Month: May 2013

Steppers, Scopes and Missing Pulses

by

As mentioned in a previous post, I was having trouble with my y-axis motors not doing as they were told. Thinking it might be at least part due to my crappy wiring job, I starting making some improvements. I bought some stepper motor wire from Inventables. This wire has four conductors plus a shielding and an uninsulated ground wire.

I crimped on some Molex style female pins on the ends and a latching connector to connect to the motor side. I then slipped on some heat shrink tubing from the untouched end and tried to shrink it with a hot glue gun. Don’t do this. It works, but the residual hot glue around the tip makes a stringy mess. For the other end I used the side of an old soldering iron. Note: I say an old iron and not my good one because it does leave some residue that will stain the iron. You can see the ground conductor sticking out the bottom of the heat shrink tube. I saved this in case I wanted it later, which I did.

I crimped on an extra ground wire to the conductor and crimped a ring terminal onto that. When crimping the ring terminal I stripped off about 3/8″ of insulation from the wire, then folded the exposed wire over the insulation. This gave me a thicker thing to crimp on. The end near the stepper wire is a bit of a mess, I tried using the base of another grounding ring terminal to crimp the wires without soldering, but ended up soldering anyway. The other side was better. On the other side I used a non-latching connector to plug into the RAMPS shield.

With my connections made I started testing my my motor using Pronterface to send code to the Arduino running Sprinter firmware. Some more on my setup:

Pololu A4983 Stepper Driver
Configured for 1/16 stepping
Trim pot set about 10 o’clock
Single Y motor (Left) connected to Z socket using shielded 3ft stepper wire.
Acceleration ramping in Sprinter OFF

Making 100 mm jogs back and forth and increasing speed in invervals of 500mm/min, I started to loose steps at 11000 mm/min. The loss was only on the negative direction though. Even more strange was that when I went back to speeds I had successfully achieved before, I now stopped short on the negative direction.

The Pololu heat sink and the motor were both remaining cool throughout the testing. I tried enabling acceleration ramping because at the higher speeds there was some violent jerking, even when it did complete a move.

// Comment this to disable ramp acceleration
#define RAMP_ACCELERATION 1
 //// Acceleration settings
#ifdef RAMP_ACCELERATION
// X, Y, Z, E maximum start speed for accelerated moves. E default values are good for skeinforge 40+, for older versions raise them a lot.
float max_start_speed_units_per_second[] = {20.0,20.0,0.2,10.0};
long max_acceleration_units_per_sq_second[] = {70,70,50,10000}; // X, Y, Z and E max acceleration in mm/s^2 for printing moves or retracts
long max_travel_acceleration_units_per_sq_second[] = {500,500,50,500}; // X, Y, Z max acceleration in mm/s^2 for travel moves
#endif

Things were less jerky, but I still stopped short when moving in the negative Y direction. I even tried reducing the max acceleration with no improvement. It wasn’t like the motor was pulling out and shuddering  it was as though it just stopped receiving step commands.

It was time to break out the big guns now. I set up an oscilloscope (borrowed from a friend) and connected it to the step pin and a ground on the RAMPS I2C breakout. I left the power to the motors OFF and just looked at the signals coming from the Arduino Mega. The pulses came in two widths. Primarily there were small pulses at 2.1us widths, but occationally there were also 8.4us wide pulses. The image below shows a trace where I am triggering on each rising edge and showing “persistance”.

Looking at a longer time scale I can see that each of the pulses are far apart and there was no chance that the pulse width would merge into the next pulse as I speculated at one point. Below is a longer trace showing two 100mm moves at 9000mm/min. The upper trace is the forward direction and the lower is reverse. (Persistance was used to keep the forward trace on the screen while taking the reverse trace.) Both traces show pulses throughout the entire 0.666sec of time they should. The motor power is still off at this point. Zooming in on the reverse trace I didn’t see any abnormalities.

I turned the motor power on and tried again at 9000mm/min. Now the motor stops short on the return and, interestingly, my pulses stop at the same time!

I repeated this test and made a video:

[youtube https://www.youtube.com/watch?v=tcTs_xJbnww]

Pulses stopped at 0.296sec when they should have continued to 0.666sec. That’s only 44.444mm of travel time. I measured the remaining distance (from the 100mm commanded) to the tape marking my start point. This was 50mm. Discrepancy may be due to the cart inertia carrying it after pulses end, or some pullout occurring in both directions during the violent starts.

It is curious that the pulses only stopped early when the motor was powered. This suggests that the motor power has some interaction with the controller being able to send signals.

Next Steps:

  • Investigate effects of endstop options (in firmware and hardware)
  • Investigate firmware options for STEP_DELAY_MICROS and STEP_DELAY_RATIO to combat missing steps
  • Add a second motor

X-Axis: Bigger Stronger Faster

by

I had trouble with my XZ Carriage flexing a bit too much for my liking. Before I tried extreme measures of adding a second piece of MakerSlide and another set of wheels, I figured I’d try some Open Rail first. I sponsored the Kickstarter project so I had some samples laying around. I cut two strips of this to the same size as my x-axis MakerSlide extrusion and bolted them on.

That’s a ton of T-nuts and M5 screw. I had to carefully line up all the T-nuts first then hold the Open Rail in place while I put in the screws. If I bumped it and shifted a T-nut away from a screw hold before I put a screw in, I would have to pull everything apart and start over. Luckily this didn’t happen.

I actually had more trouble with trying to use washers as spacers when reassembling my XZ carriage. I helped hold a stack together with a bit of tape. This worked pretty well. Also worth noting is that the orientation of these flat washers matter. These washers are stamped and have one rounded edge and one more sharp edge. The washers also curve just slightly away from the rounded edge. If you stack them so that they are all pointing in the same direction they have a smaller stack height than if you alternate them. I alternated my washers and that got me within about half a mm of where I wanted them.

You can see the alternating washers in the spacer set closest to the camera below. After struggling with that for about 3 minutes you can see my improved assembly method on the far one. You can also see the extra set of V-wheels for the new V-rail.

This modification did add quite a lot of stiffness to the XZ carriage. The z-axis does not flex nearly as much. Now the weak point seems to be a combination of two things:

  1. The carriage plate closest to the z-axis bends in the middle away from the screws. This is bending within the carriage plate itself. Modifications such as these reinforcing blocks shown about halfway down on this forum topic should also help me. (But I will have trouble trying to fit things around my Z Leadscrew.
  2. The z-axis V-wheel bearings are shifting on their races in the same way the x-axis ones did. I’m going to live with this problem for now. Things may improve when the leadscrew is installed.

The architecture for the x-axis on my machine didn’t lend itself to having the same style belt clamps as the x-axis did. I originally wanted to run the belts close to the extrusion as shown in the picture above. I tried putting some eye-bolts on the Y Carriage plate, but this would end up stealing over an inch of my usable x-axis travel on each side. The yellow cable tie shown was an attempt to make a larger radius for the belt to push it closer to the x-axis extrusion.

Next I thought of having some bracket that would push the eye bolt down from the top. This would make it much closer to the carriage plate, but it would be complicated and would not be as friendly to adjust. Also shown here is another attempt at increasing the wrap radius using some home electrical wire (single copper conductor).  I didn’t like this either.

I even tried making something out of Steel Tec parts, but this was a mess.

Next I started playing around with putting the eye bolts on the motor carriage plate instead. I had to route the belt upside down now, around some extra idler pulleys installed on the top corners of the X motor carriage plate. Although the time I put into machining the other idler pulleys so they would fit between the V-wheels would now be a waste (yes, btw, I machined those pulleys) and this setup doesn’t look quite as nice, I think this will be my best option.

I went back to CAD and some sketch paper and figured out what my bracket would have to look like. I needed to make the eye bolt a bit higher than the belt horizon because I needed room for the lock nut over top of the Y motor. If I made the bolt in line with the horizon, then the threads would be just above the motor and I couldn’t tighten it there. I thought about tapping the bracket, but that would mean I would have to remove the belt to turn the bolt and tighten it or else twist the belt. I printed a template and fabbed it up. Apparently I was paying too much attention to the vertical dimensions because my eye bolt mounting hole needed some post-design shifting in the horizontal to get it to line up with the belt. A rat tail file helped me out.

With both brackets completed I used some pieces of Steel Tec to make belt clamps. Those are #8 screws and nuts. This all turned out pretty well.

Ok, I wasn’t going to mention machining these idlers I didn’t end up using, but there are some interesting points. I did some sketching and found out that I could still fit a 9mm idler between my V-wheels when I doubled my rails. These little salvaged idlers are a machined tube with a 5x5x16mm bearing pressed in them. I knew I needed only enough room for 2 shim washers on either side of the bearing and I needed to make sure I machined it symmetrically. So, I put the two washers on the bearing and measured the depth to the outer edge. Then, on my lathe I faced the bearing end until I got to this depth.