Effector moving up/down slightly during horizontal moves

I've just built my first delta (Kossel XL) running on Marlin 1.1.0-RC5. I have calibrated the endstops, rod length and radius using the paper test at the bed centre and the feet of the three towers.

The issue I am having now is that when I move from home to any of the towers, everything is fine and the hotend is the correct height above the bed. But when I move from one tower to another, the print head moves up/down slightly and doesn't end up in the correct place.

For example, moving from home -> X -> Y results in the hotend being slightly higher than moving directly from home -> Y. Similarly moving from home -> Y -> X results in the hotend being too low (and scraping on the bed), even though directly from home -> X is fine.

I have tried various combinations of speed and segments per second with no change in the observed behavior, any ideas what might be off?

Sounds like backlash in the Traxxas joints. Try to use a pair of springs on each carbon rod pair. Put one spring close to the effector, the other close to the carriage. Rubber band works, too.

There are also things to mount the springs more easily: [www.thingiverse.com]

But could also be backlash in your motors/carriage mount. Is everything put together firmly? If you fling your GT2 belt, does it have a low to medium bass like sound? Like ... "plooooiiing". Is the sound approximately the same for all three towers?

Thanks for the tips! I have rubber bands in the center of the rods at the moment, I will try moving them to the ends as you suggested.

I have checked and tightened all the screws on the motors, frame and pulleys, and all the grub screws on the shafts. The carriages are the standard Kossel roller carriages that slide on to the openframe and are secured with a M3 bolt in the center.

I'm not sure about the correct belt tension, I pulled them as tight as I could by hand and then tightened them a bit more using the screws in the top of the frame. The belts are all approximately the same and make a very low sound when plucked... Possibly the are not tensioned enough?

Is the height dependant on the direction you got there? ie. X -> Y is a different height than Y -> X? Or is it just a case of the head being the wrong height when between towers?

If the latter, I'd highly recommend going back over your build with a pair of calipers and a set square, and ensure that your towers are parallel and not twisted.

Failing that you can always try out DC42's delta calculator: [escher3d.com]

Quote
nebbian
Is the height dependant on the direction you got there? ie. X -> Y is a different height than Y -> X? Or is it just a case of the head being the wrong height when between towers?

If the latter, I'd highly recommend going back over your build with a pair of calipers and a set square, and ensure that your towers are parallel and not twisted.

Failing that you can always try out DC42's delta calculator: [escher3d.com]

Hi nebbian, my printer seems to be suffering from the first case where the height is different when i more from +Y -> -Y as compared to when moving from -Y -> +Y.
I tried to print a calibration circle but there are places where nothing is extruded when the printer is moving from +Y -> -Y, which i assume is due to the dipping of the effector....
Is there anyway to remedy this?

Anyone has any idea?
BUMP

Quote
verdesmarald
For example, moving from home -> X -> Y results in the hotend being slightly higher than moving directly from home -> Y.

1. How much is "slightly" ?

2. What is the smallest distance you can jog the print head up and down, and have it really move every time you jog it?

3. What stepper motors are you using, and what do you have the motor current set to? Are the stepper driver chips and/or the motors getting warm or hot?

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

1. About 100-200 microns of vertical movement when I move the whole width of my print bed along the x-axis (250mm). I haven't tried movement along the y-axis but I guess I should! The movement is enough that parts of the edges of large prints don't adhere well and the print head scrapes on the previous layer during printing.

2. It doesn't move if I send anything less than 50 microns from pronterface, but I'm pretty sure this is configured in marlin?

3. Wantai nema 17 steppers with astrosyn dampers that run completely cold, and pololu a4988 stepper drivers with heat sinks. The heat sinks are hot to touch but don't burn me unless I hold my finger on them for a few seconds.

Re (2). AFAIK there is no minimum jog size in Marlin. The theoretical minimum jog size is 1 microstep, which if you are using x16 microstepping is likely to be 10 or 12.5 microns depending on whether you are using 16 or 20 tooth pulleys. It's quite common for the head to need 2 or 3 microsteps before it will move, but 5 is excessive.

You didn't answer my question about what stepper motors you are using (I am interested in the rated current in particular) and what current you are running them at. In the absence of this information, my guess is that the problem is one of the following:

1. You are running the stepper motors at too low a fraction of their rated current. Aim for 60 to 80%.

2. Your stepper drivers are configured in mixed decay mode instead of fast decay mode, so they are not responding to some microsteps.

3. Too much friction in your joints or sliders/wheeled carriages.

btw stick-on heatsinks on driver chips are not effective, except for RAPS128 drivers and for drivers where the heatsink is mounted on the PCB and the chip is on the other side of the PCB. For other drivers, you are better off using a fan and no heatsink.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Thanks dc.

Re the motors - Sorry I didn't realize there were multiple models in the Wantai nema 17 line, mine are 42BYGHW811, rated for 2.3A.

Re minimum jog size - there is a setting in configuration_adv called dropsegments that defaults to 5 with the comment "everything with less than this number of steps will be ignored as move and joined with the next movement" . I lowered it to 1 and now I can jog smaller amounts (2 steps/20 microns).

1. I have the driver pots set to 0.72V, which I believe corresponds to 0.72 * 2.5 = 1.8A = ~78% of the 2.3A rated current. The drivers are rated for 2A so I believe this setting is OK.

2. The A4988 has "Automatic current decay mode detection/selection (mixed/slow)" whatever that means? The specification seems to say it uses mixed (fast then slow) when the output is falling and slow when it is rising.

3. Loosening the tension bolts on the wheeled carriages made things way worse, but tightening them seems to have improved things a little.

I have a 50mm cooling fan for the stepper drivers but I haven't wired it up yet, do you think it is worth pulling the heat sinks off and hooking the fan up instead?

I wasn't aware of the "dropsegments" value. I guess Marlin uses that to avoid getting overloaded on 8-bit processors. I don't use Marlin.

If you are using plug-in A4988 driver modules, there is zero chance that you can run them at 1.8A without them overheating. A4988 drivers need a large PCB area to handle that amount of current, and plug-in drivers just don't have that. However, I suspect that your drivers have 0.1 ohm sense resistors (most common ones do, it is just the Pololu Black Edition and a few others that use 0.05 ohms), in which case you are actually running them at 0.9A. If that is that case, try turning the current up to 1.2A, and definitely use the fan.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Hi dc,

I'm confused about the second part of your post, the pages on the Pololu website for the A4988 say the current limit is 2.5 times the VREF setting for both the black and regular editions, why would the real current only be half that amount? Also I am not sure what a sense resistor is, or if mine are 0.1 or 0.05 ohms, sorry.

Quote
verdesmarald
Hi dc,

I'm confused about the second part of your post, the pages on the Pololu website for the A4988 say the current limit is 2.5 times the VREF setting for both the black and regular editions, why would the real current only be half that amount? Also I am not sure what a sense resistor is, or if mine are 0.1 or 0.05 ohms, sorry.

When you check "vref" you are checking the voltage across a sense resistor. By ohms law (V=IR) you can tell the current present by knowing the voltage and resistance. If you are working under the impression that the sense resistor you have is .05 (or even using the multiplier above as if it was the same) and your real sense resistor is double that, the current would be half the value you expect.

Quote
verdesmarald
Hi dc,

I'm confused about the second part of your post, the pages on the Pololu website for the A4988 say the current limit is 2.5 times the VREF setting for both the black and regular editions, why would the real current only be half that amount? Also I am not sure what a sense resistor is, or if mine are 0.1 or 0.05 ohms, sorry.

+

You are quite right, both types of genuine Pololu driver use 0.05 ohm sense resistors (I thought the regular edition used 0.1 ohms, but I was mistaken). However, most Chinese-built clones use 0.1 ohm resistors, and some may still use 0.2 ohm resistors. See [reprap.org].

The sense resistors are the two largest SMD resistors on the board. If you are lucky, they will have the resistance marked on them, e.g. R100 or R050.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42

Quote
verdesmarald
Hi dc,

I'm confused about the second part of your post, the pages on the Pololu website for the A4988 say the current limit is 2.5 times the VREF setting for both the black and regular editions, why would the real current only be half that amount? Also I am not sure what a sense resistor is, or if mine are 0.1 or 0.05 ohms, sorry.

+

You are quite right, both types of genuine Pololu driver use 0.05 ohm sense resistors (I thought the regular edition used 0.1 ohms, but I was mistaken). However, most Chinese-built clones use 0.1 ohm resistors, and some may still use 0.2 ohm resistors. See [reprap.org].

The sense resistors are the two largest SMD resistors on the board. If you are lucky, they will have the resistance marked on them, e.g. R100 or R050.

On the vendor website you can sometimes see the picture of the boards and check for this. The one with 0.1 ohm I got have also R4 0 ohm for low Current Microstepping mode.
Buy only if you can check this and pray you get what you saw.
Note that on my Geetech Prusa, I realized the driver had this 0.2 ohm, therefore were set to 0.5 A and it is working extremely well after mod. It is just the original direct extruder that couldnt keep up when running fast; so I ordered these drivers.

Thanks for the info dc, electronics is not my forte!

My drivers came in a kit so I have no way of knowing if they are genuine Pololus or not. I'll see if I can read off the resistance when I get home from work. Worst case I can hook up an ammeter in series with the motor coil and verify it that way, right?

I'll hook up the fan and adjust the current limit up/down to 1.2A and see how it goes.

Quote
verdesmarald
Worst case I can hook up an ammeter in series with the motor coil and verify it that way, right?

If you do that, the connections to the meter must be secure before you turn the power on. Connecting/disconnecting motors with power applied is what blows stepper drivers.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
Connecting/disconnecting motors with power applied is what blows stepper drivers.

Sorry for offtopic, but can someone explain why this is happening?

Quote
roboprint

Quote
dc42
Connecting/disconnecting motors with power applied is what blows stepper drivers.

Sorry for offtopic, but can someone explain why this is happening?

I don't know for sure, but it's probably because if you disconnect a stepper motor when it is energised, it will produce a high back emf and for a short time it will spark across the gap where you have just broken the connection. Such a spark will cause very rapid changes in the current flowing, not at all like the smooth current changes you get when the driver itself turns off the current. My guess is that it is these very rapid changes in current that the driver doesn't like. But this is only a guess, it could also be due to the voltage waveform being even sharper than usual. One day when I feel like blowing up a few of my old stepsticks, I may take a look at what exactly happens with an oscilloscope, and see what measures I can find to protect the driver.

Edited 1 time(s). Last edit at 05/17/2016 07:54AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42

Quote
roboprint

Quote
dc42
Connecting/disconnecting motors with power applied is what blows stepper drivers.

Sorry for offtopic, but can someone explain why this is happening?

I don't know for sure, but it's probably because if you disconnect a stepper motor when it is energised, it will produce a high back emf and for a short time it will spark across the gap where you have just broken the connection. Such a spark will cause very rapid changes in the current flowing, not at all like the smooth current changes you get when the driver itself turns off the current. My guess is that it is these very rapid changes in current that the driver doesn't like. But this is only a guess, it could also be due to the voltage waveform being even sharper than usual. One day when I feel like blowing up a few of my old stepsticks, I may take a look at what exactly happens with an oscilloscope, and see what measures I can find to protect the driver.

You get high back EMF that will most probably far exceed the chip breakdown voltage, the protective diodes may not be able to do much. These motors have quite some L and very low R, thus decent Q.

Any progress with the problem?