Jump to content

mmartin

Dormant
  • Content Count

    12
  • Joined

  • Last visited

    Never

Community Reputation

0 Neutral
  1. For Alibre users: 3D systems will discontinue the smaller versions (Elements, Pro) 20 April 2015, only the top version survives. Might be a final chance to get a elements version (if this is still available somewhere). If you live in Switzerland, http://www.alibre.ch/news/pro2expv17.php is an upgrade offer. For non-pros: Consider Cubify Design.
  2. The tilting print head idea has been around for a while, also in a Stratasys patent (see earlier post). As UM class machines suffer from the lack of high precision machining (in order to keep cost low), it makes sense to use elements with high precision/cost ratio. A ball bearing probably is the best performing construction element here, much better than self-made linear bearings. However, your sketch shows how to maximize nozzle inaccuracy for a given bearing play - the nozzles are too far away from the bearing contact points. Move the nozzle ends as close to the X bearings as possible to minimize the effect. This probably also means moving the tilt axis above the X rail. The question is if a tilting head is really necessary (the Dimension Elite works with fixed nozzles, larger layer thickness though). If nozzle collisions are caused by too much play on the print head, one might question if adding even more potential accuracy issues is the way to go. And of course you add the complication of a tilt motor. Not too difficult as there are two fixed stops, but still. It could be stationary as you'd want to go to a dump station when changing material anyway (BTW is this waste bin patented? The "purge tower" idea is IIRC). And of course you need the two extruder motors, making the print head heavy... And before you invent a single motor dual extruder head, look at the Stratasys patent describing this idea.
  3. Geomagic design (got the medium license when it still was Alibre design). There is a stripped down version of Alibre which is called "Cubify Invent". It sells for $39 and has a free trial version. If you are looking for a mechanical CAD in the style of Pro/E or solidworks, I'd try this. Not very good for organic shapes though. EDIT: Just saw this has been recommended already. Cubify Invent does not support assemblies, which makes it more or less worthless for mechanical design. Cubify Design (€149) does.
  4. AFAIK the safest way to establish prior art is send the application in to the patent office, then later not pay the fees (let it expire), as you explained in a prior post. You don't gain protection, but nobody can patent your idea either at the novelty is not given any more. You could have done this with your ideas, even if you think patents and UMs open source strategy don't mix. Otherwise you live on the remains all other players leave in the field. Wikipedia article on prior art - little chances for internet sources.
  5. Tools: I found the patent list a bit depressing. There's tons of great ideas, but none will make it into an UM! So here some ideas on what can be done, specifically what tools are worth using in development. I hope this does not go too far OT. With any motion system, the question is how closely the "tool" follows the path you commanded. For an UM style layout, it's the position of the nozzle via the topmost layer of the print on the build platform. All mechanical systems show play and vibration. If after a move, the build platform wiggles too much, it might collide with a (2nd?) nozzle and rip of the print, or it might give you zigzag surface patterns. Generally industrial sensors are expensive, but an engineer costs a firm easily a six-digit figure per year. It's an easy calculation how much time he can spend on building measurement tools himselv vs. buying them off-the-shelf (making such stuff is often apprentice work - they have fun and need practise projects anyway). High speed camera:Together with a macro lens a great tool for seeing what's going on. Some compact cameras offer super-slo-mo and there's a startup trying to lower costs. Expensive otherwise. Capacitive distance sensors: From Micro-Epsilon and such. Typically you measure settling to a fixed positon, connected to a scope. If it must be DIY I'd look at CDCs from Analog. Also an option for DIY are slotted photointerrupters. Depending on beam width, you get an OK resolution, but only a very short useful range. Acceleration sensors: Have the advantage they need no counterpart, but can't measure relative movement. Brüel&Kjaer is the Rolls here, but there's tons of MEMS sensors from smartphones that might work as well. Note that double integration of the acceleration gives the displacement. FEM (modal and static): Finite element is the method of choice for analyzing stiff structures. I guess an FDM printer is a poor application as it mostly consists of bearings and drive belts - lots of play, generally nonlinear and unknown stiffness parameters. Motors: Open loop stepper drive is rather easy and seems to be no problem. If you still think you are using steps, you can connect a logic analyser to the step signals (like a cheap saleae logic), record a ten hour print and then write a python script to sift through the data and check for glitches or too high acceleration. For debugging, you can even connect a high-res encoder to the motor in doubt (Nanotec provides this as an option) and log those signals with the logic analyzer as well. You can then measure the actual motion of the motor and compare it to the step signals. Simulation: For anything more complicated than open loop profile generation, Matlab/Simulink is invaluable. If you want to get started with BLDC FOC for example, it's great to play through in simulation. You can simulate play, encoder resolution, sampling time or whatever and see how the final system would work even before starting it. Consider this question as an example: Let's assume we want to place the extruder motor into the print head. As it has to be lightweight, we use a DC motor (say from Maxon if we ignore the cost) with a multi-stage planetary gear. As it is a DC motor, we need a closed loop control. Where do we place the encoder? At the motor shaft - the gear has play so we don't know exactly what the output does. At the extruder wheel? This might be problematic control-system wise because of the play... Having mentioned " again). I wrote Stratasys went the straightforward route with the Dimension Elite. They made the system stiff enough for the required dynamics, ending up with a 120 kg machine.
  6. Patents: I'd like to encourage you to look out for interesting patents. I'll give a short introduction - I'm no expert myself (in the industry, you always need a patent attorney, it's sufficient if he knows the fine print), feel free to add your own suggestions. Reading patents can be fun. They are a treasure trove full with ideas that were deemed working and bring some benefit. Although some might seem cryptic at first, this is not intentionally so. In fact every patent must be a blueprint from which the average but non-inventive ("skilled in the art" is the legal term) Joe Engineer can build your invention. How are these finds relevant? Of course it means no commercial machine can use them, so you can trim down your UM3 suggestion list. But noone prevents you from experimenting yourself, create a single motor dual extruder head, a tip cleaner, a filament box with desiccant... An interesting question (again) is how far one could "approach" patents. Let's consider the heated build chamber. UM can't sell this because Stratasys holds the patent. But UM could prepare the machine for it - make everything inside temperature-proof, add sensors and relay outputs for the heater, maybe even (as accessory) a hood in Makerbot Replicator 2x style. Fans, sensors and heater would have to come from a different source (Farnell, Digikey or an affiliated reseller). This would be no direct patent violation, but might be asking for legal trouble. Or the tip cleaner: Provide mounting holes and a check box for path generation in Cura, but leave the rest to the customer... My favourite access is the European Patent Office "Espacenet" patent search (just type "espacenet" in google). Go to the Advanced search and type "stratasys" in the Applicant field. In the worldwide data base, this gives you 515 patents. Many are duplicates in different countries ("patent families"). You can narrow this down by entering "US" in the publication number field, but some US patents don't have the full data in the database - try a "also published as" patent, preferably an european (EP...) one. The quickest way to get an overview over mechanisms is by looking at the drawings. From the patent list, hit a few times "load more results for export" and press "Download covers". This give you a list with all covers you can flip through. Also, when you opened a patent from the result list, select "mosaics" to get a overview over the drawings. You can press "next" on this page to flip quickly through the result list. It's a bit hit and miss. Keyword search might work or not because for example Stratasys speak is different - "liquefier" means hotend for example. Interesting patents: See the list at Scott Crump's Wikipedia article Purge towers Head tool changer Single motor dual extrusion head (nice) How to handle moisture sensitive material Tip cleaner Other finds?
  7. Build platform: The manual says replace it regularly, but in practise, it lasts a long time (not only printing twice). The solvent for the soluble support is water based, not water. But not limonene either (which I find nasty). Anyway I think if it is saturated with plastic, it needs to be disposed of properly (not by pouring into the drain), so even if the solvent would be pure water, I wonder if you could pour the saturated solution simply into the drain. Actually, it is much more difficult to build a €2k printer instead of a €50k printer. Only lightweight parts, low precision components, no fancy manufacturing. So one of the very interesting questions is: Can a top quality printer be built for €3k? (I mean with support material and out-of-the-box reliable printing) Stratasys went the easy way for some design decisions. The Dimension Elite print head holds two massive extruder motors and a huge X drive stepper motor. I guess it's over 1 kg. This will cause high reaction forces on the base with vibration transmitted to the build platform. So they added a 100 kg base, the 1:100 ratio reducing the reaction displacement by 100. This certainly works, but it is kind of a sledgehammer design. Interesting question: Will top quality printing ever work with remote extruder motors (not at the print head)? I think no as material control is too loose. The smarter way (talking of the Dimension Elite) would be to reduce print head weight. Use one instead of two extruder motors (there is a nice Stratasys patent on this) and have a stationary switching mechanism. Move the X motor onto the Y rails. Use smaller, higher geared (to improve motor efficiency) motors for the extruder. Next interesting question: Will sticking to open loop stepper motor drive be sufficient? For XYZ it seems so. But for the extruder motor inside the print head, the power density of stepper motors is pretty low. A geared (BL)DC motor would be better here, requiring an encoder and closed loop motion control. But this might get expensive. Maybe a small, highly geared miniature stepper motor with 48V drive (or higher) might do as well. A note on closed vs. open loop motion: Stepper motors can be reliably driven "open loop", which means there is no position feedback except for a reference switch. This is done by massively ovedriving the motor with a regulated current that results in a torque being much higher than the torque ever present in the driven system. Together with a simple linear accceleration profile, this is cheap, computationally simple and reliable. The drawback is the motor is huge for the usable torque and the slow speed, resulting in a low power output at high weight. EDIT: The main problem is that forces in the system might be higher than planned. Dirt on a bearing, a misaligned axis... The other issue is that if something goes wrong with the step pattern generation (interrupt during move, software bug) and the resulting acceleration would exceed the available torque. In both cases you lose a step, possibly accumulating over time. So steppers are fine if everything goes as planned. You can add encoders to make them more robust, but this negates the cost advantage. Torque at a given motor size is more or less limited by the magnetic properties of stator and magnet material, so the only way to increase power output is to increase speed. This is what DC motors do, running at up to 100000 rpm. This usually means gears (unless you want to build a dentist drill). In additon, a DC motor needs closed loop position control (which means an encoder for position measurement and typically a PID position controller). The advantage is that you can use all coil current for moving your system, not having to overdrive the motor so it doesn't miss a step. Nowadays DC mostly means BLDC (or, close, "servo motor"), where the brushed commutation is replaced by a digital commutation, either six-step or field oriented control, depending on requirements. The bottom line is you get higher dynamic motion, higher power density for higher sensor cost (encoder) and higher computational complexity (PID position control, PI current control, FOC commutation, jerk limited trajectory generation). I am with Jonny here: When can we get rid of the pathetic Arduino platform for control? Closed loop motion control (maybe FOC BLDC, maybe vibration cancelling) requires more computational power. I don't see a reason why a modern control system design does not use a Cortex M4F, take an STM32F4 for example (180 MHz single cycle floating point multiply...). On the other hand, for simple trapezoidal open loop stepper motor control, even a poor 8 bit controller might do. Coming back to print head weight, UM2 is a clever design, in its way: By having the heavy extruder motor stationary, the print head is very lightweight and an enables the use of small diameter bearings and low volume/weight base parts, all reducing cost and keeping high tool path fidelity. Another question: Will two fixed nozzles work? The Dimension Elite says yes, but it is built like a tank (reducing vibration), has a layer height of 0.18 mm (not 0.1), meaning increased spacing and a heated build volume, reducing warpage. Certainly placing the nozzles much closer to each other will help - why not 5 mm apart? This will reduce tilt sensitivity. Maybe given higher tolerances, more warping (unheated), more vibration, lower layer heights of a low cost printer will just not work without a nozzle lift. But this adds cost and weight and reduces accuracy. And when we're at it: Will an open material (as opposed to fixed, maunfacturer provided) printer ever work reliably? I think no. The obvious reason is that for every material, the print parameters need to be tuned. But much more severe is the fact that maybe no pamameter set exists for making the mechanism work 100%. Remember what everybody writes after a failed print: "I probably need to tune the parameters a bit.". With a fixed material, the manufacturer would be forced to provide a working parameter set - and perhaps discover that no set perfectly works. Maybe other deficiencies exist, too much play in the drive system, too loose temperature control, too much material shrinkage... With an open material choice, this will hardly happen, as all deficiencies are masked by material variations. And here is the dilemma: Imagine the UM3 came with a closed cartridge (with desiccant and all), chip protected. I'd accept that (up to say double the material cost), provided prints work perfect with this. But I guess the community here would sink UM in a shitstorm.
  8. Fair or not fair is irrelevant IMO. The relevant question is: What can be learnt from these machines? There are many design considerations that are worth questioning, on both sides.
  9. I decided to buy an UM2 once a working (soluble) support material solution was there. It seems this is not going to happen! I regularly see a Dimension Elite printing. It is amazing to see how effortless and without any problem this thing works. It "just prints". No delamination, stringing, dripping, oozing, warping, clogging, ringing. As many are wondering how a soluble support material solution could be designed (which I think makes it a tool instead of a toy), it is worth looking at how for example this specimen does it - an almost a decade old design! Maybe others can share their obervations as well. Of course this Stratasys machine is in a different league, but I don't see a natural law prohibiting sub-$10k machines from proper printing. I don't own this machine, I have no schematics or internal documents. I read the manual, watched maintenance videos, observed it when printing, read some patents and talked to the operator. So everything is AFAIK or IMO, if you know better, feel free to correct me! Find the manual here. The machine prints 0.178 mm layers from two materials - ABS and soluble support. Layout is pretty standard, platform is Z move only, print head moves X/Y. Print quality is excellent, limited by the 0.178 mm layer thickness. Material: Comes in cartridges (CHF 550 here, approx, EUR 500 for a 0.9 l cartridge, build and support each). Cartridges are hermetically sealed with locking pin for transport. They contain a desiccant and must not be opened; the machine keeps track on how much material is used by an EEPROM on the cartridge (maybe also copy-protection). There seems to be a feeder motor directly at the cartridge. Material lifetime in cartridge is 30 days, when cartridge case is broken (eg. when the end of the filament slipped inside or you are overly curious), it must be used within days. There is one single material you can use (different colors though). The soluble support mostly breaks off easily, the rest is removed in the water based solvent in a heated ultrasonic cleaner. The machine never prints build material on the base, it always starts with a layer of support material. The machine prints on a non-heated, removable plastic base. The build chamber is closed and heated (patented). There are no tilt adjustments for the build platform. In general, the machine is very solidly built. It weighs 120 kg, everything looks precisely machined, no play on any bearing, no user adjustment screws. Z move is via a ball screw, X/Y move via stepper motors and belt drives. The print head looks quite heavy, even the X stepper motor is located on the print head. Search videos on "dimension elite maintenance", you'll find a couple of close-ups on the print head. The print head has two fixed nozzles quite close to each other (10 mm?). It has two feeder motors - geared brushed (?) DC motors from Faulhaber with an additional reduction gear, located just before the heated nozzles. The feeder wheels look quite solid and seem to make deep notches in the material - no slipping here. There is a fan in front of the heaters. Strangely the material makes a 90° bend from horizontal feeding to vertical extrusion - the designers seemed to want to keep the height low. Because of the 90° bend you can't clean the nozzles, the specimen I watched needed a replacement (the complete head was changed) after a few years as one nozzle started clogging (not bad for mostly nonstop printing). There is a scraper/metal brush head cleaning (also patented IIRC) station at the rear. Behind it is a waste bin. After each material change (at each layer when printing with support, which is 90% of the time) the printhead goes over the cleaner and extrudes a bit in the waste bin. Quick summary on the most interesting points: - heated, closed build chamber, non-heated platform - closely controlled quality, dry material out of cartridge - print head cleaning station with waste bin - geared DC extruder motors - higher power density, much higher torque than stepper motors - extruder motors directly at the nozzle - more precise material control than with remote feeder via bowden tube - very close spacing between nozzles, nozzles fixed - extremely solid, heavy build quality and: It just prints. No webcam for supervision, load the file and get the model hours later. Any other ideas? Anyone else observing other types of printers? In addition, there are some quite interesting ideas in Stratasys patents - another time. - Martin
  10. Same here - my status is still "validating" after a couple of weeks. When I press the "resend validation" link, I get an email, if I follow the link in there, I get: Oops! Something went wrong! [#10121] You have completed your portion of the validation request already. An administrator must approve your account before you have full access. So how can I get admin approval for my account? Thanks!
  11. Hmm, I could start this topic, but not in the "dual extrusion kit" area. In my profile, it says: "Group: validating". I checked my spam folders, but did not find any email in there - but it might have been deleted because I registered a while ago, but never got a confirmation that it would have been approved by a moderator (which the registration process said would be necessary). I don't see a 'validata' button in the profile either... Anyway, would there be a specific reason why I could not post to the dual extrusion forum?
  12. ... or "you cannot reply to this topic" is something I've seen a couple of times here. Could someone please enlighten me about the posting policy for new users?
×
×
  • Create New...