consistent layer shift in both x/y using clear-path servos ?

[nhof]

@thatbigprint Please post a picture of the error you are seeing on the print.

With servo motors it may be possible to up the resolution depending on the encoder resolution of the motor and available configuration settings.

For example, my setup is like so.
My motors have 20,000 cpr encoders (input configured to 10,000 steps/rev in motor driver config)
I have fairly aggressive lead screws with a lead of 25mm (nut will move 25mm per complete turn of screw)
So I just divide (10,000steps/rev)/(25mm/rev) and it comes out to 400 steps/mm.
This gives me a theoretical resolution of 0.0025mm (assuming perfect mechanics and motor response)

To contrast are running at 28 steps/mm which gives a max resolution of 0.04mm which is quite a bit lower. I expect your motors can do higher resolution than that but will need some configuration to achieve this.

Look in your motor config software or manual for the encoder resolution and how far it will move for each step pulse, this is likely called Input Resolution or Electronic Gearing or something similar. I think the Teknic ones come with 800cpr or 16,000cpr encoders depending on the model. You would do well to read the data sheet for your motors as it will have all this information specific to your model.

You may also need to tune the PID parameteres of your motor. A poorly tuned motor may have a slow response, oscillate around the setpoint, etc which can cause poor quality prints as well. Check the motor manual for how to do this or look on youtube, there are multiple PID tuning guides available.

Hope this helps, let me know if I can explain anything in more detail.

[timothyz]

@thatbigprint

i made an attempt at using clearpaths (not on a duet, but with a mightyboard) and I was able to get it functional, but was never able to get rid of servo tuning related issues. I think the direction you should go (once you get layer shifting figured out) is to have a 2:1 reduction or greater off of the motor before your drive pulley. This should allow for sharper tuning.

[dc42]

@thatbigprint said in consistent layer shift in both x/y using clear-path servos ?:

@dc42 for the z i have skipped the enable and soldered into 5v so the z is always enabled i will probs do this to the rest in that case ?

When you upgrade to firmware 2.02 you will get warnings that the motors are disconnected.

[ThatBigPrint]

@nhof apologies for the late reply i have sorted most of the surface finish issues (single wall with an 8mm nozzle and doing infill first) and as for motors mine are 800 count or the lesser model... and for tuning, I had locked down the opposite motor shaft and run the auto-tune feature they seem to be behaving z-axis is a tiny bit noisy but I can tune that in better later this weekend when I have uninterrupted time i think its duet to weight and the fact its belt drive through a jackshaft (totally unideal but easy/cheap for a super quick turn around) but in theory i can wack up the motor counts per rev and have it actually be more accurate ? i had initially thought this would just make the motors always hunt too much?

@timothyz i will be going to a planetary reduction gearbox on all axis although I'm apprehensive due to clearpaths odd choice of motor shaft? the jackshaft on z was a last min thing as a workaround seems to be doing ok so far for a printed pulley moving around 20mm of tooling plate

@dc42 will this be the case for the super early duet wifis ? as in ones without motor feedback? and if not will there be a firmware workaround ? or just don't upgrade?

[dc42]

Why don't you connect the servo drives as we instruct on the wiki page instead?

[ThatBigPrint]

@dc42 apologies for late reply reason for doing this is because it looks cleaner in terms of cable management but if it is causing me aforementioned issues I will go to flat flex expansion header

so i am still having the issue where the bed for whatever reason is not remaining level ie goes through probing etc starts printing (you can see it adjusting z) suggesting its trying to work but one end for the print (600x600x20mm) the layer fades out completely and the other end the layer is fine ?

below are my current settings gone way higher on the t values to no avail?

Taa:bb:cc:dd (firmware 1.21 and later) Minimum driver step pulse width, step pulse interval, direction setup time and direction hold time, in microseconds

; Drives
M569 P0 S1 T10:5:1:1
M569 P1 S1 T10:5:1:1
M569 P2 S0 T10:5:1:1
M569 P3 S0
M350 X16 Y16 Z16 E16 ; Configure microstepping with interpolation
M92 X26.676 Y26.676 Z89.233 E804.81 ; Set steps per mm
M566 X1500 Y1500 Z900 E1500 ; Set maximum instantaneous speed changes (mm/min)
M203 X6000 Y6000 Z2000 E2000 ; Set maximum speeds (mm/min)
M201 X800 Y800 Z500 E250 ; Set accelerations (mm/s^2)
M906 X800 Y800 Z800 E800 I30 ; Set motor currents (mA) and motor idle factor in per cent
M84 S30 ; Set idle timeout

getting pretty desperate now as work needed to go to another company to get 2 done at a huge cost to them

layer hight may also be inaccurate but not sure on that as the test cube i print appears to be over extruding but the actual extrusion is spot on ?

also, I have looked at ways of testing the repeatability of the bed probe but not much info for the duet i did try the.....

;prob testing single point.
G30 P0 X200 Y200 Z-99999 F2000
G30 P1 X200 Y200 Z-99999 F2000
G30 P2 X200 Y200 Z-99999 F2000

method but it errors out at the end?

[nhof]

Could be an issue with the probing / compensation rather than the motor drive at this point.

Maybe try disabling the auto level and try leveling the bed with a shim or a indicator to see if you can narrow the problem down.

The reason I mention this is because I have had problems in the past with my BL Touch probes creating inconsistent height maps where it looked like one side of the bed was way higher than the other, while my dial indicator indicated that everything was actually flat.

[ThatBigPrint]

@nhof i have moved away from grid and gone to 5 point bed level now works fine think it was the probe picking up the edge of the printbite ?

[dc42]

@thatbigprint said in consistent layer shift in both x/y using clear-path servos ?:

@nhof i have moved away from grid and gone to 5 point bed level now works fine think it was the probe picking up the edge of the printbite ?

Try a smaller mesh then?

[ThatBigPrint]

@dc42 tried a smaller constrained mesh but no luck its ok for now using 5 point level as the plate is flat tooling plate just how its mounted will not allow for adjustment prints are coming out perfectly flat now

just need to work on quality now

[ThatBigPrint]

Right guys i have revisited the printer so things i have done so far

added io expansion board in a sinking 5v style control for step direction and enable

tried adjusting the m569 values

but i get the same issue the centre of the bed is closer to the nozzle than the perimeter i have mechanically levelled bed as best as i can the z motor has now got significant gear reduction now dimensionally the printer movers are accurate as individual entities but both mesh levelling and bed levelling are not doing as intended.... the motor is moving as though its attempting to level but it feels a bit rough in its movements as in it jerks not smooth

not sure what to try next i have had better success with t values all set to 4 but this adds ringing.... i have used recommended settings suggested by mr DC42

"The setup time is 25ns not 25us, so instead of 1:1:25:1 I suggest 1:1:0.5:0 in the M569 T parameter. My suggestion to use 0.5us setup time is to allow for possible skew in the opto isolators."

but i get constant skipped layers in x and y immediately as in shell is printed and fill is not within the boundary

[dc42]

@thatbigprint said in consistent layer shift in both x/y using clear-path servos ?:

"The setup time is 25ns not 25us, so instead of 1:1:25:1 I suggest 1:1:0.5:0 in the M569 T parameter. My suggestion to use 0.5us setup time is to allow for possible skew in the opto isolators."
but i get constant skipped layers in x and y immediately as in shell is printed and fill is not within the boundary

If 1:1:25:0 works but 1:1:0.5:0 doesn't work, try something like 1:1:1:0 or 1:1:2:0.

[nhof]

@thatbigprint Can you also shoot a picture of your mechanical setup including how your probe is physically mounted to the machine?

I had major problems with the probe creating 'tilted' height maps on a large printer and it turned out the X axis was twisting slightly across the stroke which caused the probe reading to vary significantly (0.5mm+) from side to side.

[ThatBigPrint]

Thanks both of you i have got it to perform better now setting t values to

T1.25:1.25:1.25:1.25 what is the end value for the one that is zero in aforementioned example ?

If 1:1:25:0 works but 1:1:0.5:0 doesn't work, try something like 1:1:1:0 or 1:1:2:0.

i thought it was set up time or stop delay ?

[ThatBigPrint]

@nhof how do i upload pictures ?

[dc42]

The last one is direction hold time after the step pulse leading edge.

[ThatBigPrint]

@dc42 said in consistent layer shift in both x/y using clear-path servos ?:

The last one is direction hold time after the step pulse leading edge.

would this being set to zero previously cause the issues i was having ?

[dc42]

@thatbigprint said in consistent layer shift in both x/y using clear-path servos ?:

@dc42 said in consistent layer shift in both x/y using clear-path servos ?:

The last one is direction hold time after the step pulse leading edge.

would this being set to zero previously cause the issues i was having ?

It could do, depending on the requirements of your driver.