Piezo Z-probe Mk III Beta version now available

Avatar

By DjDemonD 15 January 2017 15:11

Member · 415 comments

It's worth trying but I wonder if the Piezo will flex enough, and whether the hotend will wobble but I'm happy to be proved wrong.

Avatar

By dc42 15 January 2017 16:49

Administrator · 2,297 comments

I think it's a question of getting the balance right. If the piezo disc is large (i.e. inner diameter of F is much larger than outer diameter of A), it will produce a large output but the head may wobble. If it is small (inner diameter of F not much larger than outer diameter of A), it will be much more rigid but produce a smaller output. I think the same will apply to any design, except for designs that use some other means to constrain the relative movement of the hot end and the effector to be in the Z direction only.

The diameter A is about 16mm on my E3Dv6 and I expect the arrangement to be very sensitive because of the direct contact between the hot end and the piezo. That's why I think a piezo for which the outer diameter of the element (which should be slightly smaller than the inner diameter of F) is 20mm may be large enough. But I could be completely wrong.

There is also the risk that a high-speed head crash could break the piezo element. I don't think this is very likely because of the brass backing. However, the ring F could be made to come over the top of the inner 3 screws so that it acts as a stop if the hot end gets pushed too high.

I guess another option that avoids tapping the E3D and drilling more than 1 hole in the piezo is to epoxy the brass substrate of the piezo to the top of the insulator, and epoxy the bottom of the insulator to the top of the hot end heatsink. Bonding epoxy to aluminium can be tricky, but I have read that sanding followed by wiping with IPA helps, so does sanding followed by etching in citric acid.

PS - thinking about it some more, it may be acceptable to bond or screw the piezo element directly to the top of the hot end heatsink, and design the electronics to survive a short between the piezo and +24V.


Duet WiFi hardware designer and firmware engineer
http://www.escher3d.com
https://miscsolutions.wordpress.com

Avatar

By T3P3Tony 15 January 2017 19:13

Administrator · 705 comments

I think in the long term we will want to go with screw mounting, rather than bonding, as it is easier to assemble and easier toreplace a malfunctioning part


Duet Wifi Hardware Designer
www.duet3d.com
www.think3dprint3d.com

Avatar

By DjDemonD 15 January 2017 19:41

Member · 415 comments

This is very encouraging, I am sure various efforts will probably coalesce into a single design we can all benefit from. There's a good bit of long term testing to be done, Leadinglights and Moriquendi have got a good few months if not 6 or more on their under-bed piezo system which places more demand on the piezo transducers than hotend mounting (under normal conditions anyway).

Avatar

By Moriquendi 15 January 2017 19:47

Member · 39 comments

Hi all,

I've followed DjDemonDs link over from reprap. I have a couple of boards if anyone wants them.

I think it's worth mentioning with respect to mounting hot ends that the piezo effect is based on a change in pressure rather than the pressure itself. The signal you get changing from 10N to 20N of force is  the same as you would get going from 100N to 110N.  To put it another way, a steady pressure on the element produces no output only a change in pressure. I see no reason why you couldn't clamp the piezo element rigidly between the hot end and the carriage/effector, the element should respond to the additional force of a z-probe event just as well.

Moriquendi

Avatar

By T3P3Tony 15 January 2017 19:50

Administrator · 705 comments

Hey Moriquendi, welcome to the forum and thanks for the clarification. Are your boards documented anywhere?


Duet Wifi Hardware Designer
www.duet3d.com
www.think3dprint3d.com

Avatar

By Moriquendi 15 January 2017 20:19

Member · 39 comments

The eagle files with a BOM and basic tuning instructions can be found herehttps://www.dropbox.com/sh/3q38ew98kt7f … jm2Ra?dl=0 the original circuit design was LeadingLights, I tuned it to my needs and put together the PCB design

I got 12 boards made up and offered to post the spares to anyone in the UK in exchange for a £5 donation to http://www.great-dane-care.org/index.html a charity of which I'm a trustee. That offer stands for the remaining boards that I have (four I think at this point). Please let me know if I'm breaking forum rules offering these, to be clear, I'm not making anything personally from these boards.

If there is demand I can make more though I have a new PCB design that is easier to make (components on one side only) and easier to tune (trimmers actually visible to the naked eye)

Moriquendi

Avatar

By DjDemonD 15 January 2017 20:25

Member · 415 comments

This is how my effort works I clamp the piezo element fairly tight in between the two halves of the assembly eliminating the wobbly nozzle, which is what you cannot do with fsr's or microswitches.

Avatar

By deckingman 15 January 2017 20:44

Member · 715 comments

Just a thought guys. I've seen people using magnets to hold the hot end to the effector. I think the idea is that in the event of crash, the hot and just breaks away before any damage occurs but it also allows for rapid changing of the hot end. Could you use a similar arrangement but with a Piezo sandwiched between the magnets? So the magnets exert a constant force which is broken (changes) when the nozzle touches the bed? I don't have a Delta so have no idea how you guys mount the hot end. Also I don't know if a negative change in pressure works the same way as a positive change, hence "just a thought".


Avatar

By dc42 15 January 2017 20:53

Administrator · 2,297 comments
Moriquendi wrote

Hi all,

I've followed DjDemonDs link over from reprap. I have a couple of boards if anyone wants them.

I think it's worth mentioning with respect to mounting hot ends that the piezo effect is based on a change in pressure rather than the pressure itself. The signal you get changing from 10N to 20N of force is  the same as you would get going from 100N to 110N.  To put it another way, a steady pressure on the element produces no output only a change in pressure. I see no reason why you couldn't clamp the piezo element rigidly between the hot end and the carriage/effector, the element should respond to the additional force of a z-probe event just as well.

Moriquendi

Hi Moriquendi, thanks for joining in this discussion.

What you say is true of cylindrical piezo elements, but not of bimorphs, which are the type of element I believe we are talking about. Bimorphs are made from two oppositely-polarised sheets stuck together. They respond to being flexed. When you flex them, one sheet has to expand, the other has to contract, so you get a voltage between the two faces. They are far more sensitive to small movements than piezo cylinders, but of course not nearly as rigid. Conversely, they produce far more movement for a give voltage than a cylinder does, which is why they are used in buzzer elements. When I was working with CO2 lasers, my colleagues were using piezo cylinders about 50mm long for mirror positioning and needed 2Kv of drive to get 10um movement. I was building miniature lasers, so I used bimorphs and I was able to use a much smaller element and reduce the voltage to 250V.

I don;t think you would get nearly enough sensitivity from a cylindrical piezo element. If you look at the ones listed at https://www.kistler.com/gr/en/products/ … lbf_9131_b, even the smallest one senses "up to 2.5kN". We want to sense a small fraction of 1N.

So this is why the mounting arrangement I proposed is designed to let the piezo flex a little when the nozzle contacts the bed.

- David


Duet WiFi hardware designer and firmware engineer
http://www.escher3d.com
https://miscsolutions.wordpress.com

Avatar

By Moriquendi 15 January 2017 21:10

Member · 39 comments

You may well be right, it would make sense for the buzzer elements I'm using to seek to operate at the lowest voltage for the maximum volume, nobody wants 2Kv in their alarm clock. I'm using piezo elements in the bending mode under the bed of my delta but they seemed to respond well to being tapped while flat on the desk as well. I think the circuit could be tuned to work with a smaller signal without much difficulty, perhaps someone should try it.

Moriquendi

Avatar

By dc42 15 January 2017 21:35

Administrator · 2,297 comments

You might like to try the following experiment - I just did, using a 27mm piezo disc:

1. Support the disk on a ring around its periphery. Connect the piezo to an oscilloscope. Tap the centre and watch the voltage. I get a pulse of 20V.

2. Put the disc (or as much of it as you can without the wires getting in the way) between two flat surfaces, with the top one being an insulator (e.g. glass). Push down on the top surface, but not directly over the piezo, to clamp it flat,  Use your other hand to tap on the top surface directly over the piezo. I get a pulse of 100mV, so 200x weaker.

Given the size of the 20V pulse, I am confident that a 27mm diameter piezo (20mm active element) will be sufficient, despite the fact that only a ring near the periphery of the active element will be free to flex, because the central section will be clamped or glued to the hot end heatsink. The main question for me is whether the amount of flex that the piezo allows will permit too much lateral movement of the nozzle. Attaching the piezo at the bottom of the hot end heatsink would be better from a vibration perspective, but shielding it adequately form the heater block would be difficult.


Duet WiFi hardware designer and firmware engineer
http://www.escher3d.com
https://miscsolutions.wordpress.com

Avatar

By Dougal1957 16 January 2017 08:47

Member · 265 comments

David

just seen your post on the Google forum re Haydens balls and ends and can say they are superb and the Delrin ends will work fine with the balls at 90 degrees but not out the side of the robotdigg carriages.

Your idea of a PCB Effector makes me wonder if strain gauges could be used as a sensor setup in place of the Piezo disc?

Avatar

By leadinglights 16 January 2017 10:46

Member · 8 comments
dc42 wrote

You might like to try the following experiment - I just did, using a 27mm piezo disc:

1. Support the disk on a ring around its periphery. Connect the piezo to an oscilloscope. Tap the centre and watch the voltage. I get a pulse of 20V.

2. Put the disc (or as much of it as you can without the wires getting in the way) between two flat surfaces, with the top one being an insulator (e.g. glass). Push down on the top surface, but not directly over the piezo, to clamp it flat,  Use your other hand to tap on the top surface directly over the piezo. I get a pulse of 100mV, so 200x weaker.
..............

I suspect there is something wrong with your setup as I find using the piezoelectric disks in bending gives about 3 times the sensitivity as using them in compression, not the 200 times that you have found. This time 3 gain is not cost-free however as there is a considerable penalty in terms of compliance although such is the sensitivity of piezo disks that even this penalty is unnoticeable.

A big plus for using piezo disk in bending is that it is easy to engineer out the mechanical cross-talk that has plagued the piezo bed leveling in my Cartesian printer. If the three sensors under the bed can be replaced with one on the hot-end that is even better.

Mike

Avatar

By DjDemonD 16 January 2017 12:13

Member · 415 comments

Okay so MkII of my design is now on thingiverse http://www.thingiverse.com/thing:2034152 (and on my large kossel) working well, although after assembling it my bowden clip broke in the v6, so now I will have to dismantle it again. I might do a series of photos to show how to assemble it, and a full BOM of parts needed.

y3m5CUjcX7skggC9vEVDgPAqlFyFOEOv0g17STZEbe0P5EMNhG5Rpb5s5AUHyC3grQGdkhfsb866T7B_K7LwEuqcDpKdUTAjxyiEnZXqC0_f4e9VhfaEspB-0eLf_FED5AoAsAODeNWM_isUhqD7uQLIpSVGIFmZ8cCtpeUfrkR_EE?width=1024&height=768&cropmode=none

Its much smaller now the whole unit is just 40mm diameter and it holds the nozzle fairly firmly. Only two bolts to hold the upper and lower sections to one another as they now slide over one another quite tightly with 4 flanges around the outside. Two m3 bolts to hold it to your carriage/effector. If printing it use plenty of support. I brushed the domed section above the hot end with acetone afterwards as it looked a little thin, but it sits nicely on top of the hotend, and insulates the piezo from the hot end.

Next step design an effector with the upper piece built-in.

I found a 5mm spur point wood drill bit cuts a much neater hole through a piezo transducer than a twist drill or diamond core drill. Use a wood board for support, low rpm and stop if it gets hot, went through it neatly like a hot knife through ABS.

Last edited by DjDemonD (16 January 2017 13:06)

Avatar

By dc42 16 January 2017 12:46

Administrator · 2,297 comments
leadinglights wrote

I suspect there is something wrong with your setup as I find using the piezoelectric disks in bending gives about 3 times the sensitivity as using them in compression, not the 200 times that you have found.

If I don't preload then disk then it responds more to pressure. I suspect this is because the disk or the surface it is on is not flat, so it is actually bending a little. But you were talking about preloading the disk in your post.


Duet WiFi hardware designer and firmware engineer
http://www.escher3d.com
https://miscsolutions.wordpress.com

Avatar

By leadinglights 16 January 2017 14:21

Member · 8 comments

This is quite interesting and I would like to find what is giving rise to the differences that we getting. The setup arrangement that I use and the sort of outputs I get are shown in the original article of the RepRap forum thread http://forums.reprap.org/read.php?1,635075,page=1
I have not documented my studies of bending but the setup was a 27mm piezoelectric disk epoxied to a 70mm by 30mm piece of 1.6mm FR4. One end was secured and a point load applied at the other end. Once I saw how much output I was getting I tried tapping with an admittedly uncalibrated finger on both the new "flexing mode" setup and the old compression mode setup. The difference was 2.7V from the compression rig and 7.4V from the flexing rig.
I use "no-name" piezo disks as I have found that the sample to sample spread is less than the Murata devices - which also have the opposite polarity of output. The Murata disks are less temperature sensitive.

Mike

Avatar

By DjDemonD 16 January 2017 22:19

Member · 415 comments

Made a few slight mods to the design, added a small lip above the lower sections to avoid the bowden coupler being squeezed (and disconnecting) when assembling the unit. This should also decrease the contact area onto the piezo for a stronger signal. Made a slight cutout in the upper section to allow the piezo to bend slightly more. On thingiverse.

Avatar

By Rommie 17 January 2017 16:12

Member · 26 comments

the piezo method is well beyond my capability I think to be fair I may try to get rid of the tilt completely instead lol. I wouldn't know where to start or what to wire to where so unless doug makes it for me and tells me I'm screwed lol

Avatar

By (In)Sanity 17 January 2017 16:22

Member · 279 comments

I like the one company that's using an accelerometer to detect the impact,   Congrats on your work however,  hope it all works out well.

Jeff


Duet WiFi Powered FFCP with E3D legends hotend system.  BLTouch grid leveling.

Avatar

By DjDemonD 17 January 2017 18:23

Member · 415 comments

Rommie, its not very tricky really, you need a board which Moriquendi can make for you for a donation to his charity, and a sensor the total cost is about £10. Print the (3) parts and assemble it, then adjust the board to get a trigger when pressing up on the hotend, but not when just moving around (I actually don't mind if it triggers a bit when moving as I set slow speeds for bed probing and normal speeds for printing and I am not using the probe when printing so it doesn't matter), instructions are provided. Then setup your z probe (I can provide details for this for duetwifi, I will plug it into a ramps board too and work out the config required).

After that you're away. Super accurate, nozzle based probing! I am still improving the design to find the optimum between sensitivity (which is very high with piezo's - so plenty of headroom here) and nozzle firmness. The latest revision is very sensitive and works but my nozzle is too wobbly for my liking.

Jeff thanks for the thumbs up, I've seen the SeeMeCNC accelerometer z probe but then also read comments suggesting it would be difficult/impossible to make it work - thats beyond me I am not an electronics native.

Avatar

By Dougal1957 17 January 2017 18:45

Member · 265 comments

My sensor board arrived this morning (Only ordered it on Sunday)

Doug

Avatar

By Rommie 17 January 2017 23:24

Member · 26 comments

ok so need a sensor board then, Moriquendi have you got any boards left buddy?

Avatar

By Dougal1957 18 January 2017 06:57

Member · 265 comments

Sean

You need to PM him he is not normally on here he did have 5 on Sunday?

Avatar

By Moriquendi 18 January 2017 08:23

Member · 39 comments

I've got two left, PM me if you're interested.

Moriquendi