTW, with the zillions of pre-made printers out there, I would like to find some comparative reviews by experienced people, not just 'I had no problems' kind of feedback. Any good places to look? I would probalby go with a Cartesian given the filament limitations of the deltas because of the bowden tube approach. I want full access to the variety of filaments out there and squishy ones are hard to feed in the deltas.
thanks
|
Re: MakerGeeks.com Mini Kossel Kit - Junk! May 18, 2015 05:25PM |
Thanks everybody for all the cool-headed sympathy and feedback; my boiling point has risen to productive levels again.
Thanks for the feedback dc42, I appreciate it. The forums have been responsive but sometimes seem to run out of ideas. (especially jamming)
I took too much for granted with the kit. I will have to tear it all down and start over with more care. I seems to me that the easiest way to ensure squareness would be some kind of plate with at least four attachments between each tower; I am picturing clear acrylic, although this is where I hate the deltas: interfacing with non-orthogonal components. A handful of 60 degree adapters would be awesome and I've designed some for printing.
One Idea I had last night was to remove and rotate the spider assembly 120 degrees. If the errors stayed the same, it is due to the chassis, otherwise is something in the spider assy. After the swap, my skewed XY box (with perfect dimensions) turned into some kind of more complicated combination of skew and dimension errors. I will have to do some repeats to see if these are real, but if so, I will have to look at the spider assy again; I rebuilt my own rods and printed my own effector platform (on a expensive commercial printer) so that is still TBD.
I have since realized that the standard 170 diameter plate is too small for many of my desires; if I have the patience (and cost of different parts), I would definitely move up to a 220 or bigger. I can't imagine parts taller than the current Kossel, but just changing the uprights and the belts is a REALLY easy upgrade in the future. The extrusions from the MakerGeeks kit are one of the few parts I would reuse.
And thanks for the heads-up up on the Robotdigg parts! One factor I've observed of the reprap paradigm is the focus on saving $ and the general DIY mentality that tends to focus on low-cost/easy-to-procure items. If this is going to be a precision tool, I don't want to cut corners. Self-replication is a cool idea, until it gets in the way of quality and/or required baby-sitting. I'm not interested in tweaking 'my baby' to keep it running well as a project. Considering that the delta's have no diagonal components, the quality and properties of the vertex parts are critical. And I would stick with the 2020 parts are well.
Funny, I upped the current a lot because my extruder was skipping too easily. Whether the extruder skips constantly or comes to complete halt because of chewed-through filament, the end result is a failed part. What I want and have designed in my head is a much better extruder with more grip. Even a drive with a second opposing hob would double the grip. A spread-out rubber opposing tank-tread grip would be ideal but of course a lot more complicated.
Less desk space, +1. Unlimited height possible, +1. Faster speed, +1. I don't have experience with the cartesians (this is my first step into the 3D printing world) so I can't compare very well.
I am curious though how you worked out that the the electronics combo you mentioned (which comes with the MakerGeeks kit and NOT one of the issues I've had) is under-powered. Have you observed timing errors?
One of the strong arguments for the cartesians is the bowden tube. Clever solution but it was a compromise. Soft materials are hard (or impossible) to push, and the extra tube-drag doesn't help the extruder grip problem. That is another area that I have been pondering: how to bring the grip on the filament back to the effector while keeping the motor elsewhere. This is critical if you want to keep delta speeds with more than one hot-end. I've pictured a pair of steel-cabled bowdens driving the gripping mechanism on the effector platform, using an oscillatory quadrature feed mechanism instead of the current linear method, but that is definitely going into the cool-project zone, not a fix.
As for the force comments, here is an alternative perspective (the one my physics-background jumped to: less mathematical, more conceptual).
-Same extrusion rate means same amount of power required. Thinner filaments must mean faster feedrates. Since power = velocity times force, then the force must be lower. In that sense the 1.75 makes more sense unless the stepper can't go fast enough. High speeds don't seem like limiting factors in these designs whereas high torques and forces mean higher strains and lower efficiency with any power electronics. For this reason alone I have seriously toyed with converting to 1.75, although that eliminates any possibility of flexible materials (I want material options!) and I already pre-purchased lots of 3mm filament in anticipation of printing all kinds of cool designs.
Does anybody have suggestions about working out the jamming issues? That is the scary part for me because it is so hard to get feedback about what is happening. Either it is sticking to something, or the hole is clogging, or the filament flow properties have changed somehow. Trying to troubleshoot all three seems to reduce to: try,try, try! Maybe there is better forum or thread for this issue? It's not really delta specific.
Quote
dc42
1. Towers not quite perpendicular to the bed. In my case, there was no slop in the corner pieces, but I wasn't careful to ensure that the vertical and horizontal extrusions were perpendicular when i assembled it. Also, I had to hammer the extrusions through the corner pieces (as described in the instructions), which peels off shavings of plastic, and this could lead to the holes ending up not quite straight even if they were initially. I tried loosening some of the screws, pushing the frame square, and tightening the screws again, but I wasn't able to improve on it much. The effect of non-perpendicular towers was that after doing the usual 4-factor calibration (i.e. endstops and delta radius), the nozzle height was wrong along radial lines midway between the towers. I solved it by implementing 7-factor autocalibration from a single set of probe points in the firmware. What the autocalibration actually does is to adjust the tower positions, which to a first approximation compensates for the leaning towers. The firmware I am using also supports orthogonal axis compensation, so I don't have to put up with leaning prints.
Thanks for the feedback dc42, I appreciate it. The forums have been responsive but sometimes seem to run out of ideas. (especially jamming)
I took too much for granted with the kit. I will have to tear it all down and start over with more care. I seems to me that the easiest way to ensure squareness would be some kind of plate with at least four attachments between each tower; I am picturing clear acrylic, although this is where I hate the deltas: interfacing with non-orthogonal components. A handful of 60 degree adapters would be awesome and I've designed some for printing.
One Idea I had last night was to remove and rotate the spider assembly 120 degrees. If the errors stayed the same, it is due to the chassis, otherwise is something in the spider assy. After the swap, my skewed XY box (with perfect dimensions) turned into some kind of more complicated combination of skew and dimension errors. I will have to do some repeats to see if these are real, but if so, I will have to look at the spider assy again; I rebuilt my own rods and printed my own effector platform (on a expensive commercial printer) so that is still TBD.
Quote
dc42
When I rebuilt the machine to increase the printable diameter to 300mm, I printed new corners and took more care when assembling the frame. Despite this, the towers still lean slightly.
I am currently converting the machine to use 2020 extrusion because I want more printing height. As I am a perfectionist and would like to eliminate the lean, this time I am using metal corners from Robotdigg.
I have since realized that the standard 170 diameter plate is too small for many of my desires; if I have the patience (and cost of different parts), I would definitely move up to a 220 or bigger. I can't imagine parts taller than the current Kossel, but just changing the uprights and the belts is a REALLY easy upgrade in the future. The extrusions from the MakerGeeks kit are one of the few parts I would reuse.
And thanks for the heads-up up on the Robotdigg parts! One factor I've observed of the reprap paradigm is the focus on saving $ and the general DIY mentality that tends to focus on low-cost/easy-to-procure items. If this is going to be a precision tool, I don't want to cut corners. Self-replication is a cool idea, until it gets in the way of quality and/or required baby-sitting. I'm not interested in tweaking 'my baby' to keep it running well as a project. Considering that the delta's have no diagonal components, the quality and properties of the vertex parts are critical. And I would stick with the 2020 parts are well.
Quote
dc42
2. Extruder grinding through the filament, resulting in loss of traction. This plagued me for a long time, but I solved it eventually by reducing the extruder motor current from 800mA to 500mA. The reason is that if the motor can't push the filament, e.g. because the nozzle is not hot enough, or too close to the bed, or over-extruding, then it's important that the motor skips steps in preference to grinding through the filament.
Funny, I upped the current a lot because my extruder was skipping too easily. Whether the extruder skips constantly or comes to complete halt because of chewed-through filament, the end result is a failed part. What I want and have designed in my head is a much better extruder with more grip. Even a drive with a second opposing hob would double the grip. A spread-out rubber opposing tank-tread grip would be ideal but of course a lot more complicated.
Quote
dc42
With these fixed, my Kossel produces prints more reliably and more quickly than my Cartesian printer. It also takes up less desk space. So I definitely prefer Delta over Cartesian. I wouldn't run a delta on Arduino Mega/RAMPS/Marlin, because the processor lacks the performance to drive a delta easily and the firmware lacks the features I need, but it appears that a lot of of people do.
Less desk space, +1. Unlimited height possible, +1. Faster speed, +1. I don't have experience with the cartesians (this is my first step into the 3D printing world) so I can't compare very well.
I am curious though how you worked out that the the electronics combo you mentioned (which comes with the MakerGeeks kit and NOT one of the issues I've had) is under-powered. Have you observed timing errors?
One of the strong arguments for the cartesians is the bowden tube. Clever solution but it was a compromise. Soft materials are hard (or impossible) to push, and the extra tube-drag doesn't help the extruder grip problem. That is another area that I have been pondering: how to bring the grip on the filament back to the effector while keeping the motor elsewhere. This is critical if you want to keep delta speeds with more than one hot-end. I've pictured a pair of steel-cabled bowdens driving the gripping mechanism on the effector platform, using an oscillatory quadrature feed mechanism instead of the current linear method, but that is definitely going into the cool-project zone, not a fix.
Thanks, but I'm going to stick with the 2020 design I currently have.Quote
dc42
I'd offer to send you my 1515 Kossel vertices, but as you are quoting prices in $, I'm guessing you are in the USA, and I am in the UK to the postage could get expensive.
As for the force comments, here is an alternative perspective (the one my physics-background jumped to: less mathematical, more conceptual).
-Same extrusion rate means same amount of power required. Thinner filaments must mean faster feedrates. Since power = velocity times force, then the force must be lower. In that sense the 1.75 makes more sense unless the stepper can't go fast enough. High speeds don't seem like limiting factors in these designs whereas high torques and forces mean higher strains and lower efficiency with any power electronics. For this reason alone I have seriously toyed with converting to 1.75, although that eliminates any possibility of flexible materials (I want material options!) and I already pre-purchased lots of 3mm filament in anticipation of printing all kinds of cool designs.
Does anybody have suggestions about working out the jamming issues? That is the scary part for me because it is so hard to get feedback about what is happening. Either it is sticking to something, or the hole is clogging, or the filament flow properties have changed somehow. Trying to troubleshoot all three seems to reduce to: try,try, try! Maybe there is better forum or thread for this issue? It's not really delta specific.
Sorry, only registered users may post in this forum.