Piezoelectric disks for Z contact detect and bed levelling

Clearlynotstefan - Bristol, UK. That's very kind of you, I only need an offcut.

Edited 1 time(s). Last edit at 03/14/2017 09:03AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
JamesK
I'm surprised that print surface makes much of a difference for a physical contact mechanism like this, but I guess the difference between a very hard surface like glass and a thick softer surface could be enough to cause a reduction in signal. I'll certainly be very interested to see the results.

Print surface makes a big difference to the performance of accelerometer probes too, according to some extensive testing done by a user (mhackney) of the SeeMeCnc accelerometer probe. He got it to work much better on glass than he did on PEI.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Yeah I'd like a few runs (100 data points) for each surface, just to be able to say we've tested it on x,y,z surfaces.

We can then also recommend a z-offset for each surface based on the results, assuming there's much of a difference. It stands to reason that a compressible surface will generate a larger offset, however the probe is very sensitive 10g of pressure is all that is needed to trigger it.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Just checking, if I'm holding the board so it's vertical with the pins facing down, which is pot1 and which is pot 2?!

Also, why do you favor digital as opposed to analog on the dirty and what g code specifies which?

They will be marked perhaps on the reverse as VR2 and VR1. Adjust VR2 first.

I favour digital as you can set higher travel speed between probes due to the debounce value. With analogue I was having to massively reduce speed/jerk/accel before probing, otherwise I got "probe triggered before move complete" or some such error, due to the mechanical noise when decelerating to stop over a probe point. With digital it seems to look for the probe trigger for far less time and therefore filters out false triggers. Its also wise to experiment with it using the lowest motor current you can (I used 500mA) to limit any damage caused by head crashes. If it seems very insensitive reverse the piezo polarity.

So digital:
M558 P5 I1 F300 X0 Y0 Z0 ;digital piezo sensor output falls on contact, probing speed, not used to home axes
G31 X0 Y0 Z-0.2 P1 ;sensor is nozzle and debounce ;sensor is nozzle and debounce value

Analogue:
M558 P1 I1 F300 X0 Y0 Z0 ;analogue piezo sensor output falls on contact, probing speed, not used to home axes
G31 X0 Y0 Z-0.2 P700 ;sensor is nozzle and trigger value.

The P700 value on analogue is the threshold over which it considers a signal to be a trigger, You can raise it to make it less sensitive or lower it to make it more sensitive.

You can watch the analogue value on your duet web control, it will vary as you move around, see the effect of higher acceleration and observe the value rising when the nozzle travels over rough surfaces or hit blobs. However you will probably have to use lower speed/accel/jerk to get it to probe each point successfully. We were thinking it might be able to be used as a slow down or stop system to detect crashes or other obstacles, but we haven't got anywhere with that yet,

Edited 2 time(s). Last edit at 03/14/2017 12:17PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Interesting, I've been probing at f700. You found this too quick? I was thinking a little speed should give a more definitive probe impact, I look forward to seeing how 300 works.

VR1 is next to the Piezo input pins, VR2 is next to the GND pin. They are marked on the board though it's possible it's underneath the pot body.

Idris

There's not much in it, on my test rig I've tried from 100 to 1100 and seen little difference. The spread in results was higher for higher speeds but we're are only talking 0.01mm more spread. (my rig is not best placed to do really high precision as its too bendy and using marlin so only 2 decimal places output) I asked Tom from youtube for his code from his arduino based sensor rig but he didnt get back to me.

Proof is in the delta calibration deviation before and after using the piezo probe. I was getting 0.03-0.04 with IR probe and metal effector and carriages, if I had a very clean bed. Height maps were slightly undulating. Now I have a printed effector and piezo probe and I'm into the 0.02-0.03 range and my height map is flat. Looking forward to either having a metal effector or the new PCB effector DC42 is testing and using it with the piezo probe.

Edited 2 time(s). Last edit at 03/14/2017 12:35PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

For a printer with a printing enclosure heated to better than 100C I had intended to use three underbed piezos mounted on storks to keep them out of the heat but had an idea that I think is worth trying. As the cold part of my hotend is watercooled, why not use that for keeping the piezo disc at a decent temperature. I have made a first attempt in that direction and first results are encouraging.
The piezo disc is kept under direct compression between the two surfaces by a coil spring (not visible in the photo) and a leaf spring which also keeps the body of the hotend from rotating about the Z axis. One surface consists of an aluminium block which is in thermal contact with the water jacket while the other is a piece of 2mm thick FR4 glass fiber laminate mounted to an effector body. In the photo the nozzle has been replaced by a 10mm dia. steel ball, a second steel ball is used to strike the first one in the Z, X and Y axes to see what the relative sensitivities are in each axis. Also shown in the photo are another piezo disc and the striker used to get a signal.


As the disc is neither allowed to flex, nor has the mechanical amplification from a pressure release method, some loss of sensitivity was expected but a signal of about 1V was obtained with a tap from the striker and this should be quite usable. One thing that I did find was that preload pressures above about 2kg caused the sensitivity to drop.
The big test will need the hotend to be mounted in my test jig - a light milling machine that I use for all of my tests. I hope to have some results by the end of the week.

Mike

Piezo probe with active thermal management, careful you'll be working for Tesla before long! Looks great, looking forward to the 100 deg C chamber printer thats fairly extreme. Far from preventing warp you might actually anneal the parts whilst printing them.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Regarding probing speed I did start some test runs with 100mm/min, 200, 300, etc... data is on the spreadsheet, but my rig crapped-out at 400 something was moving as the probed distances started to slip. I think I am going to have to build a fancy rig, out of proper stuff.

Edited 1 time(s). Last edit at 03/14/2017 05:02PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I just installed the "heated enclosure piezo hotend" in the test setup. A bit more difficult this time as I need to set the point where it stops moving down accurately as even my little milling machine can exert more than 10kg of push and this setup is very solid in the Z direction.
The results that I get shows that the output is not too bad, about half the output of my undrilled discs fitted to an underbed mounting and about a fifth as sensitive as a 27mm disc drilled disc in bending. The plot below shows the output from 0.1mm of travel at a speed of 1mm per second - the compliance is so low that the pressure on the nozzle is about 1kg at the peak!!!. Vertical scale is 1V/cm and horizontal is 50ms/cm

Note on the plot: I think that the small wiggle on the plot is steps from the Z motor and my guess is that the little jiggle at the beginning is to do with the stepper going into ramp-down - this needs more investigation though.
More work is needed as this was only a single disc and there may be variations. The X and Y compliance is also much more than the Z compliance - probably O.K. though but needs measuring.

Mike

p.s. On probing speed: I use a speed of 1mm/sec on the test jig but on my printers I double this to 2mm/sec, f120

Quote
DjDemonD
Piezo probe with active thermal management, careful you'll be working for Tesla before long! Looks great, looking forward to the 100 deg C chamber printer thats fairly extreme. Far from preventing warp you might actually anneal the parts whilst printing them.

My real target is 80°C but if this works well the star to shoot at is 100°C so all calculations and materials are aimed at this. Back in the real world I will accept any hotness I can get above 60°C

Mike

Hey how much can I take off of the peizo before I start facing consequences? This system where I have to disassemble the entire effector to pull out my Bowden tube is unworkable. I had a thermal jam today, took me like 45 minutes to pull it out and put it back together lol. If I can take a little more out of the Piezo disk I could get a thin tweezer from the top to push down on the bowden release.

Hi you can try taking more out of it, piezo discs are cheap... experiment. I have redesigned the clamp for the 20mm version (which could be integrated into the Lykle effector which I presume is the one you're using) which allows the bowden tube to be disconnected and the hot end removed without dismantling the piezo module.

How often do you need to dismantle your hot end? I'm just curious as for me it's every couple of months....

Edited 1 time(s). Last edit at 03/15/2017 03:29PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Has anyone tried orienting the piezo vertically instead of drilling a hole through the center of it? I hooked one to my 'scope last night and could get what I think is a usable signal by tapping it on the edge of the disc. Maybe glue a #6-32 nut to the ceramic face in order to "pull" on the ceramic?

I got the boards in from oshpark last night, but I won't be able to get them assembled for a few days - maybe someone here could experiment with the idea?

tnx.

g.

Its worth looking into, they provide a good voltage rise when they bend, so adhering them to a flexible part of an assembly (like in Moriquendi's bed supports for his delta) will bend them nicely when the assembly bends.

Have you got a sketch of how you might mount one vertically yet also mechanically coupled to vertical movement of a hotend? I'm trying to think how you could accommodate one neatly in this configuration but I haven't got anything in mind. This is what I was suggesting a few posts back, if we could get custom piezo's made what would they look like?

The 20mm ones are a bit more versatile for the scale of the parts we are talking about.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

My thought was to mount one between two of the mounting spacers that attach the hot end assembly to the effector platform. You could even mount it to a thin cork disk to damp out any vibration introduced by the motion system. (assuming a small weight (a nut) glued to the ceramic face would produce adequate movement when the nozzle struck the bed)

I can't sketch it up right now, but I could tomorrow if needed.

g.

Please do I'm sure it will be food for thought.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I've got a few thoughts on how to do it vertically. Maybe I'll throw one together.

To verify, the current board with 3 inputs will use 3 discs and trigger upon first contact with any of them right? I've got some ideas that I really think would work well.

That's correct for the Z-probe boards v1.1 and 1.2.

Idris

When the discs are wired in parallel the output voltage approximates to the sum of the input pressures. If each has a 10 gram load put on it the output will be roughly what it would be with 30 grams on one and nothing on the others. The loads must be applied at the same time of course.

Mike

Good point Mike, the reverse is true too, having one move in one direction and another move in the opposite direction will result in zero signal if they both move the same amount.

Idris

Just spitballing here... I wonder if you could mount the piezo somewhere on a motor. When the force increases on the belt, it will try to twist the motor on the mount... Might save all that tedious mucking around on the effector.

I think that piezos on the motors would be so far from the nozzle that the the signal would be too weak and too slow to be useful for calibration. Having said that, the idea may have a lot of merit for detecting other things going wrong - such as the print lifting and being struck by the nozzle. Octoprint or other software could shout "Hey boss, some sh_t is happening" or equivalent. Piezos are cheap and MCUs can be had with plenty of spare pins so all that is needed is the programming time.

Mike

Well we were discussing on duet forum using the analogue values from the piezo board as a crash/rough surface detector, which could stop, or slow down the machine under certain conditions.

I think your idea Nebbian would work if you had slightly compliant motor mounts with piezo's, as we have a slightly compliant hotend mounting system with a piezo, but the elasticity in the belts especially on a corexy or large deltas would make for a much more vague signal.

As for it being messing about, that depends on your point of view this version is designed to just be a drop in module, take your hot end off, stick this in, clamp your hot end into it and connect your endstop wire. Its not there yet but it coming along. I'd love your input on this, can you think of a way to mount a piezo to something (they work bonded to any flexible structure also) to detect nozzle contact without drilling one?

Drilling them isn't that problematic, and getting some made with holes would surely be possible. But doing it without drilling would simplify the process and reduce the variability in the piezos. However you set it up though you don't want a wobbly nozzle.That 20mm alpha version the nozzle is firm, the piezo triggers when it meant to - happy days.

Edited 1 time(s). Last edit at 03/17/2017 06:41AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

A cautionary tale for anyone using ebay piezos.

I've just spent the best part of a week swearing at my printer because I couldn't get it to probe correctly.

I'm using underbed piezos, three piezos bonded to compliant supports under a round bed. The ABS supports aren't really up to working under a 100deg bed despite the protection I gave them, they've been working well for about six months but it was time to replace them. I printed a new set, bonded on the new piezos and installed them in the printer. When I probed the bed, probing beside one tower didn't work, it'd push the bed down 2-3mm before triggering. Nothing I could do would improve it, I printed a new set of supports, new piezos, checked every connection again and again and always, one of the supports wouldn't trigger but after I changed the piezos it would be a different support. Tapping the bed or the bare piezos produced a good signal even with a gentle tap. I swapped the board, tried without the glass bed installed, everything I could think of.

Finally I reversed the connector on the offending piezo and I'm back to single microstep accuracy. Turns out you can't rely on the electrical polarity of the piezo being the same, even within the same batch. Mine were connected in parallel, centre contact to centre contact, rim to rim. However, on one of them the piezo ceramic itself had the opposite polarity, pressing it down produced a negative signal so it wasn't till it stopped moving that it produced a positive signal that the board registered as contact.

Has anyone else had a similar problem? Is this limited to ebay piezos or will the Murata and Kepo ones have the same problem. For the application they're designed for (buzzers) it wouldn't make any difference what the polarity is, perhaps they just don't check.

Idris

Yes I have been commenting about polarity a lot as it seems to now make sense based on what you've said to be quite random. I presumed it was because I was using the inverted logic instead, one way around they seem to trigger but the voltage rise (maybe Mike you've seen this on your scope?) is not fast enough to trigger quickly enough and you get a crash, the other way around they are amazingly precise.

So something to be wary of with our experimental beta units, if its performing poorly, reverse the piezo connector. For pre-built units I will make sure they are the right way round, by testing them on the rig, before they go out of the door.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

For my intended use case of three point under bed mounts I wanted to be able to cope with the bed moving in opposite directions at the sensors for some probing points, so my circuit rectifies the signal before passing it to the op-amp, and uses three separate channels with the output combined digitally at the comparator output. It's not particularly expensive using quad op-amp, quad comparators.