Thursday 9 June 2011

Reliable connections

After eliminating lots of other sources of unreliability in my machines, electrical connection failures are now the most common failure mode.

The latest failure on my Mendel was that it started leaving a 10mm gap in the outline rectangle that it draws around the objects. Since a bed full of objects still seemed to build OK I decided perhaps it was due to an air bubble in the extruder while it was warming up. However, one time I saw the extruder motor stall and realised it was actually a bad connection.

I have come to realise that simple friction fit connectors do not work in the environment of these machines. I tried re-seating the motor plug but that did not fix it, so I figured the cable must be faulty. I wired both coils in series to my multimeter and waggled the cable until it went open circuit. That allowed me to locate the break and it was, as could be expected, at the point where the cable bends the most, i.e. just below the cable clamp on the  top right of this picture: -

On reflection this was not a good arrangement as the cable is only just long enough for the extremes of travel, so it is forced to bend sharply both ways at the clamp. After millions of movements the strands break one by one but the insulation holds the ends together making it only lose contact when it is stretched. When I pulled the ends of the wires three of them snapped very easily, indicating most if not all the strands were broken.

I had a similar problem with the mains wire to a heated bed a while ago. In that case the arcing melted the insulation and allowed the live and neutral to short out, blowing a hole in outer sheath of the cable. Not good! Normally you expect a fuse to protect against a cable fire, but if all the strands start breaking, reducing the current capability, or it breaks and arcs, the fuse offers no protection against fire. Even a low voltage heated bed could fail in this way because of the high current.

The XY table of HydraRaptor uses 9 way D-type connectors. These have been totally reliable moving connections because they are screwed together and have gold plated pins and proper strain relief. The professional stepper motor drives on HydraRaptor have screw terminal blocks for their connections, and again they have proved totally reliable. In contrast all the friction fit connectors fail if there is any movement or vibration of the wire. Some even burn out despite being run at well below their current rating. The contact resistance rises and they then start to heat up.

I rewired my Mendel extruder using a 9 way D-type at the extruder and a longer loop of cable. That necessitated resiting the extruder controller and I also replaced all of its 0.1" MTA connectors with screw terminal blocks. The wires could go straight into these but I added ferrules to allow them to be more easily removed and replaced. I just push the wire into the ferrule and then squeeze it with pliers.

I reprapped a bracket to attach the DB9 connector to the back of  Wade's extruder bracket.

The pins are four motor connections, two heater, two thermistor and one heatsink fan that shares a 12V feed with the heater.

Here is the new arrangement :-

The cable loop is much longer, so it bends through a much smaller angle. The top end goes through two cable clamps before it goes to the extruder controller. I found that if you put a bunch of wires though a single clamp you can get some movement at the other side of the clamp. Using two eliminates any movement of the wire relative to the board, less critical now I that have screw terminals, but still a good idea.

It should last a lot longer than the previous cable (which lasted for 15 months of continuous use) and can be easily replaced. I have seen people use corrugated tubing to protect the cable, but I didn't fancy adding any more drag on the extruder as it would increase backlash.

Interestingly, although my extruder stepper motor connections have failed several times, I have never damaged the Allegro driver chips.


  1. I couldn't agree more. I've had opto-endstop connections fail numerous times, extruder stepper motor connections fail a few times, and heater connections fail too. Very annoying. For the connection between my extruder/thermistor/heater and my RepRap, I've recently resorted to a European-style terminator block (despite the fact that it's based on screws which are affected by vibrations), tightly clamped around the wires.

  2. I've done the same as Jeff, but using two tewrminal strips side-by-side and joining the two with short lengths of stiff wire. This allows me to separate the two strips without having to undo any of the fragile equipment wires.

    I've had one wire breakage so far, and that was on a machine that didn't have spiral cable wrap on the moving wire loom.

  3. Great blog post as always, i was just wondering have you had any job offers from 3d printer companies.

    If i was a ceo of stratasys (or any commercial 3d printer company), i would hire you instantly.

  4. Jussi,
    No, why would they pay me when they get all my research for free?

  5. If they got you to work for them, they wouldn't of course let you publish your research, they would patent it.

    But i got a feeling that you wouldn't agree to that.

    "when they get all my research for free" - do you have any licenses for your invetions or can they just copy anything

  6. To license an invention I would need to patent it before blogging it, so anything I blog is free for anybody to use (assuming it is not already patented by somebody else).

    When I publish files they are subject to copyright, and I use the GPL license to match the rest of RepRap.

  7. I suggest we make a standard extruder connector pinout from this DB9 connector :-) Unfortunately there is no place for Z probe in this connecter :-/

  8. You could use a DB15 or for a hack you could use the shield.

    I don't use a probe on my Mendel, because once set it does not change.

  9. Could be worth looking to see if high current Flexible Film Cable is avalible its light weight used in std Printers. fi

  10. I would put the cables in a cable chain ( for example), attached to one end of the y-axis and to the head.