Page 1 of 1

Piezo disk bed leveling

Posted: Fri May 27, 2016 11:28 am
by leadinglights
Does anybody know of any problems that there may be with using Piezoelectric disks for bed leveling that I may have missed? It seem a bit unlikely that something as good as this seems not to have had any attention.

To give a bit of background: When I was checking out bed leveling technologies I came across several mentions of work by Njål Brekke (aka elenhinan) on using piezoelectric disks to sense nozzle contact. A couple of quick tests showed that the sensitivity of piezo disk sounders as transducers is very high so I built a simple amplifier to precondition the signal. After a couple of itterations I built this into a printer and it has now been in use for a couple of months with no significant problems.
  • So far I have checked for problems with the following:-
  • Sensitivity: huge, possibly too much.
  • Temperature: Even cheap ones are god up to 80°C, Murata to over 100°C
  • Ageing: no problems after several thousand cycles.
  • Humidity: looking good although I use a silicone compound to keep humidity out.
  • Variability: No-name brands seem to be very well matched from disk to disk.
  • Price: dirt cheap.
  • Supporting electronics and hardware: not complicated.
I put some working files on http://www.thingiverse.com/thing:1399512
A discussion on the RepRap forum at http://forums.reprap.org/read.php?1,635075

This is not an attempt to sell anything, I have no interest in commercializing it, just a genuine wish to know if this is as good as it seems to me to be.

Mike

Re: Piezo disk bed leveling

Posted: Fri May 27, 2016 12:05 pm
by Xenocrates
It sounds like sensitivity will be your biggest issue. false triggering and the accuracy of the pistons is the biggest issue with FSR based systems, which would carry over to the Piezo system. (The pistons being the mount system for the FSR's) It's very interesting, but if the sensitivity is two high, then a build on the plate might trigger them, which would be bad if you don't start ignoring them during a print. Your mount system also seems more complex than the piston based FSR setup, and mounting issues are honestly one of the biggest issues encountered. (Also, you note the voltage spikes and three legged table effect, and while you've devised counters to them which are rather clever, they still add complexity to the system, and complexity can lead to failure (Ignoring the fact that most of us have some of the most complex devices ever devised surrounding us, and have things in our pocket that would be unthinkably complicated a mere few decades ago), where possible I like to reduce complexity of a system).

All that being said, I look forward to seeing what you come up with, and it certainly is an interesting approach to a problem as yet far from solved.

I don't think you missed much, other than perhaps bootstrapping issues, as the printed parts would need reasonable accuracy, which might not be available on a poorly tuned machine. However your design would be much more fault tolerant than the closely fitted pistons used for FSR's around here.

Re: Piezo disk bed leveling

Posted: Fri May 27, 2016 12:39 pm
by U.S. Water Rockets
I like the idea, but I've always wondered why nobody had bothered to just mount a simple microphone under the bed and use it to listen for the tap of the nozzle on the bed. I believe that with the proper filters (hardware or software) that the specific frequency of the tap could even be tuned and the system would work even on noisy printers in noisy environments.

Re: Piezo disk bed leveling

Posted: Fri May 27, 2016 9:37 pm
by Captain Starfish
Xenocrates wrote:It sounds like sensitivity will be your biggest issue. false triggering and the accuracy of the pistons is the biggest issue with FSR based systems, which would carry over to the Piezo system. (The pistons being the mount system for the FSR's) It's very interesting, but if the sensitivity is two high, then a build on the plate might trigger them, which would be bad if you don't start ignoring them during a print. Your mount system also seems more complex than the piston based FSR setup, and mounting issues are honestly one of the biggest issues encountered. (Also, you note the voltage spikes and three legged table effect, and while you've devised counters to them which are rather clever, they still add complexity to the system, and complexity can lead to failure (Ignoring the fact that most of us have some of the most complex devices ever devised surrounding us, and have things in our pocket that would be unthinkably complicated a mere few decades ago), where possible I like to reduce complexity of a system).

All that being said, I look forward to seeing what you come up with, and it certainly is an interesting approach to a problem as yet far from solved.

I don't think you missed much, other than perhaps bootstrapping issues, as the printed parts would need reasonable accuracy, which might not be available on a poorly tuned machine. However your design would be much more fault tolerant than the closely fitted pistons used for FSR's around here.
Complains about complexity in a post containing nested parentheses. :lol:

If it works, great. I use piezos on dive equipment because they respond to first time derivative of displacement rather than the displacement itself, so I can filter out depth changes and still catch taps on the controls. That same filtering (bias network, cap to ground if required and a FET switch) also tunes the sensitivity so you can set them to need quite a whack to respond. All about the tuning, I guess. They are subject to cracking under point load but I suspect so are FSRs.

The only real downside to them is the tiny, tiny currents they generate - in a running machine with stepper motor power lines running everywhere, it'd be very easy for that signal to get swamped with EMI if you weren't extremely careful with layout and placement of the signal conditioning circuitry.

Re: Piezo disk bed leveling

Posted: Fri May 27, 2016 11:23 pm
by Jimustanguitar

Re: Piezo disk bed leveling

Posted: Sat May 28, 2016 4:48 am
by leadinglights
@Xenocrates, Repetier firmware ignores zeroing inputs other than when expected, but the sensitivity is the biggest problem as the printer still shakes after an XY move and the present mount system has lots of crosstalk from the X & Y channels. The obvious answer is to have a delay between an X or Y movement before advancing in Z to detect the bed. Sadly, my high level programming skills are poor.

@U.S. Water Rockets, I did a quick check with just a single piezo to detect the nozzle contact. Depending on the position of the supports and the sensor, the signal could be inverted at some X/Y positions and entirely absent at others - I think it would work but needs some research. A variation is to inject a mechanical oscillation to the nozzle with another piezo or a pager motor and detect that with the bed sensor.

@Captain Starfish, The 27mm disks I use have a source impedance of 300 Ohms and a capacitance of 20nF. This makes them the Sherman tank of the sensor world and they ignore most EMI - while being mechanically sensitive enough to detect a fly sneezing.

@Jimustanguitar, That is the very article that got me investigating Piezoelectric disks.

Downsides that I have found, all minor so far are:-
  • The firmware needs some modification to allow the printer to settle before each Z probe operation.
  • There are variations in polarity, sensitivity and temperature dependence between different manufacturers despite the disks being visually indistinguishable. Sample size so far has all individual disks within a batch are very similar.
  • The screws on the hot bed had to be lengthened to go through the mounting mechanism. Although there was no significant change in compliance, there was obvious movement as shown by the adjusting nuts unwinding. (present cure is to mage the nuts a fair friction fit)
A late edit: Piezo disks, or at least my implementation, may need a sub frame if there is no space under the lower platform of the printer.
20160527_191956[1].jpg
Photo shows view on bottom with 3 parallel mechanisms and an amplifier. Note the big Nylon nuts - made stiff to keep them from unwinding.

Mike

Re: Piezo disk bed leveling

Posted: Sat May 28, 2016 12:58 pm
by Xenocrates
Perhaps pre-positioning in your probing script in X/Y, then a G04 (Wait), then the G30? We use similar setups on robotics a lot (Admittedly not with a G04, as they use VAL or V+ rather than G), as well as for some operations on CNC machines (Lathes mostly, since the spot drill cycle has one built in), and G04 is at least nominally supported across all firmware (I have not actually looked very far into probing macros yet, so let me know if it's wrong)

Cpt. I like to reduce complexity of systems. Obviously it has to go somewhere. Be glad I haven't started doing footnote labyrinths yet. Also, shielded twisted pair is relatively easily available, and would reduce the impact of EMI, if that were a concern, which from light's description it isn't

Re: Piezo disk bed leveling

Posted: Sun May 29, 2016 11:57 am
by Eaglezsoar
Thank you leadinglights for bringing this to our attention and in great detail.
The piezo sensors is an interesting and I believe an excellent way to detect probing
and I will use your references to check it out. Happy Printing!

Re: Piezo disk bed leveling

Posted: Mon Aug 01, 2016 9:03 am
by leadinglights
A quick update to the progress with using cheap piezo disks as Z contact sensors for automatic bed leveling. As hoped for, the Rostock printer has not suffered from the cross-talk from X and Y movement that needed a reduction in sensitivity on my Cartesian printer. The trip point on the comparator is now set to 0.1N, about 10 grams and bed mapping achieved a flatness of 0.05mm with a repeatability of better than 0.01mm. Reliability has been exceptional with only a single false triggering event in over 30 bed mappings (750 Z detections)

Xenocrates said that he liked to reduce complexity of systems. The parallel mechanism which I use can be eliminated if the piezo disk has a hole in it so allowing it to be mounted under any bed mounting springs. I tried drilling holes in some piezo disks and they seem very usable: The output is clean and capacity in line with the remaining piezo ceramic. Care needs to be taken and a very sharp drill used to make the holes. Photo below shows a 27mm piezoelectric disk which has been drilled 4mm and the piezo ceramic and top electrode removed with a flat drill bit.
20160725_110621.jpg
Drawing showing how I see a drilled piezo disk being used
RingPiezo.png
Mike

Re: Piezo disk bed leveling

Posted: Fri Feb 17, 2017 5:16 am
by leadinglights
The use of piezoelectric disks on Delta printers has moved on a lot in the past couple of months. DjDemonD on the RepRap forum has developed a Z contact sensor which is fitted in the effector while Lykle has a new effector design to incorporate DjDemonD's sensor.

The ongoing forum thread is at http://forums.reprap.org/read.php?1,635075 from about page 4 onward.
Piezo hotend at http://www.thingiverse.com/thing:2069480 and Lykle's effector incorporating the piezo sensor is expected very soon - check the RepRap forum topic above.

Although this is an ongoing effort, it already seems to be a very good candidate for automatic bed leveling and automatic nozzle height setting.

Mike

Re: Piezo disk bed leveling

Posted: Sat Feb 18, 2017 7:01 am
by Zesty_Lykle
Well, we published the design now. It is on http://www.thingiverse.com/thing:2117069

In fact we published it with 3 different effectors so you can pick and choose what works for you.
DJDemonD has used this latest version and is very happy with it. So it is tested.

[img]http://i.imgur.com/pFR9qNi.png[/img]

This truly has been a very nice collaborative job, lot's of people came together to make it happen. I was just the last in the line, simply to finish up the design of the various components. All the thinking was already done by Leadinglights and DJDemonD. Kudo's to them for a great idea. Oh and by the way, it can be used with a bowden cable or with the Nimble, in case you don't have a Nimble.

Next task is to see if we can do the same but then with smaller sensors on a Chimera. But first I have to finish the Diamond nozzle effector. Almost done, but it does almost break my brain getting it to work.

Lykle

Re: Piezo disk bed leveling

Posted: Wed Mar 01, 2017 12:51 pm
by kraegar
Installed my piezo last night. I have an e3d v6 & 713maker's aluminum hotend mount. Installing the piezo wasn't bad - I put two nylon washers around the bowden to effectively make the hotend taller than it is (so the bowden collet is recessed down below the top washer), then I put the piezo on (coated side up), then a ~25mm nylon washer above the piezo, and finally the lock ring for the hotend mount. When the nozzle touched the bed, it flexes the piezo up and produces a sharp spike for sure.

I'm probing at 5mm/s with good results currently. Overall installation wasn't bad - tuning the VR's to dial in the piezo was a pain in the butt, though.

Re: Piezo disk bed leveling

Posted: Fri Mar 03, 2017 5:59 pm
by kraegar
Here are my repeatability test results so far:

https://docs.google.com/spreadsheets/d/ ... sp=sharing

Re: Piezo disk bed leveling

Posted: Sat Mar 04, 2017 9:53 am
by geneb
It still blows my mind that you can bore a hole through the middle of the poor thing and have it still work.

g.

Re: Piezo disk bed leveling

Posted: Mon Mar 06, 2017 9:04 am
by kraegar
Drilling the hole in them is definitely the trickiest part of the process. It seems I killed 3 or 4 of them and got one good one. (They work, but lost a lot of their sensitivity)

Re: Piezo disk bed leveling

Posted: Tue Mar 07, 2017 4:12 am
by leadinglights
The brass with its ceramic coating seems to blunt ordinary drills after only a couple of tries. Tungsten carbide drills are much better but 4.5mm carbide drill bits are quite expensive. The best method I found was to mount them on a bit of plastic in a lathe and use a small end mill bit in a rotary tool to machine them out. I haven't taken any pictures of discs being machined but the picture below shows the same operation on a slotted Berd type air tube.
BerdRing_GrooveMilling.png
The rotary tool is a Proxxon clamped http://www.thingiverse.com/thing:949748 in the lathe and the bit is a 1mm end mill £3.95 for 10 from China.

Edit: I have now put some pictures of the piezo disk being cut in a jig - on http://forums.reprap.org/read.php?1,635 ... msg-753999

Mike

Re: Piezo disk bed leveling

Posted: Thu Jun 22, 2017 2:43 pm
by leadinglights
I have now been using piezo discs on both of my printers for over a year now and they have proven completely reliable. What problems there were were easy to diagnose and correct. Since that time DjDemonD in the U.K. and ECJ in Brazil have independently developed piezo Z probes attached to hotends and others have produced hotend probes and underbed variants which appear to work well. In all of the correspondence that I have followed I have not seen any that showed any problem with the basic piezo disc sensor or even the simple amplifier circuits. Perhaps coloured by my own experience it seem to me that piezo disc sensors used for nozzle height setting and automatic or assisted bed leveling is pretty much the only game in town, having superior repeatability, robustness and even lower cost.
It comes as a bit of a surprise to me then that reviews of ABL technologies do not seem to mention piezo discs; even more so that comments like "I would not use a piezo Z probe" are sometimes seen in forums. Perhaps some of you could humour me by answering the following questions:-
  • Have you tried using piezo discs (purchased probe or DIY) for nozzle contact sensing and it has not worked?
  • Have you tried using piezo discs for nozzle contact sensing and it has worked but been disappointing?
  • Have you considered piezo disc based Z probes or underbed sensing but decided on something else instead?
  • Do you refuse to consider piezo disc based nozzle contact sensors?
Thanks for any replies and for any comments.

Mike

Re: Piezo disk bed leveling

Posted: Mon Jul 17, 2017 11:25 pm
by WZ9V
I'm building a V3 now with the trick laser FSR adapter plate. is there a simple way to use the underbed solution with that?

With the FSR setup you attach them to the bottom plate put the rubber washer on top and use some silicone adhesive to glue the whole stack together. I'm not sure what I'd need in the similar scenario with the piezo disks.

Re: Piezo disk bed leveling

Posted: Thu Jul 20, 2017 2:34 pm
by leadinglights
I don't know the trick laser FSR adaptor plate but there is a bit about various ways people have used piezo discs in the following message on the RepRap forum http://forums.reprap.org/read.php?1,635 ... msg-771134 also others scattered through that fairly lengthy thread.

Mike

Re: Piezo disk bed leveling

Posted: Thu Jul 20, 2017 10:27 pm
by WZ9V
I think I'll look into the effector mount option instead. It seems to be the most supported and I can avoid a long Bowden tube by mounting a flying extruder which seems to work well in my other delta printer that has a BL-Touch setup.