Piezoelectric disks for Z contact detect and bed levelling

Quote
ECJ

..............................

I do not understand why you care about this question of probing with the heated nozzle. Of course there is variation with heating, but I assume it's always the same. Here I do the probing with the cold nozzle and compensated with z offset and never had problems.
Is there really any plausible reason to probe with the heated nozzle? I wonder what it is.

To expand a little on what JamesK and DjDemonD said,

• If you probe cold and have a constant Z offset then you have to wait for the hotend to cool down in the event you need to reprobe between prints or if the print fails.
• To be good rather than merely acceptable, the first layer thickness should be consistent to within 0.04mm. Failing to achieve this sort of consistency can result in a print curling at the edges rather than adhering well. To show that it is possible to exceed that during warm up from the cold probing and starting the print consider the following: The printer is a Delta with 500mm long aluminium rails, the temperature is 15°C when you do your probing but by the time you start printing it has warmed up to 25°C because of the hotend and the bead heater and possibly you are warming the room. The aluminium rails with a coefficient of expansion of 22ppm/°C and a temperature rise of 10°C will have expanded by 0.11mm or nearly three times the ideal range.
• You would need different Z offsets for different materials as the hotend will expand to different lengths. As before, a hotend with a 40mm barrel used for both PLA at 180°C and ABS at 250°C. The coefficient of expansion of 304 Stainless Steel is 17.3ppm/°C, over the 70°C difference the length will change by 0.048mm - this is again more than the ideal range.
• The rapid change of hotends and build surfaces should be done without having to wait for everything to cool down and reheat. See the following video for example [www.youtube.com] Although nozzle probing is not shown there, if it was desirable to use a different thickness foil than there would be no time penalty.

I know that many respected members of the RepRap community think that Z probing is either a sign of laziness or just unnecessary but perfect prints are more likely if initial conditions are perfect and a quick Z probe at the beginning of the G-code goes a long way to giving perfect initial conditions.

Mike

Quote
DjDemonD
My thoughts are probe at 130 deg C nozzle temp (below ooze temp) with bed fully heated to 1st layer temperature for the material you intend to print. You wont damage your bed at this temperature for the brief contact involved.

The hot nozzle ensures that any residual plastic at the nozzle has minimal/no effect on your probed height. Of course you should clip it with a wire cutter or whatever beforehand but we're all human and forget occasionally. The heated bed because some (probably a lot) of beds, deform when heated and therefore grid levelling or delta autocalibration will be a poor fit for the actual bed surface.

The effect of thermal expansion is minimal we did a calculation based on an e3d hotend where the nozzle, heater block and the lowest part of the heatbreak are expected to expand when heated from 130 deg C to 250 deg C it was only 0.02mm, so account for it it you want but its probably not worth it, however from 20 degrees to 250 degrees its more like 0.035 which may have more of an effect.

As for not having to have any z- offset, currently this is not realistic, as there has to be some vertical compliance in the probing module to get a reliable trigger, less so with later module designs, which pre-load the piezo more, but still in the order of 0.1-0.2mm. So I set -0.1 in firmware, and occasionally babystep up/down 0.05mm.

Perhaps referring to the earlier post by Mike about getting a probe to trigger at 1g of force, we might get a result which genuinely requires no offset except thermal if you're bothered by 20 microns error.

But sure probe with a cold nozzle (but a hot bed) if you want to and then work out the offset.

Yes, I understand those questions. But I keep my nozzle always clean in any way, and I'd no doubt sound it cold.

As for the question of obtaining detection with only 1 gram of pressure, how theoretically do you imagine avoiding false shots?
I'm starting a new version. Instead of the 25mm perforated disc, I will use two 10mm discs arranged side by side and connected in parallel. This should suppress most non-vertical movements. But only 1 gram I find it difficult to achieve.

Quote
ECJ
...........................

As for the question of obtaining detection with only 1 gram of pressure, how theoretically do you imagine avoiding false shots?
I'm starting a new version. Instead of the 25mm perforated disc, I will use two 10mm discs arranged side by side and connected in parallel. This should suppress most non-vertical movements. But only 1 gram I find it difficult to achieve.

Avoiding false triggering is the problem -- the answer is that I don't know and never will without trying. As my Cartesian printer used a much higher triggering pressure I was surprised at getting 10 grams on the Delta.

Using 2 10mm discs sounds like a very good idea

Mike

Just wondering.. Is there any reason you couldn't mount the board near your control board and just run wires from the piezo back to the board?

None Brian. Thats what I do. I have a v1,0 board on my delta and a 1.1 board on my corexy, both are mounted by the printer controller, with a long wire to the piezo.

The only reason my new drop-in module will have the controller attached to it is so that its a self contained unit, and so that it can illuminate on trigger.

Edited 1 time(s). Last edit at 04/24/2017 05:49AM by DjDemonD.

---

Simon Khoury

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

Quote
briangilbert
Just wondering.. Is there any reason you couldn't mount the board near your control board and just run wires from the piezo back to the board?

Although there is some electrical noise associated with piezos and short wires do help here, long wires are still usable. The biggest source of noise by far is mechanical noise but fortunately there is not a lot of this on Delta printers as the bed is stationary. Printers with moving beds do have a major problem in that the bed will still be shaking at the end of an X, Y or XY movement.

I have recently acquired a 3 axis accelerometer and have started looking at what noise there is and how to minimize it. First indications are that mechanical noise lasts for only about 1/4 of a second after the end of an XY movement and that probing after this point on a shaky early Mendel Prusa should be possible for a sensitivity around 1 gram. It is early days with the accelerometer yet, but I can see a lot of detail that may be motor steps, frame shake, cogging of the toothed belt and other things - or it may be incomprehensible jiggling. I will publish what I find on this forum when I have got some useful conclusions.

Mike

Over the last couple of days I have been getting data from the 3 axis acclerometer and have reached the point of getting reliable tripping at 5 grams of nozzle contact pressure but have not been able to get below this point. The one useful thing I have been able to accomplish is to show that from the point of nozzle contact until the signal to the controller takes somewhat less that 1ms. As the Z speed is 2mm/sec the worst case* for repeatability is better than 2µm and is likely to be closer to 0.2µm.

As I have been doing these tests of a Mendel Prusa printer which is very shaky I will continue on a Delta printer as this has been in use for some time at 10 grams nozzle contact pressure. Electrical noise has also started to become apparent so I will also tweak the filter characteristics a bit.



Picture above shows that ADXL335 accelerometer board from Adafruit, also digital DTI and 100 gram calibration weight

Mike

* Repeatability for the piezo that is, there can be backlash, stiction and other things that make it worse.

Hey! I have sort-of-successfully used a 20mm piezo design once, however I stopped because of inconsistencies due to what I thought was slight hotend clamp tilt caused by the bowden tube (delta printer).

By now quite a lot of time has passed since this whole piezo idea started evolving...I guess people have come up with their own ideas...and something more than what's on thingiverse.

Would you mind sharing the effector setups you use ?

Thanks!

You could try putting a bowden coupler on top of the top plate and running a ptfe guide tube down into the hotend. This decouples the bowden tube from the sensor, I have to say I haven't suffered with it much but that approach does work.

Edited 1 time(s). Last edit at 04/30/2017 06:50PM by DjDemonD.

---

Simon Khoury

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

BIG thanks to all you folks that did this work!
I just added a Moriquendi board and 3 piezos to my bed mounts and it works fantastically. Over the years I have tried microswitches, FSRs, proximity and IR detectors. This is by far the easiest, cheapest, least invasive and most accurate that I have used.

Which also makes me wonder: why the emphasis on incorporating it into the hot end? The bed mount scheme seems easy and performs well.

Simplify it even more just one piezo, and because I want my bed clamped solidly not on any type of floating mount. I used to use fsrs but wasn't that impressed (although they beat inductive sensors and microswitches). Also cartesian printers can't use the under bed system easily.

Also it brings it to a much wider audience.

Edited 2 time(s). Last edit at 05/03/2017 12:48PM by DjDemonD.

---

Simon Khoury

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

Quote
swoozle
BIG thanks to all you folks that did this work!
I just added a Moriquendi board and 3 piezos to my bed mounts and it works fantastically. Over the years I have tried microswitches, FSRs, proximity and IR detectors. This is by far the easiest, cheapest, least invasive and most accurate that I have used.

Which also makes me wonder: why the emphasis on incorporating it into the hot end? The bed mount scheme seems easy and performs well.

Hi swoozle,

Can you put up some photos of how you mounted the piezos. I had to use a complete sub-frame to mount the piezos on my delta but Moriquendi has a method which is much easier. Cartesian printers are actually much easier than Deltas as long as they have a three screw bed adjustment method.

Putting a single piezo in the effector or behind the hot end comes very close to a universal solution as the RepRap community is moving towards having one or only a few informally standardized hot end mounts.

Mike

Quote
leadinglights

Quote
swoozle
BIG thanks to all you folks that did this work!
I just added a Moriquendi board and 3 piezos to my bed mounts and it works fantastically. Over the years I have tried microswitches, FSRs, proximity and IR detectors. This is by far the easiest, cheapest, least invasive and most accurate that I have used.

Which also makes me wonder: why the emphasis on incorporating it into the hot end? The bed mount scheme seems easy and performs well.

Hi swoozle,

Can you put up some photos of how you mounted the piezos. I had to use a complete sub-frame to mount the piezos on my delta but Moriquendi has a method which is much easier. Cartesian printers are actually much easier than Deltas as long as they have a three screw bed adjustment method.

Putting a single piezo in the effector or behind the hot end comes very close to a universal solution as the RepRap community is moving towards having one or only a few informally standardized hot end mounts.

Mike

It's not elegant but it was easy and quick to model and make. One fitting holds a piece of fiberglass rod to keep the plexiglass with the bed on it from moving laterally. The other fitting is simply a 20mm ID cup with an inner depression on the bottom that is slightly smaller (to put the piezo in bending) and a plug (with a smaller foot) sitting on top of the piezo. The bed of course rests on the plug.

Hi swoozle,
That looks like a nice simple design that isolates tha piezo from the heater. Piezos have been able to defeat all accusation of failings except the need to engineer around their thermal problems. Placing the piezo away from the heater and behind an acrylic bed should do that well.

Mike

Talking of which, is there any way we could build into the circuit some method of compensation for temperature variation?

This is possibly where ECJ's method might offer some advantages, there could be a "calibrate" button on the board (or potentially a calibrate signal sent from the controller at some future point), press it, then probe a few times, the board then sets the threshold and sensitivity values required, and away you go.

What about using an arduino nano as the piezo signal board they're cheap, readily available and we can program it to handle the signal any way we want? Yes there is the issue of ensuring the signal does not exceed 5V but I suspect with a series resistor, certainly for the hotend probe version with just one 20mm piezo that should ensure that even a nasty head crash doesnt send more than 5v to the arduino?

---

Simon Khoury

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

The problem with correcting for temperature coefficients that they cover a huge range - each piezo would have to be individually measured and it would only give a vanishingly small improvement. Although my investigations of piezos have shown their response to a contact to range over nearly 10:1 (about 2V to nearly 20V) and their speed of response ranging from less than 0.5v/ms to over 6V/ms, even this variation is of almost no importance. Taking the worst case above, if the voltage being regarded as showing contact is 1V then this is reached after only 2 microns if the nozzle is moving at 1mm/sec - and this will be reasonably repeatable between each probing operation. This excellent behavior continues all the way up to some indeterminate temperature above 60C where it falls off the edge of a cliff.

I think the best thing to look at to improve the frankly sucky temperature behavior of piezo diaphragms is to look at using a commercial piezo ring such as this one [www.steminc.com] At $25 for 5 they are more expensive but still quite livable if only 1 is needed in the hotend - less so if 3 are fitted to the bed.

ECJs method of having a digital filter in an inexpensive microcontroller is great but I would feel uncomfortable without input protection, anti-aliasing filtering and some buffering. With an 8 pin Microchip PIC or Atmel ATTiny and a quad dip switch to set filter characteristics this is still under $5 for parts.

Going the other way of course, you can simply wire the piezo in series with a 10k resistor straight into the RAMPS or other controller and it is likely to work - albeit a little temperamentally.

Mike

I only mention it as I am testing a hotend drop in module in my enclosed machine which can reach 55 deg C chamber temp passively and the piezo is totally out of tune most of the time, due to temperature variation. Perhaps with some sort of digital signal filtering a wider range might still be interpreted correctly as triggered?

I will obtain some of the piezo rings and try them, I also have a piezo film element to play with, its just finding the time.

I'm just looking to find a way to reduce the amount of messing around a potential customer/user of this system has to do, its no problem for a kit, we expect to have to fiddle and tinker, think ultibots fsr kit for comparison, but the drop in module is a bit harder to deliver "works out of the box", when I am having to retune test modules frequently. Maybe this is a function of the 20mm piezos outputting a smaller signal especially drilled and having a narrower optimum range of settings on the current signal boards.

Edited 1 time(s). Last edit at 05/05/2017 07:27AM by DjDemonD.

---

Simon Khoury

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

Could you put a high pass filter into the board?

Like a capacitively coupled section? Would only be a couple of extra components, and might get around the very low speed issues that are caused by temperature.


Just an idea.

I'm pretty ignorant about these things but the signal from any given piezo element is very variable so it might be millivolts or volts. Anything which filters in an absolute way is not going to work.

I think, and Mike can correct me if I'm wrong, the issue at high temperatures like 60 deg c+ if the signal just gets too small to be detected.

Edited 1 time(s). Last edit at 05/05/2017 12:42PM by DjDemonD.

---

Simon Khoury

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

I suspect that it is a bit of both. The sensitivity falls dramatically by about 60ºC but fast changing temperature can generate a big potential across a piezo called a pyroelectric potential. Example with some rough numbers below* is not as bad as it looks as you only have about one fifth of the capacity on the piezos (20nF against my 90nF) and Moriquendi's circuit has a low enough impedance (1MΩ against my 10MΩ) to ignore most of the drift.
The solution comes in several bits.

• After the piezos have stopped warming up they no longer generate this pyroelectric voltage to give them a bit of time to stop warming up
• A bit of insulation near the piezos could slow down the warm-up and reduce the pyroelectric voltage.
• The make of piezo can have an effect on the higher temperature sensitivity of the piezos, for example, no name 27mm piezos fell over at about 60ºC but 27mm Murata were O.K. to about 100ºC but with 20mm Murata ones they seemed to taper off even before the 20mm no-name piezos

Example: On my Delta I was getting the output of the conditioning circuit showing contact made even though the probe had not started down. I checked the temperature around the the piezos and found it climbed from room temperature to about 40ºC while the heated build stage climbed to about 100ºC. I looked at the output of the 3 piezos in parallel directly with the scope (Times 10 probe 10MΩ input resistance) I got 1.6 Volts of output per 1ºC per second that the temperature went up by. If the temperature around the piezos took 5 minutes to get from 20ºC to 40ºC then it is going up by about 0.067ºC per second which generates a bit over 100mv across the piezo which was about the trip value of the circuit. Note that the pyroelectric voltage can make the response more sensitive or less sensitive depending on how the sensor is arranged mechanically.
The solution was to change my input bleed resistor from 10MΩ to 1MΩ which makes it about the same as Moriquendi's circuit. Since doing that the sensitivity is great and there is no sign of thermal effects.

Mike

Edited 1 time(s). Last edit at 05/05/2017 04:14PM by leadinglights.

What about 'Gravity: Flexible Piezo Film Vibration Sensor'

£4.57($5.90)+shipping
Digital/Analog output

My modified version of a piezo effector is almost complete. It leads me to a question. How much preload can these guys take?

A fair bit, test it wire it up, preload it and see how it responds. I have enough that the hotend is not loose, I presume when used under a bed, especially a heavy one there is a reasonable amount of weight on them.

---

Simon Khoury

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

Quote
rolloo
What about 'Gravity: Flexible Piezo Film Vibration Sensor'

£4.57($5.90)+shipping
Digital/Analog output

I just bought a piezo film like this, just working out how to use it. Not using their circuit but might be worth a go see how well it works.

---

Simon Khoury

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

Just a note about electrical interference. I have until recently been saying "its not a problem to have a long lead from the piezo to the signal board", and it hasn't been a problem for me until now.

Maybe its because I am now running dual extruders and new fans and there's more wiring, especially motors, maybe its because I installed a 20mm piezo module on my kossel XL, and 20mm piezo's produce a weaker signal (on average) than the 27mm one it replaced, I am not really sure. However the piezo probe in general had been performing much more poorly than normal, so perhaps it was the second extruder. I was noticing it worked fairly well cold, but almost non-functional with the hotend heated, and the temperature near the piezo was only 30 deg C. I scoped the signal from the piezo at the PCB down the 80cm of wire and and it looked like I'd connected the scope to my TV antenna.

Anyway the short version is - if you're talking about a hotend probe, put the PCB on the effector/carriage, to keep the piezo lead short. The range of usable adjustment on the PCB ("the tuning window" if you want) will be much wider, which will then mean you will be able to work with a better range of temperatures also.

Important lesson for me, I was beginning to think I had bad PCBs (I don't) or that piezo just wasn't doing the job, maybe its not as great as I thought it was, and low and behold there is a perfectly rational explanation for the problem.

And.... regarding polarity, testing it before using the disc isnt really necessary in that once its attached if you press up on it and it triggers instantly its the right way around, if it triggers just as you let it go, then its reversed polarity - sorry if everyone already knows this or it seemed obvious but it really only just occurred to me.

Edited 2 time(s). Last edit at 05/06/2017 04:15PM by DjDemonD.

---

Simon Khoury

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

So here's the idea I've been playing with for an effector. I'd love to hear your guys input before I tinker too much with printing and adjusting it. Someone mentioned vertical positioning and I saw it as a chance to be rid of the stability rods, the collaborative effector works well for me, so I'm def trying to fix what aint broken, but hey that's reprap for ya. Relying on those rods being straight and not allowing and horizontal movement always unsettled me, so this might be a way to get horizontal stability from the probe itself.





The 3 piezos are arranged on different angles into slots that are cut out in the shape of the piezo. If they can hold a decent amount of preload when arranged vertically like this, it will definitely hold strong horizontally. Whether they will provide accurate probe results I'll find out once I get some printed. I'm just looking for general input or ideas before I proceed. If it works, you'd only need to print the inner two components, the outer ring remains the same, this is just a drop in replacement, I had my original parts SLA printed for accuracy so I don't want to redo the part that actually connects to avoid effector tilt.

Edited 1 time(s). Last edit at 05/08/2017 05:03AM by clearlynotstefan.

I see no reason why this wouldn't work and I am keen to hear how it performs, however it might be worth pointing out that in my groovemount hotend adaptor (which on thingiverse is the "alpha 20mm piezo" which has become the Piezo20 module I am shortly to be offering for sale assembled and will post the final design as its fully OSH) there are no stability rods, just 4 machine screws, with a fairly heavy pre-load, and the hotend is rigid laterally, the module barely appears to have any compliance in any direction even vertically. But then as I have been saying recently when used in this way the piezo functions closer to a microphone, than a flex-detector.

---

Simon Khoury

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

Hi clearlynotstefan,
I don't like to pour cold water on any idea but I don't think that the geometry of you idea would work. Piezo discs are primarily sensitive to bending and have a secondary sensitivity to having pressure applied across the thickness of the piezo - neither apply in your geometry which will be most sensitive in the X and Y directions.
There is a variation on your idea that I believe would work and that is to have the three piezos as if they were on three faces of a cube with the shared point in line with the nozzle. If all three are equally sensitive then the re should be an output only from pressure in the Z direction.

I think that DjDemonD was previously using piezos in bending but is now using them across the thickness where there is insufficient movement to need sliding parts.

Mike

Yeah I have a partial retraction to make, I did not realise the circular elements you show in your render are the piezo discs, I thought they were just thin sliding elements. With the piezo in use drilled as I have been using them so far. So I will say still interested to see if it works, but perhaps less optimistic that it will. That being said these things are very sensitive, so it'll do something, but you want to optimise your "tuning window", i.e. have the most potential range to work from, and I don't know if that configuration will deliver only a very narrow window of usability which will be annoying to live with day to day.

As I said a few posts back having a very narrow window to tune the disc (which I was getting from electrical interference but which might occur if the disc isn't ordinated to optimise its abilities) means temperature variation takes you out of your tuning zone and you don't get a trigger.

Edited 2 time(s). Last edit at 05/08/2017 08:41AM by DjDemonD.

---

Simon Khoury

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

How bad can you rough these things up before they stop working? What if I pre bend them to a bit of a C shape and adjust the slots to retain that bend? Then I'd get bending force if they still work at all. The only reason I tried this approach is noting that they trigger if you tap them sideways on the bed, certainly not optimal, but might work. Now if they'll still work pre bent, that'd give me a very sensitive reading as I'd be applying bend to them with pressure.thoughts?