Piezo20 probe and piezo kit now available
-
That's a good idea Idris. I'm just going to try that now, it will definitely work if the hotend is cold and it just gives it a tiny tug.
Doesn't seem to pull hard enough to really see any rise on the probe value.
Hi,
Yes very good idea!
If you want look at this company about accelerometer (it's in FACT very small piezo) kit very easy to integrate in Any hotend and duet wifi board…https://learn.mikroe.com/mems-sensors-conversion-physical-world-digital-world/
Best regards,
-
That's a good idea Idris. I'm just going to try that now, it will definitely work if the hotend is cold and it just gives it a tiny tug.
Doesn't seem to pull hard enough to really see any rise on the probe value.
Hmm, that's a shame, what's your retract speed? It should work in either direction, maybe you can apply more force by extruding, or build up pressure in the bowden tube then release it quickly. It's be great if it would work as it wouldn't require any hardware changes.
Just had a thought, you could prove the principal by manually pushing the filament at the extruder but with the extruder disengaged. Of course this wont probably work if your bowden tube coupling is in the hot end itself as there's nothing for the force to act against. I think this will only work if the bowden tube terminates on the end effector.
Idris
-
Have you tried a short fast downward head move that is rapidly reversed?
-
Moriquendi - I tried with cold extrusion allowed, 0.5mm at 80 mm/s, I can do this through the web interface console by manually ntering the commands, but if I write a macro
M302 P1
G1 E-0.5 F5000
G1 E0.5 F5000
M302 P0The console states, "unable to perform move due to heater fault" or something similar. So not sure why that is?
David - Not yet but I will.
-
The firmware probably has cold extrusion protection, I think marlin had something similar when I was using a RAMPS, you may be able to turn it off somewhere.
A sharp Z movement of the head would work for delta printers but it wouldn't work for printers that don't move the head on the z-axis, or that can't move the z-axis rapidly.
Idris
-
I tried turning cold extrusion protection off, which I can do from console, and then the extruder will move, but with a macro it ignores the cold extrusion enable and gets an error.
.
I will try the head up/down see if that helps. It would work on Duet as you can see the sensor value but on Ramps or other firmwares, how do we detect a trigger if we're not probing, agreed this is not useful if the head is not moving in Z. -
Moriquendi - I tried with cold extrusion allowed, 0.5mm at 80 mm/s, I can do this through the web interface console by manually ntering the commands, but if I write a macro
M302 P1
G1 E-0.5 F5000
G1 E0.5 F5000
M302 P0The console states, "unable to perform move due to heater fault" or something similar. So not sure why that is?
David - Not yet but I will.
Hi,
In Marlin M302 work like that
M302 ; report current cold extrusion state
M302 P0 ; enable cold extrusion checking
M302 P1 ; disable cold extrusion checking
M302 S0 ; always allow extrusion (disable checking)
M302 S170 ; only allow extrusion above 170
M302 S170 P1 ; set min extrude temp to 170 but leave disabledSo try that
M302 S0 P0
G1 E1 F5000
G1 E-1 F5000
M302 S180 P1if not work and you just want to test if you piezo sensor work you can just move fast X and Y axe and make A simple circle of 10 cm of diameter and move Z axis in the same Time up and down from 0.2mm and see the réaction of sensor. Storage the result and after compare it for futur starting configuration after G28?
-
Okay so I have tried:
M302 P1
G1 E-2 F15000Nothing really shows on the sensor output.
I also tried a macro:
G28
G91
G1 Z-50 F1500
G1 z3 F15000
G1 z-3 F15000
G90
G28Again nothing really changes on the sensor value.
However just moving z -100 causes the sensor value to rise a little if the sensitivity on VR1 is set quite high. If its a little lower not much happens but I can go down to z=10 and ask for a G30 and it probes and stops at z=0. I have taken to probing with motor current at 500mA to limit the damage if I get a head crash, which has been zero on the corexy, and occasionally on the kossel XL probably due to a loose wire.
Maybe someone else can get one of these methods to work?
For now I am manually checking the probe value on web interface or paneldue, in my setup 216 at startup, I had a slightly dodgy connection on signal wire at the piezo board, (one of my leads was loose), and the sensor value read 534. So you can easily see (on Duet at least) if you have lost contact with the piezo board. But so far (and I haven't tried the back to back sensor yet for lack of time and since both machines are working really well a slight lack of enthusiasm for taking one apart), its a manual push up on the hotend to test before beginning, to check LED changes on the piezo board and a read of the sensor value open and when given a firm push.
-
I was also thinking about tesla_power's idea to put a standard bowden pushfit on the top piece, this will work fine, since these don't allow the bowden to pass through like the genuine E3D bowden fasteners, then you'll need a PTFE guide tube coming down from the top piece through the piezo and into the hotend. This will not limit the sensor's movement. For now anyone wanting to do this is best to just print the mkIII and drill out the top piece centre-hole for whatever thread size their generic bowden coupler takes and screw one in there.
-
Okay so I have tried:
M302 P1
G1 E-2 F15000Nothing really shows on the sensor output.
I also tried a macro:
G28
G91
G1 Z-50 F1500
G1 z3 F15000
G1 z-3 F15000
G90
G28Again nothing really changes on the sensor value.
However just moving z -100 causes the sensor value to rise a little if the sensitivity on VR1 is set quite high. If its a little lower not much happens but I can go down to z=10 and ask for a G30 and it probes and stops at z=0. I have taken to probing with motor current at 500mA to limit the damage if I get a head crash, which has been zero on the corexy, and occasionally on the kossel XL probably due to a loose wire.
Maybe someone else can get one of these methods to work?
For now I am manually checking the probe value on web interface or paneldue, in my setup 216 at startup, I had a slightly dodgy connection on signal wire at the piezo board, (one of my leads was loose), and the sensor value read 534. So you can easily see (on Duet at least) if you have lost contact with the piezo board. But so far (and I haven't tried the back to back sensor yet for lack of time and since both machines are working really well a slight lack of enthusiasm for taking one apart), its a manual push up on the hotend to test before beginning, to check LED changes on the piezo board and a read of the sensor value open and when given a firm push.
Hi,
You make mistake
M302 P1 (you disable cold extrusion)
G1 E-2 F15000Try that
M302 P0
G1 E-2 F15000Best regards,
-
No I don't think thats right unless the console messages are incorrect somehow, this is what my console shows:
18:15:10 M302
Cold extrusion is allowed, use M302 P[1/0] to allow/deny it
18:15:08 M302 P1
18:15:02 M302
Cold extrusion is denied, use M302 P[1/0] to allow/deny itThe first command (at the bottom) M302 just returns the status, i..e cold extrusion denied.
I then send M302 P1
Then M302 to check status and get cold extrusion allowed.I can then send a fast retract to pull on the filament, which it does, but I don't get any noticeable change on the probe value on web control. It was only when I used a macro I was getting errors. I tried it again just now and the macro ran okay, but no change in probe status, so despite it appearing that this would be a good test it isnt! Neither is doing a violent up/down in z.
So for failsafe/testing its either going to be using back to back piezos (with the current board), which I should try tonight (I will reprint a mkIII top plate with a bigger recess for two piezos not one. Or we need some electronic solution like the multi input piezo board thats in the works. Anyway this is a minor issue, inductive NPN NO sensors have no failsafe and thousands of people use them either properly by doing an M119 and sticking a spanner under it before a printing session, or by ignoring this and every x hundred startups get a head crash.
In any case question for the electronics experts. On the kossel XL the piezo is wired with red-positive and black-negative going into the piezo board, normal operation. On my corexy I noticed the polarity is reversed (same brand piezo) but I also get normal operation - if I reverse it on either machine it doesn't work. So why does one setup happily work with the piezo in reverse polarity compared to the other?
-
No I don't think thats right unless the console messages are incorrect somehow, this is what my console shows:
18:15:10 M302
Cold extrusion is allowed, use M302 P[1/0] to allow/deny it
18:15:08 M302 P1
18:15:02 M302
Cold extrusion is denied, use M302 P[1/0] to allow/deny itThe first command (at the bottom) M302 just returns the status, i..e cold extrusion denied.
I then send M302 P1
Then M302 to check status and get cold extrusion allowed.I can then send a fast retract to pull on the filament, which it does, but I don't get any noticeable change on the probe value on web control. It was only when I used a macro I was getting errors. I tried it again just now and the macro ran okay, but no change in probe status, so despite it appearing that this would be a good test it isnt! Neither is doing a violent up/down in z.
So for failsafe/testing its either going to be using back to back piezos (with the current board), which I should try tonight (I will reprint a mkIII top plate with a bigger recess for two piezos not one. Or we need some electronic solution like the multi input piezo board thats in the works. Anyway this is a minor issue, inductive NPN NO sensors have no failsafe and thousands of people use them either properly by doing an M119 and sticking a spanner under it before a printing session, or by ignoring this and every x hundred startups get a head crash.
In any case question for the electronics experts. On the kossel XL the piezo is wired with red-positive and black-negative going into the piezo board, normal operation. On my corexy I noticed the polarity is reversed (same brand piezo) but I also get normal operation - if I reverse it on either machine it doesn't work. So why does one setup happily work with the piezo in reverse polarity compared to the other?
Hi,
The respond from A piezo sensor in the positive voltage or negative voltage is not the same curve and i think in negative voltage you have very fast voltage for very small variation of movement!
-
For me the Best sensor that you have already exist in every printer but nobody use it!!!! It's to Check the consumption of current of the Z stepper motor! It's very easy but here We need to have A high sensitive one and So the only wait is to monitor 1000 Times the consumption of the Z motor and make A average Min and Max and in the software compare each Time these values with Max and min but in micro ampere way!! Someone is ready to make some test?
-
Hi tesla_power the comment about positive and negative is something I will look into more, since there will be a preferred curve for what we are doing here, so if you know more about this please tell me more….
I totally agree with you about the stepper motors and in fact I have heard David speaking about it in the past. The TMC2660's I believe have stall detection and other features which with combined with using low motor current to probe, might be able to probe the bed. That would be great and I agree depending on if it works well and is sensitive enough would be the closest to an ideal probe so far, far closer than this idea.
In other news, many hours in the lab today. I tried out the back-to-back piezo approach and it does not work (for me anyway). The two piezo discs when pressed together become a much stiffer unit. As such the impact required to trigger them in this way is too high and the signal is lost in the noise as you have to turn the sensitivity up so high to get a signal at all. So as far as I am concerned its a nice idea but it takes the system too far outside of the sweet spot for it to function as it is currently. Perhaps the clamp could be redesigned to have a much narrower top surface which reduces the area pressing against the piezo, giving it a better chance of bending it. But for me this is a dead end. If anyone fancies seeing if a back-to-back approach works. I made a little set of wiring so I could electrically connect piezo 1, piezo 2 or both to see the effect it had on function, this would seem a sensible approach.
So to salvage a little joy from defeat I did come up with a new look for the top piece. I liked the way Sakey's implementation (page 6) looked with the piezo exposed so made a new top piece with "viewing" windows in it, so the shiny brass transducer can be seen through it. You will only see it If you are using Lykle's effector or an effector design with a hole in the middle.
-
I like the viewing window!
I think I've thought of a way to support the bowden tube without too much stress on the piezo(if any). Will model and print it when I get chance at the weekend (or next!) and hope its strong enough.One thing I have noticed I don't get as successful probing with the PEI covered bed compared to when I have it glass side up. It seems the PEI absorbs the hit on the bed a little delaying the trigger? Don't know if this is something others have seen?
-
Hi tesla_power the comment about positive and negative is something I will look into more, since there will be a preferred curve for what we are doing here, so if you know more about this please tell me more….
I totally agree with you about the stepper motors and in fact I have heard David speaking about it in the past. The TMC2660's I believe have stall detection and other features which with combined with using low motor current to probe, might be able to probe the bed. That would be great and I agree depending on if it works well and is sensitive enough would be the closest to an ideal probe so far, far closer than this idea.
In other news, many hours in the lab today. I tried out the back-to-back piezo approach and it does not work (for me anyway). The two piezo discs when pressed together become a much stiffer unit. As such the impact required to trigger them in this way is too high and the signal is lost in the noise as you have to turn the sensitivity up so high to get a signal at all. So as far as I am concerned its a nice idea but it takes the system too far outside of the sweet spot for it to function as it is currently. Perhaps the clamp could be redesigned to have a much narrower top surface which reduces the area pressing against the piezo, giving it a better chance of bending it. But for me this is a dead end. If anyone fancies seeing if a back-to-back approach works. I made a little set of wiring so I could electrically connect piezo 1, piezo 2 or both to see the effect it had on function, this would seem a sensible approach.
So to salvage a little joy from defeat I did come up with a new look for the top piece. I liked the way Sakey's implementation (page 6) looked with the piezo exposed so made a new top piece with "viewing" windows in it, so the shiny brass transducer can be seen through it. You will only see it If you are using Lykle's effector or an effector design with a hole in the middle.
Hi again,
I understand that the piezo sensor is dead So i Will take ALL my Time on current sense!
My idea is build A ready to use stepper motor in close loop mode and 256 microstep and wifi communication without ANY end stop or mesh control Bed!!!
https://www.trinamic.com/fileadmin/assets/Products/ICs_Documents/TMC2660_datasheet.pdf
Look the pdf at the point 4 but more at the point 4.1!!! I don't understand why nobody have use this possibility before??? It's amazing option ready to use to have 3D printer without end stop and have the posibilty to make an CONTINOUS BED leveling during not just the first layer by touching 3 or 4 points (linear map of the Bed) or 9 points (curve map of the Bed - prusa mk2) but have A complete map for every 0.1mm of each X and Y and every 0.05mm Z point during ALL the Printing layer!!!
Imagine the quality of object print and every point Will respect the size of the quality of your print that you choose!!!
I need you and maxi open source Guy that need to put at the maximum the capacity to manage the stallGuard2 function of the TMC 2660 driver or make our Own back coil feedback sensor..,
Good works to everybody…
-
The stall guard signal from the TMC2660 is only updated once every full step and it doesn't work at low speeds. So it's not usable for bed probing. It might be usable in place of endstop switches on a delta printer, if you then use auto calibration with a Z probe to work around the inaccuracy of homing with a resolution of only 1 full step.
-
So one full step on a 0.9 deg motor with 16t pulleys and 2mm belts would be 5 microns in the centre of the bed, which would be twice as accurate as the IR probe or piezo probe which are both around 10 micron accuracy, but with 1.8 deg motors there's no accuracy advantage and near the edge of the bed accuracy would dramatically decrease?
Plus the speed would have to be too high (even with very low motor current?) to make it a useful probe?
-
So one full step on a 0.9 deg motor with 16t pulleys and 2mm belts would be 5 microns in the centre of the bed, which would be twice as accurate as the IR probe or piezo probe which are both around 10 micron accuracy, but with 1.8 deg motors there's no accuracy advantage and near the edge of the bed accuracy would dramatically decrease?
No, 1 full step with 0.9deg and 16t would be 80 microns.
-
Yes, I see, the online calculators presume 1/16th microstepping, yes 32mm for 1 rotation = 0.08mm! So using the stepper drivers is not a useful approach at all. Best get back to improving this piezo unit then.
