3

I am trying to find a low cost high accuracy linear position sensor that can be used in a hobby 3D printer. Almost all current 3D printer designs rely on stepper motors to provide positioning but I want something that works on positioning info rather than assuming the motor will do what it's told. After a few hours of searching I have found lots of technologies that provide high accuracy sensing but they are all sold by companies who tailor make a whole system for application and require that you request a quote. I am assuming that things are probably insanely expensive and not something for a hobbyist. I am obviously not a professional and would just like to find out if any of you know of a technology I can use that should not really be more expensive than stepper motors?

PS. The type of accuracy I am looking for is about 0,01mm.

Gerharddc
  • 285
  • 5
  • 17
  • 1
    You should try taking apart some regular 2D inkjet printers. They often have a transparent tape with fine lines ruled on it and an optical quadrature sensor to read it. This should be easily adapted to your application. – Dave Tweed Jan 20 '14 at 17:16
  • @DaveTweed Thanks, I am aware of the encoders that can be found in normal printers but I am trying to design something that others should also be able to make from off the shelves components, as in not rely on having to scarp something (I for instance do not have access to a scrapped 2D printer myself...) – Gerharddc Jan 20 '14 at 17:22
  • I'm not sure if an [LVDT](https://en.wikipedia.org/wiki/Linear_variable_differential_transformer) has the required resolution. – jippie Jan 20 '14 at 17:31
  • @jippie Thanks, I have in fact read about them but I have no idea what one of this type of resolution will cost me, any idea? Also they don't seem to be intended for large distances? I need about 300mm of travel... – Gerharddc Jan 20 '14 at 17:34
  • @Gerhman My Google is as good as yours. I have about 2 hrs experience with them, during a lab in university. – jippie Jan 20 '14 at 17:42
  • I'm in the process of evaluating linear position magnetic sensors. A magnetic strip is placed on the machine and a sensor IC is moved over the strip, and due to the polar magnetization of the strip, the IC can, in theory, determine the linear travel. I am working with an AMS AS5311 IC right now, 1.95um resolution. From what I take though, these types of sensors require very good implementation to perform to rated specifications. I have a log on my website; I just finished building up a board and I'm working on the firmware now. Very early in the experimentation process, no hard data yet. – dext0rb Jan 20 '14 at 18:07
  • 1
    Is 0.01mm realistic?? NONE of the commercial systems claim that sort of resolution. In fact, none of the high precision sintering methods claim that either. There probably isn't near that accuracy inherent to the thermal extruder. – Scott Seidman Jan 20 '14 at 18:55
  • 1
    For cheap/hobby positoning I'd look at CNC Digital Readout (usually abbreviated to DRO), originally big & expensive, the eBay China ones these days are all variations on the theme of digital vernier calipers, which usually have a serial port behind a little flap. See this project for an example: http://hackaday.com/2014/01/19/three-axis-position-indicator-with-digital-calipers/ – John U Jan 20 '14 at 19:05
  • Are you absolutely sure you don't mean 0.01 *INCHES*?? – Scott Seidman Jan 20 '14 at 19:20
  • 1
    Yes 0.01mm is realistic for even the cheap digital calipers. Unless you need it guaranteed over time and temperature! –  Jan 20 '14 at 20:43
  • If you don't trust your stepper, you can put an optical q-encoder at the axis. Add sufficient reduction gear after that to drive the worm-wheel to get the required accuracy. (Assuming the wormwheel can deliver the accuracy in the first place.) – Wouter van Ooijen Jan 20 '14 at 22:03
  • @JohnU The cheap digital calipers are probably good to more like +/-0.05mm accuracy (0.02mm claimed), resolution 0.01mm. Readout speed is 3 or 40 measurements per second, so it's difficult to use them for feedback. They are sure cheap though, and usable for some purposes (like my milling machine quill) if you can slow down as you get close. – Spehro Pefhany Jan 22 '14 at 02:56
  • @jippie LVDTs have almost infinite resolution (they're analog, after all, and limited only by noise), but their linearity is not nearly good enough for that kind of accuracy, maybe 0.5% of travel, so his travel would need to be limited to a few mm. – Spehro Pefhany Jan 22 '14 at 02:59
  • @SpehroPefhany - I realise they are not going to be to the full spec, however the given application was "hobby 3D printer" which is unlikely to really *need* the spec he thinks he does. – John U Jan 22 '14 at 09:49

2 Answers2

2

The only think I know about in that range is the glass scale encoders used in machine tools. They provide a quadrature output (so you can determine direction) but are not absolute sensors, so you do need a way to home or "zero" them in a known position.

Typical resolution is 5um, so twice as good as your requirement, and claimed accuracy is 5 or 10um typically, depending on length.

You can find such Chinese scales on eBay and other places in various lengths, starting at two or three hundred dollars. We use them on a milling machine (3-axis) and lathe (2-axis) to speed up manual machining- you have to get the alignment reasonably close when you're installing them, but it's not all that difficult with shims and such like.

Brand-name ones such as Mitutoyo cost more, of course. This is a tiny fraction of the cost of Heidenhain and similar high-end encoders (I work with such a system that has a resolution measured in nanometers, and with a similarly impressive price tag).

Here's what the scales look like. The aluminum part moves relative to the slider with the cable attached.

enter image description here

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
2

You can use the optical part of a very sensitive laser mouse for a PC. I have seen that type of optical system provide extremely good resolution - and cost is down where you want it.

This one appears to have around 0.02mm resolution: http://www.parallax.com/sites/default/files/downloads/28560-MouseSensor-MCS-12086-Datasheet.pdf I am sure you can get something better.

Be prepared to work a lot with the algorithms to get reliable absolute position. Good thing is you know the approximate speed and position already plus you can have a consistent good surface to track on. Let me know how it goes?

Rolf Ostergaard
  • 4,448
  • 18
  • 24