(continued from forum … [url]https://www.ghielectronics.com/community/forum/topic?id=22868&page=3[/url])
My approach to that was to measure the filament diameter just before it goes into the extruder. Alternatively, I was going to do produce my own type of filament to use with my commercial machine. This way I know that my users will get a good quality build.
[quote=“Mr. John Smith”]
(continued from forum … https://www.ghielectronics.com/community/forum/topic?id=22868&page=3)
My approach to that was to measure the filament diameter just before it goes into the extruder.[/quote]
There are people that have been working on this for some time now. It’s certainly not as easy as it may sound and no one has really shown that it actually improves anything (that I’ve seen so far…).
Companies have also tried this and they are generally hated by their customers and some just plain go out of business (i.e. Cubify) because people don’t want to buy a printer that requires filament that costs 2x - 3x that of normal filaments. Personally, I would let filament makers make filament and printer makers make printers. Even the cheapest filaments these days have fairly consistent diameters. Its much better now than it was even a year ago. If you want to provide value to your customers then go with the thickness sensor approach and let them use the filament of their choice.
@ ianlee74 - Filament wise, just don’t make it cost 3x more than normal filament. Then you won’t be hated. With the optical sensors, it can only determine the thickness in one direction. i was thinking about having 2 optical mouse sensors so that I can get a better diameter measurement.
Maybe. I still think giving customers choices instead of telling them what to do is always a better route to take.
I think it was in that article from HaD or maybe from one linked to it but one of the guys is doing it by measuring shadows from several different directions.
@ Mr. John Smith - I am actually in the early stages of building my first 3D printer (wish me luck!). I considered several commercial offerings, but I would never under any circumstances consider a system which required me to purchase filament from a single source. If for no other reason, the 3D printer market is so young and so fluid, that even in the statistically unlikely event that this single source stayed in business for any reasonable amount of time, the specific printer and filament is unlikely to remain available for very long, simply because progress is so rapid.
What kind are you building?
@ ianlee74 - A CoreXY RepRap variant, named Fuse Box CoreXY: FuseBox CoreXY 3D Printer by _AlexY - Thingiverse
Nice! I’ve been thinking I’ll build a CoreXY once I’m done with my current delta build.
@ ianlee74 - I love the idea of a delta printer. Mostly just because they look so dang cool. They’re probably faster, too, but mostly it’s how cool they look 
I agree! But, I also suspect they’re a lot more work to keep calibrated and running than a Cartesian. Currently, I have a Mini Kossel with a 180mm bed that I built a couple years ago. I need to print bigger things now and so I’m building one with a 300mm bed. Deltas scale horribly in the X & Y directions but will go as tall as you want in the Z direction. I want the CoreXY for even larger prints. I think the CoreXY holds a close second to deltas as far as coolness and speed goes.
When everything is working well, I can usually print at 150 - 200 mm/sec. Not many Cartesians I’ve seen can do that.
Yes, delta robots have some of the fastest actuations of any type of robotic system. If you could extrude as fast as it could move then you could build objects in seconds instead of hours. I don’t think that they drift out of calibration though. Just compensate in software. Have you ever tried making one that used 4 arms? This company Adept Robotics makes a quatro that is so fast you can hear the air sing as it whips around the build envelope.
When mine is working well, I only calibrate every few weeks. Right now it is not working well… I swapped out the electronics over two weeks ago to a Duet board and also added an IR probe for autocalibration. In the process, I managed to introduce an imperfection in my frame squareness and have been trying to get a good calibration and print for over two weeks now. They are very sensitive to change. Especially when built with plastic parts. My new one will have all metal brackets. No important frame pieces will be plastic.
I haven’t and wouldn’t unless I had a really heavy effector or some other good reason to do so. Three arms and towers is the optimal configuration for a delta. Anything more is just asking for problems and added expense.
@ Mr. John Smith - I’m pretty sure you’re hearing the servomotors, not the air. I believe one of the big drawbacks to delta robots is that they’re not uniformly precise; that is, depending on where you are in the envelope, you have more or less precision (travel per motor movement). This may or may not be a problem depending on what you’re trying to do and how much precision you require…
@ godefroi - The precision of the delta is directly dependent on how accurately the arms were made. Any variations in the length will manifest as the aforementioned problem. You can however compensate for that in software, assuming the machine has sufficient resolution.
@ Mr. John Smith - Regardless of how precisely the arms are made, depending on where the head is located, one step on a motor might move the head .1mm or it might move the head .2mm (for example). The “spatial resolution” is not constant throughout the build area. Specifically, the closer to a “tower” you get, the lower your spatial resolution. This probably doesn’t matter for most applications or most people, but it is a consideration.
I still would love to build one.
A delta only works when the arms are perfect (length within 50 microns or less and perfectly parallel). Otherwise, you will get effector tilt or twist which you can’t (reasonably) compensate for in software. Lots of delta printer noobs jump into a build and think they can bolt together the parts and then just kick off the autocalibration routine and everything will just work. They usually spend the next month figuring out that they must get the mechanical parts of the printer perfect before autocalibration can even be attempted. Only minuscule imperfections such as slight (>0.05mm) bed tilt and tiny imperfections in tower angles can reasonably be fixed via software.
Resolution on larger beds isn’t generally a problem with printers because delta’s usually don’t have very large beds because they scale so terribly in that direction. By scaling terribly I mean in terms of cost & size. Arm length & bed diameter scale almost linearly.
@ ianlee74 - Indeed.
I love it when you imperial dudes see the light…
Metric is easy. Fractions are accurate 
(but I’m a lazy programmer. So, I’ll take easy.)
Someone asked me how far San Francisco was from Christchurch…
I said, knowing my metric system from my US schooling, “about 11 megameters”
But they had no idea what that meant. Finally I rephrased it as “11 thousand kilometers” - which they understood, but is a mangling of the metric system. The beauty of the metric system is the new unit for each 3 orders of magnitude.
Kilo, mega, giga, tera, peta, exa, etc.
Saying “Eleven thousand kilometers” makes as much sense as saying “about a hundred thousand decameters”
SO, I’ve decided that the whole “hah hah, the USA doesnt use the metric system - so old fashioned” is a hollow criticism, since no one actually uses it correctly anyway, Here in NZ, not only do they use “thousand kilometers” but they also announce birth weight in lb/oz “she was 9lb 4oz!”, and tire pressure in PSI. It’s so random what is metric and what is imperial. I found the same situation in the UK.