Enable strain gauge Z probe?

[adarcher]

Saw on the new listing for the V3.2 that it's using the strain gauges now.

Are the Artemis effector's going to have that enabled as well?

[Jimustanguitar]

I believe that it's built into the new hotend board, but it's not been implemented in the firmware yet.

[geneb]

It's in the firmware, there's just no official support for it yet.

g.

[adarcher]

Ah ok, do you know if it'll be something we "early adopters" will be able to "turn on" or would we have to order the newest version of the effector board?

[geneb]

I'll make you a deal - when *I* know how it's done, *you'll* know how it's done.

The v1.1 SE300 PCB has the strain gauge on it and the wiring is in the harness, but I haven't had time to chase down the details beyond that. I'll hopefully find out how it's done at MRRF and then I'll get it written up. I DO know that part of the process is removing the FSRs and changing how the bed is mounted. The probe won't work properly if the bed is "floating" like what is required for the FSRs.

g.

[wepollock]

I think hardware leveling could eventually be replaced, with stall detection (stepper drivers and software). .. stallGuard2™ high precision sensorless motor load detection..

this feature might handle bed leveling, replace endstops, and introduce collision detection (which I think should be the first implementation).

With the Artemis the big hardware gains have been had, perhaps it is time to shift more to software.

[dc42]

I think hardware leveling could eventually be replaced, with stall detection (stepper drivers and software). .. stallGuard2™ high precision sensorless motor load detection..

this feature might handle bed leveling, replace endstops, and introduce collision detection (which I think should be the first implementation).

With the Artemis the big hardware gains have been had, perhaps it is time to shift more to software.

Stall detection is low precision, much lower than microswitch endstops. You can use stall detection instead of endstops on a delta if you do one cycle of at least 3-point auto calibration after homing; but it's nowhere near accurate enough to do bed probing. See https://duet3d.dozuki.com/Wiki/Stall_de ... ess_homing.

[wepollock]

DC.. to me the collision detection would be a real plus.

[dc42]

DC.. to me the collision detection would be a real plus.

Collision detection on Delta printers is difficult for two reasons. One is that delta printers are usually designed to print fast. That means that low inductance motors are chosen, which produce less back emf for the motor driver to measure. It also means that high acceleration and jerk are used, which tend to trigger the stall detection. So you have a choice: high speed, or collision detection.

The other reason is that at some positions, one or two of the carriage motors may be moving very slowly, and stall detection doesn't work at low speeds.

[wepollock]

I guess in the future there will be other ways to get "positional feedback;" where there would be an error-check between the expected position and the actual position.

Just thinking about it, doing is harder than thinking for sure.

perhaps a single tower could contain (and read) a light-read timing stripe. The count on the timing stripe could then be compared with expected position.

I have seen some robotic projects which use a potentiometer to provide positional feedback; here a wheel with a rotary encoder might travel with the linear rails. For my scant programming skilz the rotary encoder was a bitch.

[geneb]

Closed loop positioning would be pretty spendy to do right.

g.

[adarcher]

I'll make you a deal - when *I* know how it's done, *you'll* know how it's done.

The v1.1 SE300 PCB has the strain gauge on it and the wiring is in the harness, but I haven't had time to chase down the details beyond that. I'll hopefully find out how it's done at MRRF and then I'll get it written up. I DO know that part of the process is removing the FSRs and changing how the bed is mounted. The probe won't work properly if the bed is "floating" like what is required for the FSRs.

g.

Thanks! I'll keep a watch for the this thread's updates or your thread about it

[dmo]

You could always download the configuration files and the bed configuration file and compare them. I find my calibration results are identical between the FSR probing on the Artemis and my v3 using the Duet board and accelerometer. About .02 sometimes .015.

[adarcher]

You could always download the configuration files and the bed configuration file and compare them. I find my calibration results are identical between the FSR probing on the Artemis and my v3 using the Duet board and accelerometer. About .02 sometimes .015.

I'd like to move away from the frs because I'd like to run the calibration points at actual printing temperatures. Currently, the FRS fails on my Artemis at anything above 40C on the bed.

[adarcher]

Any news on this?

I see the repo has the strain gauge set as default now...

I'll start looking at the differences in the construction docs, I know I've already got the wires up at the top, just don't know where to connect them.

Probably need to make harder mounts for the bed as well.

[wepollock]

My current thinking about "auto leveling" is that the Artemis being very ridged, and the bed being very flat; that there is really no reason to auto level. On the rostock max you have to correct for differences in tower height by fiddling with the endstop screws. Here you can do that in software. The biggest deviation between towers on my machine is 1.2mm. With that adjustment Doing a bed map the largest offset I have is .04!

IMHO the best way to get the right Zero setting is with a .2mm gauge. Especially so if you want to use a heated bed.

All I need to do is slip a gauge under the nozzle in the center, and change my offset to a level where I am happy. There is really no need to do a bed level.


; delta radius, diagonal rod length, printable radius and homed height
; M665 R150 L351.155 B145 H530X0 Y0 Z0
M665 R150 L351.155 B145 H466.5 X0 Y0 Z0 ; Titan areo to zero

; Y X Z are tower angle offsets
M666 X0.707 Y-1.915 Z1.208 A0.00 B0.00

[geneb]

Understand that there's a difference between auto-level and auto-calibration. Broadly speaking the terms can be interchangeable, but when you're into the weeds like this, it's important to know which is which.

What G32 does is auto-calibration. The series of probes it does allows a solver to ensure that the effector platform is traveling perfectly in a flat plane.

Auto-level really isn't a thing with delta configuration printers. The constructed geometry of the machine is such that barring any gross build errors, the bed is assumed to be flat - it's bolted down after all. That being said, there is a method for compensating for a bed that appears to be out of level - that's mesh bed compensation. What that does is build a map of the bed and then uses that map to wash out any gross surface issues with the bed.

When you create a bed map on the Artemis, you should get a heat map at the end that is mostly green. If you don't see that then there's one of two issues going on - the bed isn't parallel with the travel plane of the effector, or you've got a z-probe issue.

Can you run a mesh map and then post the image it gives you here? I'd like to see what that mesh looks like.

Note that to use the strain gauge, the FSRs have to be removed and the bed has to be solidly bolted down. I don't know what hardware is involved in that.

g.

[wepollock]

Good to know that G32 is really checking how the effector travels in a plane. There is not much error there.
The elements involved in moving the effector are the least rigid - belts - arms (they are pretty ridgid). With a extruder mounted on the effector I am introducing more mass into the most flexible parts of the machine.

In Re Mesh Bed -- the outliers are less than 1/10th of a millimeter with most being within 5/100ths..

RepRapFirmware height map file v2 generated at 2018-02-24 23:08, mean error 0.004, deviation 0.036
xmin,xmax,ymin,ymax,radius,xspacing,yspacing,xnum,ynum
-120.00,120.10,-120.00,120.10,140.00,30.00,30.00,9,9
0, 0, 0.026, -0.028, -0.095, -0.031, -0.048, 0, 0
0, 0.056, 0.030, 0.020, -0.028, -0.030, -0.064, 0.003, 0
0.019, 0.007, 0.020, 0.002, -0.031, 0.001, -0.003, 0.033, 0.100
0.011, 0.020, 0.005, -0.019, -0.049, 0.020, 0.010, 0.045, 0.066
-0.058, -0.031, -0.027, -0.016, -0.027, 0.025, 0.059, 0.075, 0.074
-0.010, -0.011, -0.003, -0.021, -0.010, -0.037, 0.011, 0.056, 0.079
-0.039, -0.045, -0.011, 0.019, 0.034, 0.024, -0.006, -0.015, 0.010
0, 0.003, 0.010, 0.016, 0.034, 0.001, -0.025, -0.027, 0
0, 0, 0.039, 0.019, 0.016, 0.014, -0.011, 0, 0

If anybody can do a better job (than Seemecnc) building machines, let me know!

[adarcher]

I went ahead and played with this a bit ago.

I unplugged the FSR cable
Plugged in the "tucked away" probe cable
Found some small "washers" to clamp the heat spreader to the metal frame under the bed clips
Tightened it up nicely
Changed over to the probe config.g settings and added a multiple-attempt value (also made it wait less time)

Seems to work really well so far. Not perfect, but at least for me, far and away better than the FSR setup.

Here is my modified line:
I added the A6 for 6 attempts and T15000 to speed up the movement between probe points. Also, I reduced R0.4 to R0.1 because I was impatient that day, but it never seemed to be a problem so I kept it there.

Code: Select all

;STRAIN GAGE PROBE
M558 P5 R0.1 F2500 H20 A6 T15000           ; Strain gauge probe settings

Previously I would get G32 values between .037 to .07 and now it's always lower than .032.

I'll make some better, more solid, mounts for the thing later.

[wepollock]

Great to hear that you fixed the bed leveling concern.

Are you using the four spare "probe" wires?

Also, how does the "skirt brim" feel? <--to me that subjective indicator as to whether or not I will have a dialed in first layer.

I would be interested in finding out how the ir sensor works.

I was getting good prints before, but after converting (the Artemis and Rostock Max2) to direct-drive I am getting prints with details such threads that are terrific both in PLA and PETG. I also printed out a tray of small parts that I would have printed out separately on the Rostock.. I was using my "paperweight"-cartesian printer for small parts; now the cartesian is really a paperweight. On the Rostock there is some "backlash (may not be the correct usage) on very dense infill, the printer is just not as firm as the Artemis. I have seen folks mod the daylights out of the Rostock. Yet, I think the Artemis presents a better platform for modding. The reduction of noise has improved the harmony of the household. Folks looking for tunkey might be the main target customer.

[adarcher]

I'm using those 4 spare probe wires. I don't remember if they came naked or in the connector, but there were bundled up there already in a connector--plugged it in and blam-o, first try worked.

Since then, I've only hit the max probe count for a single point once, and it was because I left a failed print attached to the nozzle.

I'd post some pictures, but currently, I'm using some chopped up PCB that had the correct holes for the bolts close to an edge. All my metal washers were too wide and would have interfered with the glass bed.

I've debated on getting a titan aero (seems to be the lightest of the direct drive options). I'll see what the future holds. I like that you are doing all the research for me, thanks!

[Infringer]

Those four wires that the directions tell you to bundle up in the wire management are the Z probe wires for the Artemis machines made without FSR (or is it FRS). I just put mine together and was getting a Z probe error until I hooked up those four wires to the probe port on the Duet board.

K

[PartDaddy]

I did not have time To red everything here. But if you are switching from FSR to The Strain gauge hot in the probe then you must rigidly mount your bed or it will not work.