Let's make 3D printing more user friendly.

So I've recently had a misadventure getting RepRapFirmware to work. Frustration set in, I verbally abused a man who was trying to help me, and it was pointed out to me that 3D printers aren't to be expected to "just work."

Says who?

My RRF misadventure ended when I got utterly sick of trying to manually adjust for the automatic calibration system. I gave up when entering z-probe trigger heights per the manufacturer's instructions made the problem worse not better. I was monkeying with the direct measurements of the printer, what I'm going to call "low level parameters" since it has to do with the physical machine itself.

Say I had it printing OK near the back and left tower, but too close at the right tower. If I changed the trigger height at the probe point near the right tower, it would change the rest of the grid in strange ways. I had no way to directly tell it to raise that point and leave the rest where they currently are. I couldn't make adjustments in a high level way. The automatic system wasn't doing the job, in fact it seemed to be fighting me, resisting simplicity.

This is a system with a built-in graphical interface. Why can't I get a graphical represntation of the bed with points scattered around that let me adjust the carriage movement to the contour of the bed? If I see it is printing high between the back and left towers, just click the down arrow a couple times until it looks about right?

RepRapFirmware has a very nice feature: Instead of mucking around with diagonal rod length or whatever the way you have to in Marlin, there's a G-Code that lets you directly correct for dimensional inaccuracy in terms of percent. "I need it to be 0.037% bigger in the X direction." So you put in that number, no multiplying steps/mm or anything. I want to base an entire printer system around that concept. That the user tells the printer the change he needs to see, and the printer does the tedious math and repetitive processes. That's what computers are for.

Part of the RepRap movement is getting desktop manufacturing into every home. Let's start work on operator friendliness. Let's make a printer that just works.

I understand your frustration but try to remember reprap was born out of making a machine at home that could emulate what £30000 stratasys machines were doing until their patents ran out in the late 2000's. That's pretty miraculous when you think about it. The fact that a derivative of that machine can be bought for £159 today is also quite miraculous,but you'll need knowledge and patience to make it work, it's still a tool for (hobby) engineers not a microwave oven.

The rub is that these are not consumer machines, inherent in reprap is that you need to be technically minded, methodical and patient or at least willing to learn these skills to get it to work.

Turnkey machines exist already machines like the Cubicon [3dcubicon.com]

But you pay £2000+ for the ability to just turn it on and press print. Why? It's not innovation its that to make and sell a turnkey machine you need to have done extensive testing, certification, ce marking, radio frequency interference testing, have a user manual in 20 languages etc.... As well as have a well engineered reliable machine, made from expensive brand name parts you can trust. The cheap machines from China circumvent this by being kits or being sold as Diy machines.

I am with you in wanting 3d printing to just be easy. A machine you buy for your grandad and he could print something within a few minutes of opening the box. That's Apple's trick, the iPhone is intuitive and non- techies can work them. But they spent billions to get there (and make even more billions). There's much less demand for 3d printers than iPhones, when that demand is great enough apple will make a 3d printer and although I will hate everything about it, they'll sell millions and everyone will think they invented the 3d printer like people think they invented the pocket mp3 player.

If you want to make a cheap turnkey 3d printer then make one, I'll help you as will the rest of the reprap community, but trying to get it out there for less than £1000 will be hard work.

As for calibrating your machine the issue is/was probing tech, the ir probe is good but it's very limited by the way it works. Try our new Piezo probe and you'll have a perfect calibration every time, once you've got the probe working which requires a little skill, patience and knowledge.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

hey Mr Dj how do you print this?

0.2mm layers with support 70% overhang threshold. However take look at the 20mm version easier to print smaller and goes between groove mount and hotend.

[www.thingiverse.com]

Or look at Lykle's version.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Ah support not done that yet
Spent all mrning on my new hotend mount/shroud and still not ready to push the button
have to make sure its future proof-ish.

Edited 1 time(s). Last edit at 03/15/2017 07:33AM by MechaBits.

I know it'll take effort. Once upon a time, it was really hard to come by a quality reliable print head because machining such a thing (particularly the nozzle orifice) takes a pretty skilled machinist. Now you can grab an E3D hot end off the shelf, fully assembled for about $80.

You want to talk turnkey machines, Let me tell you about a local printer manufacturer, Fusion3. Their first product, the F-306: CoreXY machine with a cubic foot (306mm^3) build volume. Rambo board and 2004 reprapdiscount controller running Marlin. ~$3000 machine that IS a reprap. By and large the same logic board and software I use on my $300 kit-built Prusa. They discontinued that machine and now they build the F-400. Larger build volume, big enclosed box-chassis, Duet 0.85 and paneldue running RepRapFirmware. It too is a RepRap, built out of the same tinker toys in my toybox. Long as that's happening, I'm not worried about UL certification and vendors for high quality non-Chinesium screws.

I modified Marlin firmware to make the "paper test" easier to perform. Just wrote a little G-Code script and stuck it in ultralcd.cpp. Now I don't have to hitch it to a PC and type in lines of G-code or monkey with macros, I can do it from the built-in UI. I wrote filament load and unload scripts so I don't have to manually feed filament down the bowden tube. Little details that made a Marlin-powered Kossel that much easier to live with, that certainly didn't take billions of dollars in R&D. Neither did RepRapFirmware's excellent firmware update process, I wager.

BTW, in the shower earlier, I was wondering. Some folks have used three force-sensitive resistors under the build plate to detect when the nozzle has crashed into the bed for Z-probing. I wondered why not one FSR on the effector, and you're already there with your piezo approach it seems. Piezo disc generates a small current when the nozzle strikes the surface, so it'll probably need some amplifier and logic circuitry to generate an end-stop like signal for the controller. Does it just give a rising/falling edge, or if you park the nozzle on the bed and interrogate it, does the Z-probe read triggered the way the IR probe or a mechanical switch would?

There is a very extensive thread here in general and in duet forum about piezo's. However it gives a sharp voltage rise, which is fed through a differntiator and comparator, which are adjustable so we can tune any piezo disc. The key innovation was realising we could drill a hole in the piezo disc through which the filament passes, and it would still work.

Also it does help to have the perspective of the beginner in 3d printing as the software is designed by people who've been printing for years and take everything for granted.

Edited 1 time(s). Last edit at 03/15/2017 09:12AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Yeah I wouldn't have had any notion that you could take a power drill to a piezo disc and not kill it. I usually think of them as rather delicate. I'll monitor that development, though at the moment I'm not in the market for a Z-probe; trying to get rid of one in fact.

avr81, I understand your frustration. The automatic delta calibration in RRF works very well provided that you have a Z probe that has a consistent trigger height all over the bed. Likewise the mesh bed levelling, which is useful on large delta printers to compensate for residual geometrical errors and non-flat beds after auto calibration.

I suspect the problem you had was that your delta printer has geometrical errors that cause the effector tilt to vary with XY position. This is an extremely common problem. I have listed the main geometrical errors that cause varying effector tilt at [duet3d.com]. But getting all of these geometrical alignments right is easier said than done. Until 3 days ago, my own delta printer had significant varying effector tilt with XY position, causing the nozzle-to-Z-probe height difference to vary by more than 0.1mm This was despite my efforts to get all 6 rods exactly the same length and the bearing spacings exactly the same at the top and bottom ends of each pair of rods.

Varying effector tilt has at least two consequences. First, if you are using any sort of Z probe that is displaced sideways from the nozzle, tilt in one direction alters the relative heights of the probe and the nozzle. This causes the probe trigger height - expressed as the height above nozzle height - to vary with XY position. Second, it causes distortion of the print, for example printed lines that should be straight may come out bent.

My efforts to combat effector tilt so far have been:

1. Try to eliminate it as far as possible. See the link I gave earlier.

2. Design and manufacture a Z probe (the differential IR sensor) that is small enough to fit below the heatsink of a standard E3D hot end. This allows the sensor to be placed nearer the nozzle. The closer the probe is to the nozzle, the less its height changes relative to the nozzle when the effector tilts.

3. RepRapFirmware uniquely (AFAIK) allows you to add a trigger height correction to each auto calibration probe point. So if you can't get rid of the effector tilt, you can at least compensate for it when doing auto calibration. But measuring the trigger height of the Z probe at each auto calibration point is tedious, so this is by no means an ideal solution - and if you have varying effector tilt, this will still introduce distortion into your prints. Also, it is not practical to provide trigger height corrections to the mesh bed levelling system, because there are many more points in general (up to 441 for the Duet WiFi).

I recommend highly that you mount a bulls-eye spirit level on your effector, so that you can see how the effector tilt varies as the effector moves in the XY plane. Your issues may not be entirely a result of effector tilt, because it is also possible that the IR sensor is not giving a consistent trigger height over all areas of your print bed, because I understand you are using a transparent bed and the surface below it is not a uniform matt black.

In recent weeks I have pursued two avenues in my goal to make calibration and mesh levelling easier:

1. Delta geometry that is accurate by design. I have implemented a series of upgrades to my delta printer using a "correctness by construction" approach. This has been successful, and for the first time my delta has no significant effector tilt. See [www.duet3d.com]. So it is possible to avoid effector tilt by using precisely-made parts. But cheap kits are very unlikely to use parts made to high enough standards. That's why I recommend that anyone who wants to build a 3D printer on a tight budget avoids delta printer designs.

2. Z probing using the nozzle itself as the probe. This has been popular for some time using FSRs under the bed supports, and more recently DjDemonD and others have developed a new way of doing it using a piezo sensor in the print head. I am now working on two different types of PCB delta effector with nozzle-based Z probing built in. In fact now that I have eliminated effector tilt, I could use the IR probe without any trigger height corrections; but I am hopeful that the nozzle-based probe may work well on a wider variety of bed surfaces than the IR probe does.Time and experimentation will tell.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

It seems you are using your frustration at the difficult delta calibration process to disregard the million other ways RRF is superior to marlin. Keep your duet. Use this new Piezo approach, it's absurdly good, and enjoy the many other benefits of RRF, even things as basic as line segmentation.

I appreciate your advice, David, and I'll read those links. "Correctness by construction" is interesting. Let the record show I owe you a pint for putting up with my crap. But I'm not passive aggressively complaining in another subforum here, I genuinely want to brainstorm ways to make 3D printers more accessible and easier to use on a day-to-day basis, and used my latest frustrations as a case study. I'm not even specifically talking about deltas here. Right now, my Prusa, my Kossel, and a $3000 factory-built machine require equal know-what-you're-doingness to run.

I guess what I'm getting at is social engineering. I have some history with this from my aviation days. I wrote the checklist for my flight school's trainer--a little 10 page document that included every procedure from preflight inspection through the whole flight. There were two engine start procedures depending if the engine was hot or cold. I originally put them one after the other (the whole document is a collection of numbered lists), and every one of my students would invariably start down the Cold Start procedure, and with the engine running, start down the Hot Start procedure. I actually had to stop them from trying to start a running engine. I revised the checklist and placed the two procedures side by side to encourage the pilot to pick one or the other. I noticed that students wouldn't complete checklists that wrapped a page, they always stopped reading at the page break. So I carefully formatted the document so no procedure wrapped a page. I also tried to format the document so that the pilot would tend to turn pages when he didn't have much to do, ie, the cruise descent, before landing, landing, and after landing checklists were on the same page, and the pilot could turn to that page while his workload was low during cruise flight and just read from it when he had a lot to do. I printed the checklist on half-letter size paper,so it was easier to handle in the cramped cockpit than the manufacturer's letter-size checklist from the manual. I had it spiral bound rather than corner stapled, so that the student could more intuitively tell if he could just flip the book over to read the next page or if he had to turn the page. I also tried to order the steps in the procedure in a way that made sense--the manufacturer's preflight checklist had you check several things like "all switches off, flaps up, fuel valve open" but did it in a seemingly random order. I put them in order from the left side of the cockpit to the right side where possible to limit the "Where is it" factor of new students. With these changes in place, students learned checklist usage faster and developed better checklist usage habits. I spent less time teaching them how to read the checklist and more time teaching them how to fly the plane.

Now compare the setup process of Marlin vs Repetier. In Marin you have to download a standard set of files, if you're not using a Cartesian machine, swap out some of those files for others, then manually edit the firmware in C++ to put in the dimension settings etc. That's not how most people interact with their computer. I dodged that bullet, I was already playing with Arduino when I got into 3D printing so I had the IDE downloaded and knew how to open and upload sketches, but for most that's brand new. In Repetier on the other hand, you fill out a form on a webpage with nice, human readable text fields, drop down menus and check boxen, and it builds the file for you. It's done in a way people are used to looking at. Each page looks different so you don't mix them up. If you have to go back and change a setting, you can say "Where was that thing, here's the page it was on, Aha!" Instead of walls and walls of allcaps #DEFINE statements, seemingly half of which are called some kind of offset.

When jumping into RepRapFirmware, some things I found pretty intuitive, some things I didn't. Note I did upgrade the main firmware but never got around to updating the web interface, so I've only run what came on the board. When typing G-Code into the console and pressing the Enter key or the onscreen Send button, it would highlight the text. It often didn't give any feedback if it "took" or not. I found myself pressing the enter key, and then reaching for the mouse and pressing the onscreen send button. Most text entry fields--and indeed most G-code consoles--don't work that way. The entered text disappears and you get an echo in the log. There's a check box in the Settings tab to force notifications. That does make it reply to every code sent via the console, but it also creates a pop-up notification for every single thing, which quickly gets in the way of the UI. There was also no "history." Most such text entry boxen allow you to press the up arrow key to cycle through entered text in reverse chronological order, useful for repetitive tasks.

I've also got a makerspace full of people from diverse backgrounds who are complete newbies to 3D printing to try things on. (evil laugh, lightning strike) So yeah, I want to go in a human factors direction, make it easier to understand and package it in a way that you accidentally do it right.

As for an effector piezo sensor and such--I think I'll let you boffins do a bit more R&D before I jump on board. I use my machines for work, so tried and true beats new and cool. My delta is in there printing a copy of a motorcycle part for a customer as I type.

Edit: Clearlynotstefan, if I can't calibrate the printer, what good do all those neat features do me?

"I can upload G-code fast across the network, upgrade the firmware the same way, I can control the whole printer from my smart phone, it's got 256x microstepping so the motors are smooth and quiet, it doesn't segment delta moves or approximate the geometry like Marlin does, it's just so awesome."

"Cool, let's print a part!"

"Umm, yeah about that..."

There are a lot of cool features in there. That's why I bought one. But as I said in line 1 of my Delta printer thread, I am a doofus. If it's beyond me it's beyond me, screaming at it more won't help. And it's too much money to put in the parts bin.

Edited 1 time(s). Last edit at 03/16/2017 01:04AM by av8r1.

Okay so lets brainstorm. Might I suggest we make a list of desirable features for an easy to use 3D printer.
Lets bear in mind though that if we go all the way to ideal, we are talking about the Star Trek replicator. That's the ideal 3D printer.
So presuming we are not going to be ordering "Tea, Earl Grey - hot" anytime soon what do we want to see (I'll kick off):
-A bed/z axis that is automatically physically orthogonal to the X and Y axes
-Automatic calibration of all axes for steps/mm and where appropriate auto delta/corexy adjustments for linked axis kinematics.
-Automatic calibration of (or closed loop control) of extruders, so after a material change the e steps/mm is automatically determined by the printer (or can be trimmed by some sort of measurement system in the hotend).
-The printer is a web accessible device using passwords and up to date WiFi security, to avoid Internet of Things security risks.
-Being able to use a wide range of materials, not being limited to proprietary cartridges.
-Having a printing surface which is removable, durable, which sticks to everything and releases everything when cool.
-Is enclosed (heated if necessary) to prevent fingers being trapped/burned and to ensure large parts do not warp.
-Is filtered to ensure fumes/particulates are not emitted from the printer above acceptable low levels.

Can I also suggest we complete the list of desirable's before delving into a discussion about how to do it. Lets decide what we want first....

Edited 1 time(s). Last edit at 03/16/2017 06:47AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

A point on user friendliness, particularly as it applies to the RepRap community as a lot of the members are "Makers", "Geeks" and others who are interested in the printer, the technology, the process and the science as much as the printed object.
User friendliness aimed at new users is often perplexing to those who have previously invested a lot of time in understanding the previous way. Take Repetier Firmware for example: Inveterate tinkerers would be at the Configuration.h, Pins.h and uiconfig.h on almost a daily basis and knew where every setting was. The introduction of the Repetier-Firmware configuration tool did make initial configuration easier, but it did so at the expense of the easy ability to change settings generated before the introduction. The new Configuration.h had lost almost all of its comments, had been laid out in a new order and had even had some familiar names changed. Any changes made by hand to the Configuration.h will not be seen by the configuration tool.
Please note that I am not saying that the configuration tool is not useful, nor am I disrespecting Repetier. What I am saying is that it is worth the effort to keep continuity - after all, todays newbies will be tomorrows experienced users being faced with yet another new paradigm.
On automatic calibration, geometric correction, ABL etc.: Having had nozzles crash into beds or scrapppped across Kapton film, prints start in mid air with the nozzle somewhere between Z=5 anbd z=i, I think it is time for all of the setting procedures to be standardized, clarified and even disinfected.

Mike

As DjDemonD put in his brainstorm request while I was writing the above, might I put in a thought that much of the calibration should be more of a tool to say that something has gone out of wack, just as a paper printer warns that it is out of paper or ink or that something is jammed. This would have the thoroughly desirable effect of eliminating the common claim that "nobody needs ABL, just build it right".

... to step further into direction of the "Star Treck replicator" I'm actually starting a project to expand the accessible FDM/FFF-material to glass, ceramic and metall (essentially all materials, that can be melted with a laser or other "pointed" energy source).

Im "tinkering" with this ideas some ten years now ... and already did some "3D-fabbing" with sheets (LOM-fabbing) and micro-wires.

But now It's focussed into a comercial R&D project, so should bring results in much shorter time (the first project is targeted for next year) ... so let's see, what's possible then

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

I am quite on board with Mike's comment above that the wants/needs of the user change as their experience grows. Plus there are several types of consumer of 3D printers, Home users who want it to just print stuff and never need to open the cover, hobbyists/tinkerers who are probably quite happy with the status quo, professional users who are already quite inclined to buy turnkey (expensive) 3D printers already.

Edited 1 time(s). Last edit at 03/16/2017 08:43AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I'm sure hobbyists and tinkerers are quite happy with the status quo. We here are inherently interested in 3D printing for its own sake, so we design and build 3D printers for tinkering. But in doing so, we're dragging those home users who want to just print stuff and never open the cover along for the ride.

I've got a high school student building an FT-5 because he's a big Fallout fan and wants to print props from the game. I've got a friend building another FT-5 because he wants to print ukuleles and guitars. Another friend of mine--a mechanical engineer, no less--runs a side business making EDC/Tacticool stuff and wants to prototype his wares, but doesn't want to have to learn how to level the bed. These guys aren't interested in robotics, they could care less about learning G-Code, and they're not interested in new and exciting ways to tell that the nozzle has crashed into the bed. They want a working tool that they can make stuff with, and they don't want to spend several grand on an Ultimaker. They're begrudgingly willing to deal with the technicals because it's either that or don't have a printer.

Quoth the reprap.org homepage: "RepRap is about making self-replicating machines, and making them freely available for the benefit of everyone." I'm interested in the last half of that sentence right now, in bringing desktop manufacturing to the home in a way that doesn't require the end user to become a robotics technician. Home desktop manufacturing is a reality, but I had to learn two programming languages to do it.

The professional users, engineers, folks like that, they're served already by the likes of Stratasys, or a little lower down the ranks, by MakerBot, Ultimaker, Formlabs, etc. I'm not worried about that crowd, they've been 3D printing longer than I've been sucking air.

I'm envisioning a machine that's as easy to use as Cura. I'm not aiming for "a particularly dumb Corgi can print the Eiffel Tower." I'm aiming at "You can handle Microsoft Office, you can handle this."

I'm not so interested in "it calibrates itself" as "I can easily calibrate it." The frame should be square and straight enough to not need much in the way of auto calibration. I think the user should be able to make adjustments in a human readable, high level way. That M579 code in RRF, for instance. "I asked for a 50mm cube, I got a 49.8mm cube" is a human level 3D printing problem, and "Make it 0.4% larger" is a human-level answer.

RepRapFirmware's All G-Code, All The Time approach is great--for those who know G-Code. But all the buttons in that web GUI just send G-Code behind the scenes, right? What if the web interface got more comprehensive so that you could get it all done via GUI? The G-Code layer is still in there, ready for tinkerers or those who outgrow baby's first FDM machine. Compare Arch Linux to Linux Mint.

It should be able to detect if it's running out of filament and pause with the nozzle off the part for the user to change spools. I agree it should take standard spools. I don't think ability to tell what kind of filament is loaded is necessary, I think the user should be able to read the label and pick the matching thing out of a dropdown box in the slicer.

Quote
av8r1
I'm sure hobbyists and tinkerers are quite happy with the status quo. We here are inherently interested in 3D printing for its own sake, so we design and build 3D printers for tinkering. But in doing so, we're dragging those home users who want to just print stuff and never open the cover along for the ride..........
.............

Hmm, I seldom do irrascible, but I will do my best.
Few of the members of the RepRap community are in the 3D printer business, even fewer cover their costs and fewer still are Gods. It strikes me that you are offering neither money nor sacrificial goats. Why then do you ask for changes to hardware and software that would help the naive while stripping the tools from the hands of those bringing about the changes?
By all means point out incompatabilities. Maybe address individual items that could be improved and improve them. You could even suggest the things that you want on the respective RepRap or non-RepRap forum. You could even write or edit a page on the the RepRap Wiki.

Mike

So having a frame that's manufactured straight and true is a laudable aim. See Tiko "Unibody" Delta printer. I know they're struggling to deliver via kickstarter but the idea is sound.

As for calibrating itself, the reason I say this is that is IMO either we want it to be ridiculously simple, i.e. no user setup required, or we might as well leave things as they are.

To quote "That M579 code in RRF, for instance. "I asked for a 50mm cube, I got a 49.8mm cube" is a human level 3D printing problem, and "Make it 0.4% larger" is a human-level answer. "

There's some sense to this but I'd argue that if you are a user willing to do this calculation to solve your problem then it doesn't take much more intelligence (although a little more time admittedly) to figure out how to calibrate the printer in the conventional manner. If you're talking about a cartesian machine then it is already easy to calibrate isn't it? Feed some filament and measure it and then adjust a number by the correct proportion. But that's if you know how to heat it, insert the filament, heat the filament to the right temperature, set the idler tension, and get it to feed 50mm of filament. The Prusa I3 Mk2 can auto compensate its bed and calculate non-orthogonality of its axes and compensate for that also. That's a step in the right direction, but you can't just attach a spool and watch it analyse the filament, print a few test lines and adjust the extruder calibration automatically.

I am talking about making a truly no-user-input printer, not about just making the machine one-notch less technically demanding. That might suit your requirements, but what about my hypothetical 80 year old non-techie?

We could use encoder strips on the x,y and z axes to avoid the need to have endstops, and to feedback position information to the controller. This might even allow the use of cheap DC motors, but to be fair steppers are not expensive anymore.

As for filament measuring/monitoring that's the biggest variable. The diameter can vary, the roundness can vary. A truly zero-user-input machine will measure the cross section of the filament as it's inserted, and determine how many turns of the drive gear are required to get the right amount of extrudate out of the nozzle. The fine tuning is currently done by eye and good judgement which our user cannot be expected to have. That's an engineering challenge which so far no-one has come close to solving, but unless you want a chipped cartridge based system with tight tolerances on the filament, which means a kilogram of filament is going to be £50, not £15 then that problem has to be solved.

Nozzle auto cleaning - the machine will have to have automatic nozzle wiping/cleaning, the user isn't going to want to open it up and have a go at it with a wirebrush, that's not necessary on my inkjet printer, which is a consumer easy-to-use device.

I think for the machine we are thinking about, the web control would have to take the user by the nose through what they have to do, inserting the filament etc... So that would only be applicable to the one specific printer. Duetwifi is sold by enthusiasts to enthusiasts (and OEM's) who happily learn the gcodes they need, or who will customise it for their customers requirements.

Making the machine a delta would be asking for trouble, I think they're great but they rely on precise mechanical build and they have to stand up to transportation without being knocked further out of whack than the automatic calibration mechanism can cope with. Making a cartesian machine out of repeating units of motion-stages like the Trinus is a good idea. It simplifies the design and standardises the parts.

I have a feeling this discussion will run and run. In reality I think things are evolving slowly towards the goal rather than making huge leaps forward. Manually calibrating a delta isn't that hard, but auto-calibrating it (with the right sensor) is an amazing leap forwards.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

How available are the components to do a closed loop system?

Quote
DjDemonD
As for filament measuring/monitoring that's the biggest variable. The diameter can vary, the roundness can vary. A truly zero-user-input machine will measure the cross section of the filament as it's inserted, and determine how many turns of the drive gear are required to get the right amount of extrudate out of the nozzle.

Making this truly zero input would require smarts to know the current nozzle size as well. It would be frickin awesome though.

One thing to note, there are what I'd call idiot proof 3D printers, but they are far from cheap. My old company had such a printer. It was a Stratsys uPrint SE machine. It was great, it would print anything and everything you threw at it primarily because it ran material and support material all the time. It was deadly accurate too, but used they're software.

Now the downside, it was expensive. The printer itself was $15,000 and was take it out of the box and run a little calibration routine and enter 2 values from the printed calibration. From that point, it was ready to print. However, it required a plate to print on which is how they guaranteed the print would stick without lifting mid print. These plates were roughly $5 a plate and were a single use only. It also had idiot proof material cartridges that were $200-$210 each and would only run ABS plus. It also required a solvent bath that would dissolve the support that was also in the $10-15,000 range in cost plus concentrate material.

Your $3000 kit wouldn't even come close to these idiot proof professional 3D printers, and I've yet to see a 3D printer kit that compares. Even the makerbot lines have issues with printing and calibration. The home 3D printer sounds great, but I don't think it's as close as people thing. I give it at least 5-10 years before we're at that point for $3000 or less. Even the kick starters aren't working as advertised with regards to that Tiko.

I think this is what I'm getting at really. All of these ideas will become part of enthusiast machines gradually in the same way autolevelling, autocalibration and other such innovations have. Eventually much of what we do now by skill will be automated functions.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
briangilbert

Quote
DjDemonD
As for filament measuring/monitoring that's the biggest variable. The diameter can vary, the roundness can vary. A truly zero-user-input machine will measure the cross section of the filament as it's inserted, and determine how many turns of the drive gear are required to get the right amount of extrudate out of the nozzle.

Making this truly zero input would require smarts to know the current nozzle size as well. It would be frickin awesome though.

I have got one idea, have a glass section of the print bed, which would have a scanner type cmos sensor under it. Then after autolevelling it would lay down some filament and measure its thickness and length. The resolution of a scanner should be enough to measure the width and thickness of the extruded filament and presuming you trust your levelling you've got the volume.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

What a nice discussion!
In my mind there are 3 types of 3d printer users.
1 Business, it needs to print always, when I need it to with perfect quality
2 Consumer, who wants to print stuff. It needs to print whenever needed, but quality is less of an issue
3 Hobbist, who want to tinker with 3d printers, build em upgrade em and mod them.

1st group, already covered with some expensive systems out there
3rd group, hey we are all here right?

2nd group, that is where the pain is right now. You need a printer that will print, every time. That does not need to be upgraded and does not need a lot of maintenance. That understands if you put PLA in there or ABS. That knows that you switched nozzles because the whole head is a cartridge with a chip in there. That knows that if you use ABS, it needs to warm up the chamber a bit first. Those kind of things.
It boils down to a printer that is used by the 1st group, but at a price the 2nd group can afford.

Afaik, the Prusa i3 mark2 comes pretty close to all this, except for the chamber. Who needs switchable nozzles anyway, at this level?

The costs for all these "business" features are coming down, more affordable smart control boards are coming out, there is more processing power available to add smart functionality. The new idea of the PCB effector allows smart effectors in the future, all little things that will really make printers shine. We ourselves are working on a smart complete printer that includes a lot of the "business" features, trying to get it to a price where consumers will want to buy it.

In my mind, there are 2 issues that still need to be tackled for the 2nd group of people. 1 is software and 2 is hardware

Software, we need more user friendly software control. The web interface for the DuetWifi is great, but compare that to a inkjet printer and it is horribly complex. It would be great to develop an interface that is clean, simple and only dives into complexity when it is needed.

Hardware, there is one thing still missing for the perfect printer and that is the perfect print surface. A surface that is universal for all types of materials, is forgiving in the 1st layer and will release relatively quickly. Simply does not exist yet.

I am still amazed how quickly things are improving, especially the last couple of years, the quality and the functionality of standard printers has made great strides and I am proud to be a small part of that industry.

---

Lykle
________________________________________________

Co-creator of the Zesty Nimble, worlds lightest Direct Drive extruder.
[zesty.tech]

Computer:
Print me Benchy

Sorry Dave I can't do that
that is 50 years in future!

Perhaps a DaVinci Jr ? Mini

[us.xyzprinting.com]

I think one of the other factors, going back to Lykle's quite accurate description of the market, the consumer demand for 3d printers just isn't there. Pros like them and enthusiasts like them but my Dad just doesn't want to buy one. He's slightly curious about mine, but even with a background in the automotive engineering industry a while back, he doesn't have any interest in how it works.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

What the "3D printer in the home" revolution needs is a product. The "Computer in the home" was partly driven by the internet and mail which were sort of expected, but it really exploded with the arrival of the camera on every phone - when neither the mobile phone in every pocket nor the camera on every mobile phone were expected.
Now while it would be nice to make the home 3D printer completely reliable, what is really really needed is to think of what to print with the damn things.

Mike

From what I understand, a closed loop movement system would fix any calibration issues for a cartesian machine. It can run a self-calibration just by running the axes back and forth a time or two at startup or something. That would eliminate the need for the end user to print and measure.

For something like a cartesian, I think you could trust the build of the machine to be square. A buddy of mine bought a Prusa MkII, and the way he says you square the gantry is hilarious: you crash the gantry at the top of its run, and it's square. So we have the gantry running accurately in 3D space, now all you've got to do is level the bed.

Telling what kind of filament--I don't think that needs to be automatic. When I load paper into my epson, it asks what kind I just put in, size and type. I have to key in "letter, 8.5x11" or "Envelope, 11x4". I'm going to trust the end user to read "PLA" on the spool and select "PLA" from a list. That usually happens in the slicer rather than on the printer, anyway.

Filament diameter--I'll be honest, I've got it set to 1.75 and I leave it there. I only take calipers to my filament with a new spool from an untrusted manufacturer. I usually use Atomic, MakeShaper or Taulman filaments, and they tend to be high quality and dimensionally accurate. However, a filament diameter sensor before the extruder gears isn't a bad idea, and it can double as an out-of-filament sensor. I would recommend using hardened steel nozzles rather than brass--longer lasting

I think continued development in slicers will help out in this area. None of them really do printer or material profiles in a straightforward intuitive way, I gave up on Slic3r because I couldn't determine how that worked, whether printer, print quality and material profiles were dependent or not. They're not so good at detecting overhangs or lines over nothing (for instance, Simplify3D tried printing the bottom of a hole entirely over the gap in infill) They tend not to make elaborate decisions about supporting the structure above, or layer cooling. That "It also used their software" when talking about a Stratasys says something. Also, I don't think any of them come with manuals. I KNOW Simplify3D doesn't, I asked. Side note: I started into technical fields at 13 when I took my first shop class, and I've had RTFM yelled at me daily since. I feel the need to say back "I will RTFM as soon as you WTFM."

I agree with Mike, there isn't one big killer app for home FDM yet. To print, you need a model. Where do you get models? Four places: CAD software, Sculpting software, 3D scanners, or someone else that has one of the above. Intuitive modeling software is a discussion for another time, and probably another group of boffins. Having just built a FreeLSS scanner, I can say that might be a technology we just start building into 3D printers, it goes together so well. Then there's Thingiverse. Thingiverse is a good idea that wasn't implemented quite right, but the pain in the neck force will straighten that out.