Fighting with computers

I have been losing a fight with my latest extuder/hotend combination. It works ok as far as I do not want it to work too fast. I want to keep the PG35L motor not too hot or it will break. So I am powering at 500mA, which seems to be well beyond its maximum rated current, so it gets hot. But I've added a passive cooler to it so won't get dangerously hot. Should I used a higher current, as someone suggested, maybe my mileage might vary, but at the expense of needing a fan blowing directly to keep the motor from frying itself. One of the things that was bothering me was the maximum printing speed I could use with this new setup on my Prusa i3. Extrusion speed on Pronterface software (the host I use) is user selectable but measured in millimeters per minute, while travel speed is usually configured in the slicing software in millimeters per second. But doing the apparently simple math of dividing by 60 will not help here, among other things because the extrusion speed is the 3m

I am very happy with the Prusa I built, and so I am with the Printrbot, which much smaller but it does a nice job too. In fact, most of the parts (on different shades of blue) of my new Prusa i3 have been printed with the Printrbot, but the black ones, which are ABS parts, have been printed with my Prusa. Having two working 3D printers seems to leave you in a difficult position to explain your wife why in hell you're building yet another one, but somehow I managed. I was curious to learn how this new version from Josef Prusa would do compared to what I was familiar with. I've done a short picture album of the process with some comments in them whenever something different or creative was done there. My first impressions are that this printer can work really fast and it seems very stable at high speeds. It feels much sturdier than the previous Prusa's and more stable than Printrbot. To its advantage has the reduced mass of the new compact extruder which I have not mas

A few days ago I learned in the RepRap forum about a new stepper driver from Pololu using the DRV8825 from Texas Instruments. I have read good things about this family of drivers being more robust and higher current than Allegro A4988. Well, thanks to the kind offer from Pololu I've got a few units for testing (at a huge discount) and my first impression is pretty good. This may well be the answer to get a direct drive extruder that does not suck. I have detailed in past posts how I was unable to get a successful direct drive extruder working in my system, as I was needing more current that my electronics were capable of, for my extruder to work reliably at medium speeds. This new driver seems to be able to push 1.5A without the need of active cooling. The manufacturer claims that up to 2.5A per coil are possible with proper cooling. Next stop is to use them for powering a CNC machine with beefier nema23 motors.

When I saw this summer that Texas Instruments was giving away ($4.99 shipped is a steal) Stellaris boards sampling their Cortex M4 with floating point unit I thought I could not go wrong, as it was already cheaper than any other kit I have bought in my life. After a while I've received a big box at home, than contained a smaller box, that eventually had the small card and USB cable inside. The first impression was very good. But once I've got the board up and running I wanted to start making some sample programs for testing its capabilities. There is where I felt a bit dizzy. Texas lists a lot of tools to be used, some can be downloaded from their site, others are free versions of commercial products with a limited functionality. I guess  that these years using Arduino have spoiled me, as  I have been able to complete different projects without having to have a look at any AVR datasheet. Even worse, I get used to the idea I could be developing code in different platforms (L

I have just finished a ShapeOko but because I did some mods: I've got some leftovers: some MXL belting and two 5mm MXL pulleys. What best thing to do than to start a project so they get some use. So I planned on building a new Prusa i3 3D printer. Not that I need yet another printer but a new model to try it out. I built a Printrbot earlier this year to get a smaller model than the Prusa for home use. At the time the printer dimensions were not very clear and much guess work was done on my side. Besides, I used a direct drive that worked but I was never happy with. This time I am, again, doing some guess work for Prusa i3, like figuring out that z-axis rods are M5 threaded rods or taking the overall dimensions out of the DXF of the aluminium frame off the Prusa i3 github , while learning that some changes have been done and are not yet into the master branch. As I was planning on using MXL belts instead of what apparently it is used in Prusa i3, which my guess work suggest is

I have been dealing with a large size CNC project for a while. I thought a smaller version come come in handy at home. So I decided it will be a gift for my wife. I bought a ShapeOko from Inventables (Chicago, IL) with the fear that customs might be expensive. I ordered just the metal kit which was a steal at $199. It is shame that I ended up paying $90 for customs. When I mentioned to Zack (from Inventables) he was shocked too. Anyway, for those of you that have never heard of ShapeOko let's say it is an open hardware project for a small CNC milling machine based on another open hardware project called Makerslide . The latter is a linear movement guide made of aluminium extrusion. I am almost done building the thing and it only took a few hours. The kit is well packaged and it is easy to build following the wiki instructions. However there are several improvements that I may already include in my build: dual Y-axis motors, nema23 motors for X and Y axis and an alternate r

Over the last seven years I have published six articles on different topics on Circuit Cellar magazine. It is the only magazine I keep on buying myself at home whatever that means. This summer I was approached by one of their editors about the possibility of running an interview. Being my first time, I wondered why other people might be interested on what I could say, I am not that interesting. But I thought it won't cause me any harm either (I honestly hope so) so I agreed. The process was quite painless and you can read an abridged version here . Given the fact that I have not thought I have accomplished anything specially great, I guess the only meaning of that is that I am growing older. On the other hand, there is some ego boosting associated with the episode I cannot deny either. If only my employer cared about it a bit ...

I've got a new computer in the office and it's a Windows 7 system (not my choice). I did all the required dance to enable Wake-On-Lan, both in the UEFI bios and on Windows but still my Remote Desktop did not happen. Nmap reported port 3389 was filtered but Windows Firewall was disabled. It turns out that Kaspersky Endpoint Security 8 (which was installed) does have its own Firewall, which was activated in my setup, and apparently does not care whether you activate Remote Desktop in Windows or not. You have to go to its management app: Configuration/Antivirus protection/Firewall and click the packet filtering rules to enable TCP port 3389. I wish they either warn you when they see Remote Desktop is activated and there is a rule preventing it to work or to just self-adjust to that change. But it is clear that Antivirus software real task is just make the life of Windows users miserable, isn't it?

One of the cool features of USB is that using Human Interface Device (HID) you can get a new device to behave as, for example, a keyboard. You can do this now with a Teensy or an Arduino UNO (but not the FTDI-based older Arduino boards). The key is that by using the proper USB messages a device can talk to the system in a way that looks like a mouse, a keyboard, etc. The good thing about talking like a keyboard is that applications that expect a keyboard input may be fed directly with the data from that device. This many times the case with devices like a barcode reader, that can input the scanned code to the application software as if the user was typing that number on the corresponding box of the program, which is convenient just in case the code is unreadable for any reason by the reader but not by the user (or if the barcode scanner is broken or stolen). My current project required reading RFID codes and for that I already bought a RFID reader with I2C interface that could be

When one of my kids told me his Kindle had stopped working I thought it was his fault. On closer inspection I could not see any damage to the screen. None of the suggestions of the troubleshooting guide worked so things did not look good. After some reading on Amazon forums it was clear I was not alone and other customers had experience a similar problem that required a replacement of their unit. I was very happy to learn my unit was less than a year old, so I contacted customer service to see what to do next. I had bought the unit while being in the states but I was now living in Europe, so I was uncertain about my choices. Once I explained the problem I was having with the unit, to a very nice representative named Stacey, I was told unit had to be replaced. I was told a replacement unit has been sent to my address in Europe and paid with a gift voucher. I was told that using that same box I should return the defective unit once I've got the replacement. I was even provi

The initial idea was to slice 3D models into a sequence of thin slices. Once that was working it was suggested that thicker slices may reduce the total machining time. The problem is that vertical cuts for the slices will create an ugly rendering of the desired surface that will later need to be fixed by an artist. The next idea was to put together the milling action for each slice sides and the nesting algorithm that will distribute the slices into the available stock sheets. As usual it is easier to say than to do, and though it seems Aspire and Rhinoceros allow you to do a similar process, my code is now performing both tasks without user intervention. Blue or green color on the outline represent whether the slice has to be flipped or not (each slice has to be machined from top to bottom).

There is always a better way, in my case, the better way was to get rid of my crappy implementation and to let others do the work for polygon buffering. And trust me, I tried to patch my simple implementation and next I tried to use the better but still not error-proof JavaGeom one. I ended up using the fine implementation of Java Topology Suite, that worked like a charm (once I managed to understand the subtle details of how things work over there).  So now you can see how the gray color lines are an accurate offset of blue polygons in my software. It did not hurt that JTS is a pure-Java library, so my tool will be multi-platform the easy way. I learned along the process that though SWIG framework does include Java bindings for CGAL library, it does not provide full access to all library functions and, once again, polygon buffer seems to be one of the parts missing. Besides, CGAL install is platform specific (though CGAL is still a terrific library). Along the process I have learn

I should have settled with a graphics library of some sort, but I did not. So now I have all sorts of problems some of these libraries have one function call away. Unfortunately many library creators do expect to be paid if you do a commercial product using their code. I find this to be fair, but not knowing how much or for how long just stir me away. My current problem is a simple one: I have collection of 2D polygons and I want to obtain the union of all of them. Sometimes it is a bigger polygon, sometimes a set of disconnected areas. It turns out that non-convex polygons make the solution harder than I expected. As I am writing code in Processing (Java) I could used JavaGeom , but it is based in GPC library . I could use JTS library , but I was lost in the process of learning how to integrate it with my code, but this was before I learned the problem took me longer to get it coded than learning another library. You have been warned! Update: If you can settle with integer coor

The project I am working on at the moment requires solving several computer graphics problems. I mentioned a while ago the polygon offsetting problem. Today my problem is about placing polygons on pages to be printed. I want to make the listing as short as possible, thereby packing the polygons in fewer pages is desired. Trying not to reinvent the wheel I googled around first. Of course the first problem of the googling thing is to figure out what is the popular name of the problem you are interested on (and believe me, I am yet to find an "original" problem never thought and wrote about it before). My first searches suggested Cutting stock problem , but  it was not exactly my case. It is a pity as in 2007 a dynamic programming solution was published . So moving on I found the best fit to be the Bin packing problem .  Again, an NP hard problem. Wikipedia article suggested the heuristic of the first fit algorithm , that is known to provide good (but not optimal) solution

While I am not a king (inside joke for Spaniards only) I do make mistakes quite often. One of them was not to disconnect my RAMPS-controlled 3D printer from the computer and mains supply for a quick solder job (just one pin of heating resistor was loose just before starting a print ...). Unfortunately, my trustry JBC solder of the last twenty years was not properly isolated anymore and I had a big flash and all the house lights went down. When I restored the power the computer was booting up happily, the printer power supply light was on but the printer was dead and the computer was unable to detect the printer anymore. All the lights on the board and on the Arduino Mega were off too and my system shown a warning message that USB port was disconnected for excessive power consumption: no good news at all, but at least I learned the USB port was not completely dead. I removed the RAMPS board from the Arduino Mega, connected it to the USB cable and then the lights come back on, surpri

Last week I have been researching about a task I initially did not even know the name: Polygon offsetting. A polygon offset (in case you ignored it too) is a computer graphics primitive (though it can be done by hand too) that given a polygon will trace a inner or outer version of it that it is either totally contained inside the polygon or that it will contain it entirely (if you want an outer offset) whose perimeter will be at a constant distance from the perimeter of the initial polygon. And given that I seriously doubt that my explanation is good enough, here you have an image (of an inner offset, gray-line perimeter): I was looking for an available implementation of this process, as I wanted to be able to use it for an ongoing project. This time Google did not helped much and though I was able to find the question I borrowed the picture above from and that CGAL library did have polygon offset as one of the functions or that there was Clipper library or a Python implement

A recent project required acceleration measurements to study a person's gait. I thought this could be done easily with one of the many MEMS (MicroElectroMechanical Systems) devices available in the market under the name of IMU (Intertial Measurement Units) or just accelerometers. I settled with the cheapest acceloremeter I found on eBay with an I2C interface to connect it to a Funnel IO powered by a rechargeable Lithium battery. Funnel IO (or FIO in short) is an Arduino-like board that includes a USB-powered battery charger plus an XBee module interface on board. It can be programmed wirelessly but this proved to be a real PITA to get it working (and even after that, it does not work reliably). Once I got some code grabbed from Arduino forums to read my sensor (MMA7660FC from Freescale) I was a bit shocked with the results. My sensor was a 3-axis accelerometer and  I was expecting a measurement of X=Y=Z=0 on my first test with the sensor static. I was wrong. I should have rea

Last week I had a "USB cable incident" with my Prusa (computer and printers are on different tables in my office). It was entirely my fault as I walked across the USB cable and I almost threw my printer to the floor. As a result, the ferrite torus I put in the USB cable to prevent EMI problems broke. Without it, I know prints may interrupt anytime due to interference. While on eBay looking for a replacement, I realized that a Bluetooth module was costing less than $15 now. So I thought I could give it a chance to the idea of operating my printers wirelessly.   Later, I learned of an even cheaper module at $6.60 , but this one is 3.3v only (which is ok if you are using this voltage for your Arduino, but I am using 5v). A quick look at Sprinter firmware and Arduino Mega and RAMPS board specs revealed that Serial2 port using 16 (TX) and 17 (RX) pins was easily available on the ext4 connector of RAMPS board. And so was GND and +5V. Making it very easy to connect the Bluetoot

This time I am interested on a particular application of this technology: My university has moved to RFID picture IDs for students and I was curious about how difficult it was to read the cards from an Arduino board. Our system is using 13.56 MHz cards, so I ventured into eBay territory. Almost anyone can type "Arduino RFID" there to get plenty of cards that allow you to read (and write) RFID tags/cards. For those of you living in a cave, RFID is a wireless technology that allows you to both power and interact wirelessly with a card or tag that contains a digital unique identifier that optionally may include some rewritable memory too. Tags or cards themselves do not have nor need a battery but they contain a microchip and a coil to gather energy from nearby readers. That means that the card and the reader need to be pretty close for this to work (that is why the technology is dubbed as NFC or Near Field Communication). After buying one card and waiting the usual two w

I had a visit this week of a teacher interested on creating 3D prints out of equations. The thing is that most of the tools I used need to have a polygonal mesh in the form of an STL file in order to print an object. I have found a couple of interesting tools. On one hand, Wolfram Alpha allows you to download any generated volume out of your equation, but you need to have PRO account. It seems you can have free PRO account for a couple of weeks, but later you have to pay a monthly fee. I did not explore that route any further. The other is a piece of software called K3DSurf that does the job for free. It is interesting though it did not work in my Mac, not even after installing Qt3, but it worked in Windows XP. It is worth a try!

Not one but two friends brought to my attention the apparently innovative new product by HP touted as a 3D scanner and printer  HP TopShot LaserJet Pro M275 . I was surprised by the marketing information and curious to see what this device was capable of. If you've followed my blog, you've read about my late interest on 3D scanning and printing, so a 3D scanner that was being made by the biggest printer maker seemed like a big deal and maybe the signal of new devices to come. However, after cutting through the marketing campaign, I was able to read some of the details of what the device can really do . That was the disappointing end of the history: what HP Topshot technology is about is not 3D scanning but to take pictures of objects. Do I really need to say that of course the objects you can take pictures from can be 3D objects? Are not all the objects we use 3D? Ok, maybe you can claim that paper pages are not 3D (we can argue on the scale of it). But to me, HP mark

I mentioned a while ago the cool software by Autodesk 123D Make that enable those with access to a laser cutter to create 3D objects from cardboard slices. The problem was that the software was only available for Apple computers. Now there is an online version that does the job no matter what operating system you use. The image shows a cardboard version of my happy new year design you can get on Thingiverse.com Update: A big oops if you use Linux ... (I do).

I borrowed another graphics card from a friend and this time it was the right one: Nvidia GTX 560 Ti with 1GB of RAM so I was able to test drive the new open implementation of the Kinect Fusion. First result is really promising. A couple of times I've got out of memory errors when starting the application but it was unclear to me whether it was main system RAM or the graphics' card internal memory. Pressing 3 you get the current point cloud written to disk. I am really shocked by the accuracy of the real-time results. I am getting around 20 fps. Definitely worth trying, but remember you'll need a powerful power supply too. I borrowed a 1.000 watts PC supply, sweet.

Once I realized all the work done for compiling PCL 2.0 was useless with my current graphics card I mentioned to a colleague the problem who was kind enough to lend me a Nvidia FX 5800 CUDA-enabled graphics card with 4 GB of RAM. I thought it was beyond the requirements of Kinfu so I did not bother checking it out before doing the installation. First problem was that this card hardly fit inside my computer. Later I found out my power supply just lacked the proper connectors to power that graphics card. I asked for help and I have got a 1000W PC power supply. This one has all the connectors needed but I am not fitting it inside my computer box. Once I was convinced of the right way of connecting the 8 pin PCI-express additional power socket, which was all but obvious to me, I fired up the system and it came up nicely. But, once again, I was defeated by the hardware: It turns out this board is not a good match for running Kinfu either. Though it has 250 cores, it lacks the computin

I was impressed when I saw the results presented in SIGGRAPH'11 about Kinect Fusion by a group of people from Microsoft Research. They manage to get a fantastic real-time scene scene registering using a software they named Kinect Fusion. A couple of papers were published explaining the system architecture and some videos shown the great performance they were obtaining. Today I have learned that an open source version is being built around the Point Cloud Library and the sample video looks equally promising. But development is only going to be supported on the upcoming 2.0 version of the library. I am going go give it a try as soon as I manage to compile the whole thing .   Ouch: After fighting with CUDA install (driver, tools, sdk) plus VTK library that was not installed plus some more fiddling with cmake and ccmake I manged to compile the current version of PCL2.0 but to no avail, as kingfu_app is not happy with my oldish video card (GF8400GS) and I get this lovely mess

Nicolas Burrus ' company has been kind enough to make available software for 3D scanning using a Kinect device that anyone can use for both Windows and OSX. What you obtain is a colored point cloud and not a mesh, but using software like meshlab you can turn that into triangular mesh. The way you use this software is by handholding your Kinect and moving it around the object or room you want to scan as if you were painting all the space. The point cloud is growing larger every time you move to a different area of the scene or object. Result is not as impressive as Kinect Fusion demos but, contrary to the former, this one is available. I guess the will keep on improving the system to obtain better spatial resolution with less noise.