Delta frustration - ready to give up.
-
Phil
Surely it is more likely to be a pulley that is off in some way have you physically counted the number of teeth on each one?
Know what your steps/mm are currently set to may give a clue as well?
Doug
-
I agree that the pulley (or belt?) would be a much more likely culprit than the motors.
-
Phil, I really think your X pulley is slipping on the motor shaft, even though I know you have already checked that. Try energising the motors, send M906 I100 to prevent idle motor current reduction, then see if you can move the X carriage up and down a few mm by hand
-
I haven't yet counted the teeth….. but some work in a spreadsheet shows that it can't be.
every tooth variance with a gt2 belt = 2mm difference in travel per revolution.
at 150mm of travel I have 3.75 motor revolutions
I assume if it was a tooth count error travel would be out by 7mm+So some further number crunching reveals something interesting.....
Pulley 20T
Motor 0.9deg
mm/rev = 40mm
distance for each full step = 0.1mmfor 150mm of travel I measured 148.55 to 148.6mm actual.
If we round that to 148.5mm for a minute we have 1.5mm of error
Now 1.5/3.75 = 0.4mm per rev of error
mmmm..... this is exactly 4 full stepsthat seems far to coincidental for me.
-
Phil, I really think your X pulley is slipping on the motor shaft, even though I know you have already checked that. Try energising the motors, send M906 I100 to prevent idle motor current reduction, then see if you can move the X carriage up and down a few mm by hand
Ok tried that,
it does not move, not even with a very heavy hand.
Cheers
-
Maybe stepper motors run the other way around in 'stralia.
-
How about if you try to move it out at the extents you were having trouble with? Like at x290 or y290
-
PS - also, please wind back the acceleration and jerk settings to these values and test again:
M201 X1000 Y1000 Z1000 E800 ; Accelerations (mm/s^2)
M566 X600 Y600 Z600 E1200 ; Maximum instant speed changes mm/minuteFurther Update:
Done this no change…Next test: Swap the X and Y cabling and see if issue moves or stays with tower.
Result: (not fully tested, too cold outside) however initial tests shows the issue stays with the tower and this, for me, atm clears any firmware issue.
Further evidence that this is a partially failed/failing stepper is that with power off, moving the X carriage by hand reveals a very clear (hard to describe) rough then smooth movement. rough is what i feel moving the other two carriages at all times. whilst i haven't measured it the rough/smooth cycle looked like 40mm so 1/2 rev is "normal like the others the other 1/2 rev is smooth/less resistive.
-
-
How about if you try to move it out at the extents you were having trouble with? Like at x290 or y290
The more i test/research/discuss and think about this and the more evidence i gather the more it becomes obvious the the moves to the extents enable the last issue of incorrect X movement to be clearly seen rather than a pure cause.
This machine has 674mm long effector arms to cover the print area which means the for the first 100mm of print radius the carriages really don't move a lot.
Think i'll end up parking this till I can replace the motor.
-
Phil
What angle in relation to the bed do you get at full extent with those Arms?
Mine will be 750 long for a similar sized Delta with a slightly smaller bed than yours (around 550 mm dia)
just wonder if your getting close to or below the 20 degree angle that is the recommended lowest angle. if so maybe that is compounding the error's.
(Would be good to know as I haven't cut my rods yet so may be able to reduce them).
Doug
-
Hi Doug,
I just went a did a quick check
Angle at around full extent is approx 30 to 32 degrees.I cheated a little for the long rods, i placed the effector at the outer edge, and measured to the carriage at a 30 deg angle to get the arm length, subtracted the end offsets to then get the cut length.
the final arm length will depend on the carriage design and the arm ends and you may be better to physically measure at your desired min angle.
-
for 150mm of travel I measured 148.55 to 148.6mm actual.
If we round that to 148.5mm for a minute we have 1.5mm of error
Now 1.5/3.75 = 0.4mm per rev of error
mmmm….. this is exactly 4 full stepsthat seems far to coincidental for me.
Try reducing the acceleration?
-
David:
already down to 1000mm/s, 5x less than i normally run.New motor (plus spare) is ordered, I just need to be patient now and wait for them to arrive.
Thanks to all contributors so far in this saga, I feel that without your input this thing would be dismantled by now.
Note that during all this the Duet WiFi has performed flawlessly with next to no timeouts (ajax error).
-
Phil, I suggest you remove the belt from that motor, disconnect it, and check whether turns freely. If not, remove the motor and take out the 4 screws holding it together. Take the rotor out and remove any debris, including magnetic debris clinging to the rotor. Then re-assemble the motor and see if it now turns freely. If it still doesn't, wait for your new motors to arrive.
-
Phil, I suggest you remove the belt from that motor, disconnect it, and check whether turns freely. If not, remove the motor and take out the 4 screws holding it together. Take the rotor out and remove any debris, including magnetic debris clinging to the rotor. Then re-assemble the motor and see if it now turns freely. If it still doesn't, wait for your new motors to arrive.
Not a bad idea
, I was going to pull it apart anyway, it actually moves too easy rather than being sticky.The weather here hasn't been the best with tops under 10c which i find to be freezing so my time outside in the unheated garage with the printer has been minimal the last few days….
The new motors have been shipped and will hopefully arrive by the end of the week so I may as well pull this one out tomorrow.Cheers
-
OK, time for an update.
Murphy paid me a visit as I've been running test prints. The integrated mounting for the IR sensor had become cracked and consequently the sensor was not quite vertical causing some interesting probing results. I suspect but can't prove that due to it's close location to the hot end the same said crack was changing the sensor angle with some heat transfer.
Added some bracing to the mount and printed a replacement on the trusty Flashforge. I'm just waiting on the high temp silicon to cure that ensures the sensor stays in place.The X stepper motor has been replaced and after a tweak of the actual rod length and a reset of the steps/mm back to 160 I now have less than 0.2mm distance error over 500mm of travel in both x and y directions.
I still seem to a have a layer offset happening in the +x+y direction over the first couple of layers but if I didn't specifically look for it I likely wouldn't see it so i'm going to park that for the time being.
What did I find pulling the X stepper out and apart:
Well David was spot on with a loose pulley but it was loose enough to move significantly more that the measured error however never found it by moving the carriage up and down by hand…..
Suspicion is that the belt tension pulling at right angles to the shaft on the 10mm wide pulley was enough to lock it in for relatively slow or consistent moves but the faster moves required at times was enough to cause it to slip...... the puzzle is the slip was absolutely measurably consistent.
Rotating the motor by hand didn't inspire that the motor was the same as the new one so a new one got used anyway.I wrote in post one that the pulleys showed no sign of slippage and this was the case right up till I had the stepper in the hand rotating the shaft by gripping the pulley.
Despite the absolute frustration I have learnt a lot with this exercise both about physical delta movement and all the variables but the firmware configs etc.
Thanks to everyone for your help and support
-
If the motor has a flat on it for set screws then its possible the set screw was slightly loose but not enough to free spin so it would lock at x and y rotations but free-spin between the two sometimes.
