Posted on

How making calibration foolproof drove me insane…

I am working on a assembly guide, but when printing a part on the new printer I found a simple calibration issue. The part came out ever so slightly oval. That set me on a path that took up quite a lot of my time… Too much, but what is what happens when you are on leave, and you have the time to play a bit…

Anyway, the idea is as follows.

  1. Print out SCARA-Cal, a special phantom I made for Morgan that initially helped me tweak steps per angle.
  2. Measure the following lengths marked on the phantom, Theta, Psi, X, Y and enter into a spreadsheet
  3. Said spreadsheet calculates Theta and Psi steps per degree, and the X and Y scaling required to make perfectly round circles.

My calculations are still a little flawed, but I am trying out a print right now…  If it works right this will be the easiest way to ensure a perfect first calibration.  If it does, I will re-add SCARA-cal and a quick guide with the spreadsheet to Git

 

10 thoughts on “How making calibration foolproof drove me insane…

  1. Hi Quentin,
    The more detailed assembly guide is necessary!
    Awaiting more pictures and more videos.

    1. Absolutely. It is a multi tier approach at the moment. Nicholas Seward ( Reprap Simpson fame) is working on a simpson style assembly diagram for Morgan at the moment, but converting OpenSCAD to Inventor is a very daunting task indeed. Will take a little longer than expected. There is also a possibility that there might be a picture guide in the latest Reprap magazine… Lets see how that pans out.

      I the mean time my photo guide should be done by the end of next weekend, if all goes well.

  2. A little suggestion:
    Since the bed is aluminium, why not make it a touch type calibration where the hot end and the bed form a closed loop when they touch…

    1. I would love to do that… But unfortunately you do not print straight to the aluminium. It has to have something the plastic will adhere to, be it a layer of Kapton tape (excellent insulator) or some ABS juice. These are unfortunately not conductive, and therefore you need a mechanical method of probing the bed surface.

      I have a mechanical workaround that is not too hard to do, and does not take long either, but there will be a mechanical probe update for morgans everywhere soon.

  3. Just add the tape afterwards, especially if it is a one off thing and the bed is quite open so it should be easy to add the tape afterwards.
    Or leave tiny dots open for the calibration points.
    It just seems to be a very fool proof way to do the calibration.

    1. You just gave me an idea… The print surface is loose in order to facilitate easy swops. If you turn the bed around ( given that is is completely flat of course) you can use that method with ease.

      It will then be a run once use many routine ( which is working fine at the moment anyway) but wont require so much interaction from the user.

      The biggest calibration issue remains the dimensional calibration, but I am getting close to automate that as well.

      Thanks,
      Q

      1. If you then stuck a non conductive circle on the other side so which the head has to follow around without touching a conductive piece of material it can give you a reference shape…

        1. Phineas, I know what We are going to do this weekend…

  4. I’m new to 3d printing and would like to find out the precision, accuracy and sizes of prints that can be made using your Morgan printer.

    Regards,
    Mike

    1. Hi Mike,

      Morgan is a bit of a freak in that regard, because you can choose what you want.

      The build envelope is currently 200x200x220mm, and I have build parts that challenges the limits in X,Y and Z.

      In terms of accuracy and precision, you get what you wait for. Morgans use SCARA inverse kinematic that can make it work very fast, at a lower quality, or much slower for very detailed prints. The reason for this is two fold. First, the plastic needs time to be placed accurately, and the microprocessor is only capable of about 200 inverse kinematic calculations per second. It should be clear that slower movements will have more calculated moves per unit length, than fast prints.

      I will post a photo of some froggies I printed recently The tiny one has the same level of detail as the bigger ones, but printed at more than double the resolution.

Leave a Reply

Your email address will not be published. Required fields are marked *