Horizontal Skips

[n9wxu]

I think I am down to my last major problem.
Horizontal Skips.
In the attached picture you can see three representative objects. All of them exhibit the problem.
In every case they skip AWAY from the Z column.

All prints are supplied from the repetier host program running on my Mac.

What I have tried.
1) Increase the stepper motor current (175 default -> 195)
2) Improved stepper driver cooling (see the thermal images in the other post)
3) Changed the firmware to repetier 91
4) Moved PEEK fan, extruder power, extruder temp and 4 spare wires into a PVC tube separate from the extruder stepper cable. Stepper cable goes down the side slot in the Z extrusion. Cheapskate seems to run freely over the cable.
5) Laced and routed all limit switch wiring around the top of the ROSTOCK base (underside of the top deck)
6) Laced and routed all stepper wiring around the bottom of the ROSTOCK base (topside of the bottom deck)
7) Checked the bearing covers for burs. There does not seem to be any protrusions on them.
8) adjusted the cheapskate for slop. It appears to run free.
9) corrected motor pulley set screws. They are all secure on the flats
10) Ran a random X,Y,Z break in sequence (attached)

I have observed that on occasion the Z steps simply stop. On some occasions this is when the print is over and the head is parked at max X, Y0. During the failure the head collides with the X column because the Z is "stuck". Pressing home after the motion stops always works correctly.
It almost looks like the issue is periodic. The two column prints have a 10mm step but there are a few smaller steps that are hard to see. If it is 10mm periodic, then I can expect the issue to be a 3mm diameter rotating thing (2*Pi*R = 10, R = 1.6mm) That is getting close to the motor shaft diameter but not quite. A sticky spot in the cheapskate bearing would occur every 80mm or so.


What can cause an "electrical" stop of the steppers that I have not addressed?

ToDo:
1) Move the Extruder Stepper wires out of the Z column into a second PVC pipe.
2) Disassemble the Z cheapskate, clean, polish, reassemble.

[daftscience]

IIRC, my steppers are set to 175. If you're stepper motors are overheating try lowering it back down. When you say it appears to run freely it should freefall when it's not connected to the motors. [youtube]http://www.youtube.com/watch?v=5LCHGMivJeE[/youtube]

If you want to rule out the endstop interference, go into configuration.h and change this:

Code: Select all

#define ALWAYS_CHECK_ENDSTOPS true

to this:

Code: Select all

#define ALWAYS_CHECK_ENDSTOPS false

With that set you will home just fine but when it's printing you can press the endstops all you want and it wont stop the motors.

Aside from that, maybe double check your motor wires, make sure they are crimped on there well and there isn't any loose connections.

[lordbinky]

You seem to have addressed most possibilities. Have you done prints at lower speeds? After the depth of steps you've covered so far I know this is a bit silly, but you've confident in your belt tension right?

[n9wxu]

Moving the wires away from the endstop wires seems to have fixed the problem.

The Cheapskates may be slightly tight. They need a nudge to free fall.
The belts are good.
I could reduce the current but even at 175 it seemed a bit warm. I suppose I should lower the setting and retake the pictures.

Lower print speeds did not fix the problem earlier. I think that slower retract settings may have reduce the problem occurrence and knowing it was crosstalk this makes sense.
If I had known earlier about the endstop check code I would have tried it. It would have saved a few hours of rerouting wires and recrimping the ends.

Now I need to complain on the slic3r forum. Apparently slic3r treats the brim and skirt as layer 1 so my code speeds up to print the actual portion of the part that touches the bed. This makes it hard to get good adhesion. I set the reduced speed for layer 1 but this is apparently layer 2. Getting ABS to stick does seem to be an art form.

[mikefazz]

Hi all,

I have been battling this for a while now (Y axis for me) with no luck as of yet. I have tried many things:
- Belt tensioners
- RC type Ball ends
- Switching to 1.75mm filament
- loosing the cheapskates
- adjusting up stepper driver power

I have the endstops only checked during homing.

I have the EZExtruder mounted on top with the wires all running down along a threaded rod I added. I would expect the thermister reading to be an issue if there was EMI.

The Y axis motor is the closest to the board so I would be surprised if it was the problem.

I have the old plastic pulleys on the motors but they don't seem cracked when I looked at them closely.

While this is probably a bit paranoid I think I have noticed the blips sometimes occur when the garage door is opened. My garage door is a jackshaft drive one and does special stuff like variable speed during opening and closing that could throw out noise? Or it could be it letting drafts in?

I feel I am running out of ideas on what could be causing this... a bad board? Has anyone put their board in a faraday cage (metal box)?

My electrical setup is a large server power supply, RUMBA, DVR8825 for axes steppers. Stepper wires twisted by coil, plugged in to board.

I also have a custom water cooled hot end which is pretty heavy. I have the acceleration set to 300 and the jerk at 20.

Any suggestions much appreciated...

[lordbinky]

Aha! Reduce your jerk to 4 and do a test print to see if that clears the issue up. You have (relatively) ALOT of mass on the extruder platform.


Edit: I have the metal pulleys and I haven't cause my belts damage yet and I've locked up the belts in various ways (more than I'd like to admit). So just because you don't see belt damage doesn't mean it's not jumping a tooth or even skipping a step in the motor.

[Eaglezsoar]

Have you tried moving the end stop wires away from all other wires as was done in the thread above yours?

[n9wxu]

There are a few reasons why the limit switches reveal EMI issues and not the thermistor.

1) Input impedance
2) Current induced
3) System Affect

Lets take them in order.
The susceptibility of a cable to EMI is related to the impedance of the circuit the cable is feeding. If you are inducing 1mA into a 1K load you get 1V of noise. If you induce 1mA into 200ohms of load you get .2V of noise. The endstop resistance is the internal pull-up of the AVR I/O pin is listed in the data sheet between 20k and 50k ohms. The input Vil level is listed as 1.5v maximum. Therefore it takes between 70uA to 175uA of induced current to trip the end-stop. That is not much current. If the design had a 200ohm pull-up resistor on the end-stop pins it would be far more robust.
The thermistor has is 100k but it has a 4.7kohm pull-up. Assuming 175uA of noise I will see 0.8v at the ADC of the MCU. But there is a 100nF capacitor creating a low pass filter on the thermistor. As the frequency goes up (and induced stepper spikes are VERY high frequency) the effective impedance goes down and the noise at the ADC pin goes away. Adding a small capacitor to the endstops would also help.

Induced current. The more current you put into a long wire, the more the wire will radiate as an inductor and the larger the affect will be. I bumped my stepper currents up (175 -> 195) this aggrevated the problems. Distance will attenuate the issue. The problem comes from creating a magnetic field around the wire. The magnetic field is changing so it will create a similar current in a parallel wire (just like moving a bar magnet near a coil). Because the motor wires are running in parallel with the stop wires for 400mm there is a large amount of coupling so the 1.5A can easily reach the 175uA in the parallel wire.

The affect on the system for a falsely tripping e-stop is immediate and obvious. The affect on the system for the temperature is subtle and almost unnoticeable. You may need to chase the set point a little bit during certain extrusion rates, but most likely the filter will cancel most of the issue and leave a very slight DC offset in the voltage leaving a temperature bias when extruding. This can be adjusted by bumping the temperature a little up or down and chalking it up to the "personality" of your machine.

[smiley]

snip/

I have the old plastic pulleys on the motors but they don't seem cracked when I looked at them closely.

/snip

I had a cracked plastic pulley. You could feel (and hear) the skip from the crack by moving the platform around with your hand with the power off. Sounded almost exactly like a rock or a screw caught in your car tire, tick! tick! tick!

[mikefazz]

Hi all thanks for the suggestions, and thanks n9wxu for the detailed explanation of how end stops pick up noise.

I have the 'always check endstops' un-checked in repetier so unless the wires from the end stops pick up noise like antennas and it somehow gets into the rumba board I don't think that is a problem. I have the end-stop wires inside the 8020 with no other wires next to the end stops. They do cross the stepper wires at the bottom but they don't follow those wires for more than a few inches.

lordblinky I set the jerk down to 5 and did a long run over the night still offsets... Could try 2...

lordblinky/smiley when I initially got the kit one of the pulleys had an obvious crack in it and I got a new one. The others didn't show any cracks... If I ever do a tear down I'll try remember to replace them.

So my latest idea from talking with my dad (a proper EE unlike this ME) made me think my issue is the automotive relay I am using... The RUMBA doesn't have a powerful enough MOSFET so I put a 30A automotive relay running off the heat bead output. Well apparently mechanical relays are noisy (internal arcs) ;( The Y axis stepper wire crosses near this relay. It also kind of explains why offsets seem to occur more when the garage door is opened as the heat bed may cool down some resulting in more relay tripping. I don't think I had offset issues back when I used the rambo.

So I am trying a print with the relay disconnected and thus no heated bed ... fingers crossed.

[lordbinky]

Seems like you got it then.

[n9wxu]

When you added the relay did you place a diode across the coil pointing at the supply?

If you did not put the diode there you will get a high voltage flyback at the top of the MOSFET. At a minimum this will eventually destroy the MOSFET. It can also couple massive noise into the PCB causing any number of issues such as missing/extra pulses to the stepper drivers.

Have your father check it for a snubber circuit across the coil. If he told you to add one (probably did) make sure you put the snubber in properly.

[mikefazz]

n9wxu: I didn't add a diode, snubber, or capacitor across it ... I ordered some SSRs (something I meant to do a while back). Do you still suggest adding circuitry with an SSR? Could the 'unblocked' relay board have caused any permanent damage to the rumba board? Other than the loss of steps and fidely USB connection the board seems to work fine.

So my latest test (no relay connected) still had issues. I actually did a print with t-glass on my SpiderBot at the same time which is right next to the MAX and also got offsets with it!

Both printers use the RUMBA board so my next test is to put the original rambo back in the MAX and see if that fixes things. I have seen a couple mentions of RUMBA being susceptible to noise so if the rambo fixes this I'll look into using megatronics2 in the future since they can handle 3 extruders and seem a bit more reliable...

[mikefazz]

So before I left on a week long vaca I ran a long print that was almost finished before going to airport... I had put the original rambo in. The prinit looked good so next I'll try tightening the cheapskates a little and increase the speed. If all goes well I'll eventually switch to megatronics... Almost ready to kick rumba to the curb...

[mikefazz]

So I ran a 40 hour print at around 70mm/s feeds with no offset issues. So after quite some time trying to track this down looks like it was rumba all along.